Your Federal Quarterly Tax Payments are due April 15th Get Help Now >>

DVP Manual for CSIRO 1994 by p4P9G0H


									                      Manual for creating a

                    NATURAL AREAS

                               on the


                             prepared by
                        Morgan M. Robertson
                                for the
 Cooperative Research Centre for Tropical Rainforest Research and

       Sponsored by the Australian-American Education Foundation

Morgan M. Robertson                1                   Creating a DEM
This manual was prepared in October, 1994 at the CSIRO Tropical Forest
Research Centre in Atherton, Queensland. It is a cooperative project conducted
by staff of the CSIRO Division of Wildlife and Ecology and James Cook
University of North Queensland Department of Tropical Environment Studies
and Geography, under the auspices of the Cooperative Research Centre for
Tropical Rainforest Ecology and Management, Cairns.

This project was funded by a 1993 Fulbright Scholarship awarded by the
Australian-American Educational Foundation for the study of ecology in
Australia. This work was submitted as partial fulfillment of a Postgraduate
Diploma of Science in Tropical Environment Studies and Geography.

Morgan M. Robertson                    2                      Creating a DEM
     “Once you have tasted the exciting new world of computer-aided
       photogrammetric stereocompilation, you will never want to look back.”

                                              Leica DVP Manual, pp. 1-1

Morgan M. Robertson                  3                      Creating a DEM

This is a manual designed to lead you, the user, through the steps necessary to
construct a digital elevation model (DEM) from aerial photographs using a
scanner and the Leica Digital Video Plotter (DVP) currently located at the
Atherton CSIRO. It was written to aid the Wet Sclerophyll Cooperative
Research Centre members at the CSIRO and James Cook University of North
Queensland (Cairns Campus) in constructing highly accurate elevation models
of the gradsects surveyed by the Wet Sclerophyll field teams.

This is a tutorial; it is structured with only the goal of creating a DEM in mind,
and is not for use as a general reference manual for the DVP. There is a
reference manual accompanying the DVP which is well-suited to trouble-
shooting and answering specific questions, but is not of as much use as a start-
to-finish guide to procedure.

The Digital Video Plotter
The DVP is a software program that has been loaded on to a normal PC, whose
sole function is to record the three-dimensional coordinates of features visible
on aerial photographs or satellite images. It will allow the user to view two
overlapping digital images of terrain, which is known as a stereoscopic model,
and extract data about the three-dimensional coordinates of features on the
images. By recording the three-dimensional location of the ground at various
points, you will create a file of coordinates that can be used to make an
elevation model. The elevation information from models covering an entire
gradsect will be stored in files until you have extracted information about all the
photographs of your study area. From there, the files can serve as the base for a
number of GIS applications.

Morgan M. Robertson                      4                        Creating a DEM
Structure of the Manual
The manual is divided into four chapters, each with several main sections. At
the head of each section is given an estimate of the amount of time it will take
to perform or learn the process(es) described within that section.

Chapter 1 deals with the acquisition of photographs, and describes the
fieldwork which must be done in preparation for the modelling process.
Chapter 2 explains the procedure of scanning the photographs using a digital
scanner. Image files are produced, which will be transported to the computer
which hosts the DVP software. Chapter 3 describes the process of orienting
images digitally so that two images can be viewed in stereo, and to fix image
coordinates to real-world coordinates. Chapter 4 describes the process of
capturing data from the model, and exporting it to be analysed on other
cartographic software.

This manual is designed to explicate some, but not all, of the processes outlined
in the Leica manual. A read-through of the Leica manual before starting work
on the DVP is recommended, particularly of chapters 4, 6 and 7.

You will be using:
 an A3 or A4 scanner
 a networked PC connected to said scanner (the “scanning computer”), with
  the appropriate scanning software, as well as Adobe Photoshop, loaded onto
 a networked PC with the DVP software and hardware installed on it (the
  “DVP computer”)
 DVP hardware (includes: a mirror stereoscope mounted on the main monitor,
  a digitizing tablet, a digitizing puck, a mouse, two monitors, dongle inserted
  into the printer port).

Morgan M. Robertson                     5                        Creating a DEM
In case of difficulty
The Leica representatives in Sydney are always available for queries; do not
hesitate to call them with the smallest problem. Their numbers are:
        Brian Nichols, Photogrammetry Sales Engineer -
                            office - (02) 888-7122 ext. 116
                            mobile - (018) 415-569

        Briony Lincoln, Photogrammetry Sales Assistant -
                            office - (02) 888-7122 ext. 115

Brian Nichols’ business card should be affixed to the body of the computer as

Keep in mind that the UNIX operator at the CSIRO (currently Guy Carpenter)
may also be able to address some of your problems, particularly those dealing
with network connections.

Here are definitions of some of the more cryptic words scattered around both
this manual and Leica’s manual.
Capturing data: recording the three-dimensional coordinates of a feature on an
   image on the DVP.

Control points: points marked on the DVP screen by the user to bring two
   coordinate systems into orientation. Thus, the user will place a relative
   control point at identical spots on two separate images, and the identical
   points will be used to bring the coordinate systems of the two images into
   Relative Orientation.

Driving (to a location): moving the split-screen display of the DVP to a
    different location on the model by selecting a point on the reference
    photograph with the digitizing puck.

DVP: the Digital Video Plotter, developed at the Université Laval in Québec,
  and marketed by Leica.

DVP computer: the PC on which the DVP software and hardware is installed.

Morgan M. Robertson                     6                       Creating a DEM
Fiducial mark: the marks in the corners of aerial photographs which serve to
    impose a coordinate system upon the photograph.

Groundfile: the DVP file in which the absolute control points are kept.

Information bar: The edge of an aerial photograph with information about
    altitude, flight time and date, and site of the flight.

Model: stereoscopic model - the area of overlap between two photographs
                    taken from different angles.
       digital elevation model - a computer file containing coordinates which
                    represent the location of the surface of the ground (or
                    canopy) across an area being modelled.

Orientation: bringing two coordinate systems into alignment so that they can be
    mutually convertable.

Pointing: bringing the floating marks together on the DVP split screen, using
    the puck and mouse together, to mark an identical point on both images.

Puck: the instrument connected to the digitizing tablet, which moves the cursor
   in most operational modes of the DVP, and can be used to select coordinate
   locations on the digitizing tablet.

Reference photograph: The photograph laid on the digitizing tablet next to the
   DVP computer, to aid the user in driving around the model.

Scanner: a machine which operates, and looks, much like a photocopier, except
   that it makes a digitised image from a physical image, for viewing in
   graphics software.

Scanning computer: the PC to which the scanner is attached, and on which the
   scanning software is installed. The digitised image created by the scanner is
   saved on the scanning computer.

Split screen: the upper portion of the DVP’s monitor display. It displays
    together, side-by-side, portions of two images to be viewed in stereo.

Vector: a line, point, arc, rectangle or circle which is drawn on the DVP screen
   by the user, and whose three-dimensional coordinates match the feature
   over which it is traced. This is the method by which the user can capture
   the location of features of interest. Vectors exist only as connections
   between points marking the same feature; points are grouped into vector
   elements (see below).

Morgan M. Robertson                     7                       Creating a DEM
Vector element: a collection of points marked on the DVP screen , and the
   vectors connecting them, which defines the shape of a feature of interest on
   a DVP stereoscopic model. The coordinates of vector elements are stored
   in a vector file, so that the three-dimensional location of the feature can be

Vectorizing: tracing lines, points, rectangles, circles, or arcs over features
   shown on the DVP screen so that coordinate data about those features may
   be recorded.

This manual is stored on the UNIX computers at the CSIRO in three parts
under the filenames DVPMAN.DOC, DVPINTRO.DOC and
DVPTITLE.DOC. The associated diagrams are stored in the same directory
under the filenames DVP*.BMP, where * is a number between 1 and 15.

Good luck!

Morgan M. Robertson                     8                       Creating a DEM
                               TABLE OF CONTENTS

        The Digital Video Plotter
        Structure of the manual
        In case of difficulty


 1      Chapter 1: PRIOR TO SCANNING
 2      1.1 Obtaining the Photographs
                1.1.1 Colour vs. Black and White
                1.1.2 Ordering properly exposed photographs
3 1.2 The Camera Calibration Certificate
  4     1.3 Obtaining ground control points during field study
                1.3.1 How, when and where to obtain ground control points
                1.3.2 Problems in obtaining ground control points
                Identifying your location on the photograph
                Canopy interference

 8      Chapter 2: SCANNING IMAGES

 9      2.1 Correctly positioning the photograph on the scanning plate
                2.1.1 Using an A3 scanner
                2.1.2 Using an A4 scanner
                Trimming photographs to fit an A4 scanner
                Positioning the photograph on the A4 scanner
                Performing sample scans to check inclusion of fiducial marks
 13     2.2 Determining cosmetic changes to the image
                2.2.1   Disk space
                2.2.2   Setting Scan Mode and Resolution
                2.2.3   Preview scan
                2.2.4   Determining appropriate brightness settings
                2.2.5   Determining appropriate contrast settings
 19     2.3 Scanning the image
                2.3.1 Scanning with 386 computers
                2.3.2. Scanning with 486 computers
 24     2.4 Transporting large files
                2.4.1 Storage options
                2.4.2 File Transfer Protocol
                2.4.3 Moving image file to the DVP computer from the scanning computer
                2.4.4 Moving image files from the scanning computer to a net computer
                2.4.5 Storing the image files on magnetic tape

 Morgan M. Robertson                            9                              Creating a DEM
       ON THE DVP

29     3.1 Structure of the DVP
30     3.2 Steps to Orientation
               3.2.1 The reference photograph
               3.2.2 Getting around in the DVP
32     3.3 Naming or choosing a model
33     3.4 Photo reference
34     3.5 Internal Orientation
               3.5.1 Getting around a bug
               3.5.2 Making a camera file
               3.5.3 Recording internal reference points
              Moving the floating mark
              Placing the internal reference points
               3.5.4 Photo base
               3.5.5 Calculation of Internal Orientation
               3.5.6 Modifying the Internal Orientation
40     3.6 Relative Orientation
               3.6.1 Selecting the method of pointing
               3.6.2 Selecting relative control points
              Getting around
                       More about Puck Mode
              Driving to the control points
              Monocular pointing
              Stereoscopic pointing
                       Problems of stereo pointing
              Monocular or Stereo pointing?
              Marking the relative control points
               3.6.3 Computing the Relative Orientation
49     3.7 Absolute Orientation
               3.7.1 Importing absolute control points
               3.7.2 Recording absolute control points
              Point Numbers and Ground Files
               3.7.3 Computation of Absolute Orientation

53     Chapter 4: VECTORIZING ON THE DVP

53     4.1 What is Vectorizing?
54     4.2 Modes and Codes
               4.2.1 Getting around in Vectorization
              The Second Screen
               4.2.2 Selecting a code
               4.2.3 Selecting a mode

Morgan M. Robertson                            10                       Creating a DEM
58     4.3 Placing an element
               4.3.1 Lines
               Open and closed polygons with point-by-point lines
                        Creating closed polygons
                        Creating open polygons
               Stream lines
62     4.4 Editing an element
               4.4.1 Snapping
               4.4.2 Deleting elements
63     4.5 Other Features
               4.5.1   Making codes visible and invisible
               4.5.2   Vector weight and line type
               4.5.3   Text
               4.5.4   Importing other vector files
66     4.6 Creating the DEM
               4.6.1 The DEM function on the DVP
               4.6.2 Building a better DEM
               4.63 Marking elevation control points
71     4.7 Output
               4.7.1 Creating an export file in ASCII
               4.7.2 Transporting the vector file

Morgan M. Robertson                           11                             Creating a DEM
                                    Chapter 1
                           PRIOR TO SCANNING

Before you begin using the DVP, you must go through a number of steps to
acquire the photographs and make digitised images of them. These procedures
are broken down into the first two chapters of this manual. This chapter deals
with obtaining properly exposed photographs of an area and the accompanying
documentation, and with the fieldwork necessary before sitting down at a

You can skip this chapter if...
You can skip this chapter if you are working on photographs of the Mt. Baldy
area. All of the photographs from that area have already been acquired and
scanned, and you should begin at Chapter 3. In fact, if you are working on any
gradsect being examined by the Wet Sclerophyll Project, you can move on to
Chapter 2, because the photographs for all Wet Sclerophyll gradsects have
already been obtained. They are kept in the photogrammetry lab of the Division
of Wildlife and Ecology Wing at the CSIRO. Examine them carefully; you may
need to re-order certain photographs (see comments on photograph quality in
section 1.1.2).

Furthermore, users working on the Mt. Baldy, Paluma, Mt. Windsor, and Mt.
Haig areas, reading section 1.3 is highly recommended, as ground control
points were not obtained for these areas during fieldwork by the Wet
Sclerophyll field teams.

Important note
Obtaining the photographs can take up to a week, and obtaining ground control
points must be done during the field expeditions to the gradsect you are
working on. Both of these are rate-limiting factors: you cannot do anything
until you have obtained photographs, and while you will not need ground

Morgan M. Robertson                    12                       Creating a DEM
control points until Chapter 3, they are virtually impossible to obtain after the
survey field trips to an area have concluded.

1.1 Obtaining the Photographs
Time: allow at least a week for sending photographs back and forth.

1.1.1 Colour vs. Black and White
Aerial photographs of an area may be obtained from the local Queensland
Department of Lands office:
        Atherton: 913-222
        Cairns: 523-434

Although black and white prints are preferred for use with the DVP because
they allow for slightly better resolution, most photographs are available only in
colour, for a price of $22.50 each.

The Lands Department will not make black and white prints from colour
negatives. If you want black and white photographs, be sure that black and
white negatives exist for the area you want before going to the trouble of
ordering them. Although black and white photographs are preferred for use
with the DVP, colour photographs are quite acceptable since colour
photographs can be scanned in such a way as to produce a black and white
digital image.

1.1.2 Ordering properly exposed photographs
Each aerial photograph belongs to a “run” (pass of the aeroplane), and a
gradsect may be covered by several runs. Usually, photographs within a run are
all of the same general exposure quality (the same levels of contrast and
brightness), but between runs there can be considerable variance.

You must make sure that the runs are all fairly bright; a run that is consistently
dark will cause problems later in the digitising process. This particularly

Morgan M. Robertson                               13                  Creating a DEM
 applies to rainforested areas, which are unfortunately both the darkest areas on
 aerial photographs and the areas of most interest. Even the darkest rainforest
 gullies on the most dissected terrain must show some resolution of detail in
 order to construct an accurate elevation model for that gully. If you can’t see a
 feature clearly on the photo, you won’t be able to extract data about it.

 If one or more of your runs has large dark portions, or the exposure seems to be
 quite dark in general, you will want to return them to the Lands Department and
 have new ones sent to you. The Lands Department will do this for free, if you
 send them the following:
  The original photographs which you want replaced with brighter exposures.
  A cover letter explaining why the photo quality is inappropriate for your
  A photograph of satisfactory exposure (if such exists) from a nearby run, to
   show the photo technicians what your idea of acceptable brightness is.

 Photogrammetry work is done in the Lands Department’s main office in
 Brisbane. Staff at the Brisbane office have been contacted about this problem
 before. Contacts are:
                Dennis Williams, Senior Cartographer
                Bob McIntosh, Manager
                Remote Sensing Section
                Queensland Department of Lands
                Locked Bag #40
                Coorparoo Delivery Centre, QLD 4151

 Bob McIntosh’s phone number is (07) 896-3218.

1.2 The Camera Calibration Certificate
    Time: if you operate by fax, one day should suffice.
After you have obtained the photographs, you will need to get the camera
 calibration certificate for each camera used in photographing your area. This
 certificate will be needed when you orient your photo on the DVP (see section
 3.5.2). The Queensland Department of Lands in Brisbane holds these

 Morgan M. Robertson                               14             Creating a DEM
certificates; speak to Ron Samson, or another Senior Cartographer, in the Land
and Administration Boundaries Program at (07) 896-3229. Have the photos at
hand when you call him; he will need information from them to determine
which certificate to send.

1.3 Obtaining ground control points during field study
   Time: this should be done at the time of fieldwork on each gradsect. If it was not, it can take
several days of independent fieldwork on your part to obtain proper ground control points.

When using the DVP, you will be required to enter AMG (Australian Map
Grid) coordinates for the model you are working on. Therefore it is absolutely
vital that while in the field spot heights, using a Global Positioning Satellite
receiver (GPS), and altimeter readings be taken for features identifiable on
both photographs of a model. GPS receivers, and instructions on their use, are
available from the Technical Officers in the Division of Wildlife and Ecology,

The taking of physical coordinates must be incorporated into the field trip
agenda, rather than arrange for an inconvenient second trip. Some models are
largely roadless and trackless, and it makes sense to take the readings whilst
stomping through the bush anyway, collecting other data. Unfortunately, at the
time of writing, the fieldtrips to several gradsects (Mt. Baldy, Paluma, Mt.
Windsor, and Mt. Haig) have already been completed without precise ground
control points being recorded.

1.3.1 How, when and where to obtain ground control points
You must tailor your field method to the needs of the DVP when acquiring
ground control points. Below is a guide to acquiring a sufficient number of
accurate points for use on the DVP:
 Where? Take readings only in areas where two photographs of the gradsect
  overlap. Mark all these areas on a topographic map before going on a field
  trip. Models will be constructed from the overlapping areas of two adjacent

Morgan M. Robertson                                15                              Creating a DEM
   photographs, so don’t accidentally take readings outside areas of overlap

   Take spot heights at evenly distributed points around the model, especially if
   you are taking only a few readings. This will be hard to accomplish if the
   model is difficult to traverse. The points must be evenly distributed along
   both the x and y axes of the modelled area. If the points are all distributed
   along one line (a road, for example), then there will be great lateral distortion
   away from the line.

 How many? The DVP requires an absolute minimum of four spot heights.
  However, try to collect at least seven spot heights, in order to achieve an
  acceptable level of accuracy.

 How? There are two requirements for obtaining GPS readings:
     1. you must take the reading at a point you can perceive on the
        photograph with precision (a bend in the road, the edge of a clearing,
        the corner of a boulder, etc.).
     2. you must take the reading at a spot unobstructed by dense rainforest
        canopy, so that the GPS signal can penetrate to your receiver.

 The need for precision cannot be overemphasised. You may want to make a
  drawing similar to Figure 1.1 for each site, so that when you begin working
  on the DVP, you know where the reading was taken to within several pixels.
 Warning: The GPS readings that are already a part of the Wet Sclerophyll
  gradsect proforma are not recorded accurately enough to be of use. The
  location of the reading must be pinpointed on the image to within 8-10
  pixels, with certainty.

Morgan M. Robertson                      16                       Creating a DEM
              Figure 1.1: field sketch for obtaining ground control points

1.3.2 Problems in obtaining ground control points Identifying your location on the photograph
One of the greatest difficulties of using the DVP to analyse non-urban images is
the lack of recognisable and accessible points on photographs. Even if there are
several features on a photograph which are easy to pick out and would make
good ground control points, it is often difficult to try and find these features on
the ground. This becomes more difficult the farther away you are from clear
features such as roads. If you are not precisely sure of where on the
photograph you are standing while taking a reading, the GPS reading you take
will be of no use. Furthermore, if there is a single, straight road running
through the area of the model along which you collect GPS readings, you must
obtain at least one reading some distance on both sides of the road.

“The Alphatonia method” has proven to be a solution to the problem of keeping

Morgan M. Robertson                          17                      Creating a DEM
track of location on the photograph while in the field. In photographs of
rainforested areas, the roads are often covered by rainforest canopy; however,
the paths of those roads are clearly visible by the pattern of the flat-canopied
Alphatonia trees which commonly grow along the side of the road. These trees
make it relatively easy to pinpoint your location on the road, even if you cannot
see the road itself on the photograph. Canopy interference
Another difficulty is that, even on a road, rainforest canopy may not be light
enough to allow the penetration of a GPS signal. If your model consists of
nothing but rainforest, you have a problem. Take readings at the most exposed
sections of road you can find, or a rockface: any accessible point where a break
in the canopy occurs will do.

You might come to a clearing along a road where you can achieve a GPS
reading, but still be unable to see the ground at that location. In these cases, go
ahead and record the GPS and altimeter readings, but also note the height of the
canopy. Later, when you are marking your control points on the DVP, you will
place the control point on the canopy above your location, and when entering
the coordinates you will add the canopy height to your altimeter reading.

Morgan M. Robertson                      18                       Creating a DEM
                                       Chapter 2


In this chapter you will make a digitised image of the photograph using a
scanner. This consists of three processes:
1. Positioning the photograph on the scanning plate of the scanner so that all
   four fiducial marks on the photograph fit within the scanner’s range.

2. Ascertaining which cosmetic changes will be necessary to make the image
   acceptable for viewing on the DVP, using sample scans of small areas of the

3. Scanning the entire image. This process includes:
     Executing improvements upon the image using the settings determined
      in step 2.
     Rotating the image to the proper orientation for the DVP (only if an A4
      scanner was used to scan the image).
     Saving the image in the correct format.

You will need a scanner hooked up to a PC, which has the Adobe Photoshop
and AVR Scanpack software installed on it. Preferably, the PC should be a
486-speed computer.

You can skip this chapter if...
All the images of the Mt. Baldy gradsect have been scanned and saved on long-
term storage cassettes, so if you are working on the Mt. Baldy area, consult
with the UNIX technician about importing them to the DVP computer. Skip
this chapter and begin at Chapter 3.

Important notes
 The scanning process can be quite a long and frustrating one; unless you are
  very familiar with scanners and scanning software, allocate most of two
  weeks to produce satisfactory images from all the photographs of a gradsect.

 It is useful, throughout the rest of the process, to make a clear distinction

Morgan M. Robertson                       19                      Creating a DEM
   between photograph and image. The photograph, of course, is the piece of
   developed photographic paper, while the image is the information contained
   in a file created by the scanning of the photograph. The two words will not
   be used interchangeably.

 Henceforth, the sides of the photograph will be referred to by letters, as
  shown in the diagram below (the information bar is Side A, etc.).

                         Figure 2.1: sides of an aerial photograph.
2.1 Correctly positioning the photograph on the scanning plate
Time: allow up to two hours for each photograph.

The image must be positioned on the scanning plate so that all four fiducial
marks are included in the image. The fiducial marks are the small circles
within the black border at each corner of your photograph. They are necessary
for determining the true scale of items shown in the photograph; their use is
explained in section 3.5.2.

If you are using an A3 scanner, including all four fiducial marks is a simple
process. However, if you are using an A4 scanner, inclusion of the marks
requires trimming the photograph and positioning it on the scanning plate, and
performing sample scans to determine if the manipulations have been

2.1.1 Using an A3 scanner
An A3 scanner is sufficiently wide to include the entire photograph in the

Morgan M. Robertson                                20                 Creating a DEM
image without requiring any manipulation or special positioning of the
photograph. To position the photograph:
1. If the photograph is part of an east-west run, turn the photograph so that
   north is at the top. If the photograph is part of a north-south run, turn the
   photograph so that east is at the top. This is how the image will be viewed
   on the DVP, so this is the position in which each photograph needs to be

2. Lay the photograph face down on the scanning plate, with the top of the
   photograph towards the front of the scanner. The photograph should be
   flush against the front left corner of the scanning plate.

3. Close the scanning lid. Do not open it again until you have finished
   scanning the entire image.

4. Proceed to section 2.2.

   Hint: The information bar, with location, altitude and scale information, also
   indicates the direction of flight. It is located on the side of the photograph that
   is towards the nose of the aeroplane.

2.1.2 Using an A4 scanner
The scanning width of A4 scanners is too narrow to include an entire
photograph. To fit all four fiducial marks on the photograph within the
scanning range requires two processes: trimming and positioning. To
determine if these manipulations have been successful, you will perform sample
scans of small areas around the fiducial marks. If they have been unsuccessful,
you will repeat the positioning process. Trimming photographs to fit an A4 scanner
Use a scalpel and a metal straightedge, with a cardboard backing, to trim your
photographs. Do not use a plastic straightedge, as it may snag your scalpel.
Trim Side D (to the left of the information bar) to within a millimetre or two of
the fiducial marks; this usually means 10 or 11 millimetres must be removed. Positioning the photograph on the A4 scanner

Morgan M. Robertson                         21                          Creating a DEM
The photograph should be positioned face down on the scanner, flush against
the front-left corner of the scanner. This means that the A-D corner of the
photograph must be at the origin of the scanner axes, and Side D (the trimmed
side) should be against the y-axis of the scanner (see Figure 2.2) Note that on
the scanner, the black triangles on the two axes mark the limits of the scanning

                Figure 2.2: placement of the photograph on the scanner Performing sample scans to check inclusion of fiducial marks
Having trimmed and positioned the photograph, you will now perform small
“sample” scans of the areas around the fiducial marks to determine whether the
marks are within the scanning range. The margin of error is extremely small;
the scanning width is only 3.3 mm wider than the distance between fiducial
marks. The chances are good that on your first try, you will not be successful in
positioning the photograph so that all four marks are within the scanning range.

To perform sample scans around the fiducial marks:

Morgan M. Robertson                         22                         Creating a DEM
1. Place the photograph on the scanning plate, flush with the lower left corner
   of the plate.

2. Turn the scanner on.

3. Turn the computer on.

   Accessing the scanning software:
4. Enter Windows by typing win. The two software applications you will be
   working with are the AVR Scanpack EasyScan for Windows, and Adobe
   Photoshop. Make sure that these programs have been installed on your PC
   before proceeding any further. AVR Scanpack allows you to scan (digitise)
   an image of the photograph, while Adobe Photoshop allows you to
   manipulate the brightness and contrast levels of the scanned image.

5. Enter Adobe Photoshop by clicking on the Adobe icon.

6. From the Adobe File menu, select Acquire, and then TWAIN. TWAIN is a
   program which allows files that are scanned using the AVR Scanpack
   software to be transported automatically to Adobe for viewing, without the
   user having to exit AVR and enter Adobe. Selecting TWAIN from the
   Acquire menu should cause the AVR Scanpack window to appear.

   Within the scanning software:
7. Without worrying about the settings on the right hand side of the screen for
   the moment, press the Preview button on the screen. This will execute a
   Preview scan to display what is on the scanning plate in the Image Window
   (the blank space on the left side of the AVR window).

8. The striped lines within the Image Window are known as the “crop box”, and
   its borders are movable. The scanner will only scan within the borders of the
   crop box. To move the borders, click on them with the mouse, and drag
   them to a new position.

9. Position the crop box around the area of a fiducial mark (near the corners).
   Keep in mind that the photograph doesn’t extend all the way down the length
   of the Image Window, but is only about ten inches long (the numbers along
   the side are measurements in inches).

10.The amount of computer memory that the image will occupy is given below
  the Image Window. Try to limit the size of your “sample” image to 5

Morgan M. Robertson                    23                       Creating a DEM
   Megabytes or below.

11.Select Scan.

  Observing the result of the scan:
12.When the scanner is finished, select Exit from the File menu of the AVR
  window, and the Adobe window will be reactivated, with the new image
  displayed in the middle of it. At this point, all you are looking for is the
  presence of the fiducial mark within the image. If it is not present, either you
  have scanned the wrong area of the photograph, or the photograph is badly
  positioned on the scanning plate.

13.Adjust the photograph if necessary. Conduct sample scans of the rest of the
  fiducial marks, adjusting the photograph until all of the fiducial marks are
  within the range of the scanner.

2.2 Determining cosmetic changes to the image
Time: at first, this process may take up to a day for a single image; the time consumed will decrease as
you become more adept.
The image must be made cosmetically acceptable for viewing on the DVP.
This will be accomplished in two stages.
1. You will set the Scan Mode and Resolution settings on the AVR window
   prior to performing any scanning. These are settings which remain constant
   throughout all scanning operations.

2. You will then perform scans on selected sections of the photograph, using
   them to find appropriate contrast and brightness settings to use later, during
   the scan of the entire image.

Before these operations, however, it is necessary to ensure that you have
adequate disk space to perform all the manipulations necessary.

2.2.1 Disk space
In the AVR screen, note the amount of disk space available on the C: drive.
The available space is displayed below the Image Window. You will need
three times as much space available as the image requires. computer will need
this extra space when it attempts to transfer the scanned image from the AVR

Morgan M. Robertson                                24                              Creating a DEM
software to the Adobe software, and rotates the image.
 The photograph should require about 54 Megabytes, so about 170 Megabytes
  of memory must be available.

 If there is insufficient memory, exit AVR and free up some disk space, either
  by delete some old images, or move them off of the computer (see section

2.2.2 Setting Scan Mode and Resolution
In the AVR window, the following two settings, both located on the right side
of the screen, must be adjusted as described below. These settings will remain
constant throughout both the sample scanning process and the scanning of the
entire image. There is no need to adjust them again.
 Set the Scan Mode to 256 Greyscale. This will scan the image in black and
  white. Vegetated areas are easier to examine on the DVP when scanned in
  black-and-white than in colour.

 Set the Resolution to 800 dpi (dots per inch). The DVP can handle colour
  images of resolutions between 300 and 450 dpi, and greyscale images of
  between 600 and 800 dpi.

2.2.3 Preview scan
Performing a Preview scan will display an image of the item on the scanning
plate in the Image Window. This allows you to select and scan sample bright
and dark areas of the photograph, in order to find the most appropriate
brightness and contrast settings for that photograph. It is not a proper scan,
merely a preview of what is on the scanning plate.
 Move the pale blue bar in the Image Window down to approximately 12
  inches or so (enough space for the entire image). The scanner will not scan
  past this bar.

 Press the Preview button.

2.2.4 Determining appropriate brightness settings
Brightness is a measure of the ambient light level of an image. To find the
appropriate brightness value for scanning your photograph, you will have to run

Morgan M. Robertson                     25                       Creating a DEM
sample scans at different settings of the Brightness value in the AVR window.
A sample scan is a scan performed on a selected portion of the entire image,
and will allow you to judge the effect of various brightness settings. Once you
have found an appropriate brightness value for a photograph, you can usually
use that value for all the photographs in the same run.

Do not adjust the Contrast setting in AVR. Contrast value will be adjusted in
Adobe Photoshop (see next section, 2.2.5).

To choose a brightness setting:
1. In the AVR window, the Brightness setting has a default value of 50. This is
   a good starting point for determining the appropriate value. For most
   photographs, a setting of between 50 and 60 will work well.

2. Select an area to scan by moving the crop box on the previewed image.
        The area should include very dark and very light features of the
        The area should take fewer than 10 Megabytes of memory to store.

3. Press the Scan button to scan this sample area. It should take about two to
   four minutes to fully process the image.

4. When it is done, choose Exit from the File menu of the AVR window, and
   you will return to Adobe where the image is waiting to be examined.
        Since you are adjusting only Brightness in the AVR, and not Contrast,
         do not expect to see a good image when you return to the Adobe
         window from AVR.

5. The image may be a bit washed-out, but as long as it has no very dark areas,
   you should continue on to adjust the contrast in Adobe. In fact, it is better to
   err on the side of overexposure: the image will almost certainly need to be
   brighter than you think at first.

6. If the image is too dark or too light, return to the AVR screen and perform
   another scan at a different brightness setting. Please note that AVR restores
   the brightness value to 50 each time you return to the AVR screen.

7. Write down the brightness level that works best, and use it when scanning
   the entire photograph (section 2.3).

Morgan M. Robertson                     26                        Creating a DEM
2.2.5 Determining appropriate contrast settings
Adjusting contrast is a matter of making the output values of the image, which
is what you see on the screen, different from the input values of the image.
which is what the scanner reads from the photograph, and what is stored in the
image file.

Contrast can be thought of as the degree to which two pixels of different input
value are differentiated by output brightness; e.g., if pixels of different input
value are not differentiated at all in the output process, then you have an image
all of one shade. Though in 256 Greyscale mode, the scanner can assign each
pixel to one of 256 levels of brightness, in practice the scanner will only receive
input from the photograph at about 150 of those levels. To increase contrast,
then, you will change the output values of each pixel so that an input value of
170 equals an output value of 255, and an input value of 30 equals an output
value of 0. You have now stretched the image input values so that the output
image is displayed in 256 shades of grey rather than 150: this is called a
histogram stretch.

You must perform a histogram stretch on the sample scan you have just
completed in section 2.2.4. The histogram stretch will be performed in Adobe
Photoshop, not AVR.

To perform a histogram stretch:
1. When you are viewing your sample scan in Adobe, select the Adjust
   function from the Image menu, and then the Levels function. The histogram
   for the image should look something like this:

Morgan M. Robertson                      27                        Creating a DEM
                           Figure 2.3: a contrast histogram.

        The input values are along the x-axis, while the number of pixels at
         each value are along the y-axis.
        The black arrow at the far left end of the x-axis marks the input value
         where the output value will be 0.
        The white arrow at the far right end of the x-axis marks the input
         value where the output value will be 255.
        The 254 shades of grey in between white (value #255) and black
         (value #0) are divided up on the x-axis between the white arrow and
         the black arrow.

2. To change the output values, click on the arrows and move them near to the
   positions shown in Figure 2.4. Each image will be different, of course, but
   these positions usually produce satisfactory contrast.

3. If there are two peaks to the curve, a thin one and a large one, then the
   thinner one represents the dark black of the photograph border. In these
   cases, the arrow should usually be placed in the middle of the thin peak.
        This ensures that the darkest black of the output range will be
           reserved for the photo border, and not for any of the features of the
           photo itself.

Morgan M. Robertson                     28                        Creating a DEM
             Figure 2.4: settings for the adjustment of brightness levels.

4. If there is no thin peak, place the black arrow a little to the left of the base of
   the curve.

5. It sometimes improves the contrast to move the grey arrow, which marks the
   output position of the middle value between 0 and 255.

6. The image, still visible behind the histogram box, will change as you move
   the arrows. Experiment with the positions of the white and black arrows
   until the desired result, a light image with good contrast, is achieved.
        You want to achieve a contrast that is high enough to show
           differences in texture and tone, but low enough to avoid creating large
           black areas where the image output values are lowest.

7. When you have achieved a satisfactory level of contrast and brightness, write
   down the arrow positions that produced the best contrast. These will be used
   when you scan the entire image in section 2.3.

8. Select OK from the histogram window to close it.

Figure 2.5 is a graphic representation of the effect of a histogram stretch on
output brightness values.

Morgan M. Robertson                       29                        Creating a DEM
   Hint: A good way to see if even the darkest areas have achieved good contrast is
   to use the magnifying glass icon on the left side of the Adobe screen. Once you
   have performed the histogram stretch, you can magnify “problem” areas of the
   image. The scale is indicated on the title bar of the image. Click with the
   magnifying glass icon on the darkest area of the photo until the scale reads 1:1; if
   you can imagine picking out detail in three dimensions from the image you see at
   1:1, than your image is fine. To minify, press Alt and click.

                       Figure 2.5: “stretching” the brightness values.
2.3 Scanning the image
Time: approximately 1 hour 15 min. for the entire process.

Now that you have positioned the photograph correctly on the scanning plate
(section 2.1), and found appropriate settings for brightness and contrast (section
2.2), you are ready to scan the entire image. This section will lead you through
the process of scanning the entire image into the computer, performing a
histogram stretch on it, rotating A4-scanned images to the correct orientation,
and finally saving it to the hard disk. Procedures are presented for two types of
computers, 386-speed and 486-speed. If you are using a 386 computer, read
only section 2.3.1. If you are using a 486 computer, read only section 2.3.1.

Morgan M. Robertson                              30                      Creating a DEM
2.3.1 Scanning with 386 computers
386 computers are relatively slow, and are prone to breakdown when dealing
with files as large as these. If you are working on a 386-speed computer, you
will have to leave Adobe and enter AVR independently in order to scan. The
reason for this is that slower machines have trouble bringing entire (54
Megabyte!) images through to Adobe using TWAIN as an intermediary, so you
cannot scan the image by accessing the scan software from within Adobe.


1. Do not return to AVR from Adobe, instead Quit Adobe.

2. Enter AVR from the Windows Program Manager screen.

3. Set Brightness to the value recorded in section 2.2.4.

4. Make sure that the Scan Mode is still set to 256 Greyscale and that the
   Resolution is still set to 800 dpi.

5. Preview the image.

6. Expand the crop box to cover the image; it will be a 54 Megabyte image, so
   be sure that you have at least 160 Megabytes free on your C: drive.

7. Press Scan.

   Saving the image in TIFF:

8. Select Save from the File menu. The Save dialog-box will prompt you for a
   filename. The filename for the image should probably indicate which study
   site the image is from, as well as the photograph number (e.g.

9. The File Format selector must be on TIFF, as the DVP will accept no other

10.Exit AVR and enter Adobe.

Morgan M. Robertson                     31                      Creating a DEM
   Adjusting contrast:

11. Select Open from the File menu and bring the newly-scanned image up for

12. Select Adjust from the Image menu, and then Levels. Adobe will take
  some time to construct a histogram.

13. Perform a histogram stretch, using the values you recorded in section 2.2.5.
  This will take some time for the computer to think about, as the software has
  to manipulate the brightness value of approximately 16 million pixels.

14. When the image is satisfactory, select OK, and Adobe will save the
  histogram settings. This will take some time.

   Rotating images scanned on A4 scanners:
   * skip this section if you have used an A3 scanner

15. Under the Image menu in Adobe, select Rotate.

16. If, on the photograph, the flight direction (indicated by the information bar)
  is to the east, select 90 CW (90 degrees clockwise).

17. If the flight direction is westward, select 90 CCW (90 degrees

18. Adobe will require approximately 15 minutes to execute this command.

   Saving the image again:

19. Select Save from the File menu.

20. In the Save dialog-box, although the image has already been saved in TIFF,
  make sure that the Save File as Format Type selector is on TIFF(*.TIF).

21. Make sure that the Drives: selector is set to the C: drive.

22. Make a note of which directory you are saving the image to.

23. Select OK.

24. In the TIFF Options window, make sure the Byte Order selector is on IBM

Morgan M. Robertson                     32                        Creating a DEM
25. Select OK. Adobe will save the document to the hard disk.

You have now scanned the photograph, and saved the file containing the image
on the scanning computer. Scanning with 486 computers
If you are working on a 486-speed computer, congratulations; the process is
faster and simpler. On 486 computers are fast enough so that TWAIN ought
not to crash whilst moving the 54 megabyte file between AVR and Adobe.

   Scanning the image:

1. Acquire the AVR screen through Adobe.

2. Make sure the Brightness is adjusted to the setting that you found most
   satisfactory in your sample scans.

3. Expand the crop box to cover the entire image. The image will occupy 54
   Megabytes of memory, so be sure that there are at least 110 Megabytes of
   memory available on your C: drive.

4. Press Scan. The greyscale scan takes approximately 10 minutes.
   Warning: Occasionally TWAIN will break down and be unable to
   transport such a large image between AVR and Adobe. In these cases,
   you must exit Windows and restart the scanning process.

5. When the image is scanned, Exit the AVR window.

   Adjusting contrast:

6. Back in Adobe, select Adjust from the Image menu, and then Levels.
   Adobe will require 3-4 minutes to construct a histogram.

7. Set the arrows below the histogram (use the values you recorded in section
   2.2.5 as having produced a satisfactory image)

8. Select OK when the image is satisfactory. The saving process will take
   about 10 minutes.

Morgan M. Robertson                    33                       Creating a DEM
   Rotating images scanned on A4 scanners:
   * skip this section if you used an A3 scanner

9. Under the Image menu in Adobe, select Rotate.

10. If the flight direction is eastward, then select 90 CW (90 degrees

11. If the flight direction is westward, select 90 CCW (90 degrees

12. Adobe will require approximately 15 minutes to execute this command.

   Saving the image file in TIFF:

13. Select Save from the File menu.

14. In the Save dialog-box, the Save File as Format Type selector should be
  set to TIFF(*.TIF).

15. Make sure that the Drives: selector is set to the C: drive.

16. Make a note of which directory you are saving the image to.

17. Name the file (the name should reflect the area shown and/or the
  photograph number), and select OK.

18. The TIFF Options window will appear. The Byte Order selector should be
  on IBM PC.

19. Select OK. Adobe will require approximately 10 minutes to save the

You have now scanned the photograph, and saved the file containing the image
on the scanning computer.

Morgan M. Robertson                     34                        Creating a DEM
2.4 Transporting large files
Time: 15-20 minutes.

Before you can use the DVP to analyse the digitised image, you have to move
the image file onto the computer which houses the DVP software. This section
will describe how to move the image file from the scanning computer to either
the DVP computer or a mode of long-term storage.

2.4.1 Storage options
The following are your options for storing the image file.
1. Store it in the scanning computer. Since all available computer space is
   needed for the scanning and rotation processes, this is not recommended.

2. Store it on the DVP computer. If you are planning to build a model with this
   image in the near future, this may be a wise choice. The DVP computer can
   hold only seven images, so it may not be able to hold all the images of the
   gradsect you are working on.

3. Store it in a directory on one of the UNIX net computers (TAMMAR, TROFIS,
   or DINGO). This has the advantage of getting the image out of your hair,
   while ensuring that it is readily accessible. Keep in mind that the net
   computers are not an endless wellspring of disk space, and please consult
   with the UNIX operator before loading them onto one of his/her computers.

4. Store it on magnetic tape. The DVP computer is equipped to accept input
   from magnetic tapes. These are a very efficient way of storing large amounts
   of data (images from an entire gradsect will fit on one tape!), but they are
   relatively clumsy and time-consuming to load data onto and off of. They are
   a good place to store images that you don’t anticipate using soon, or images
   that you have already finished modelling.

The execution of the last three options is described in detail below, but will
require familiarity with the File Transfer Protocol (FTP) method of transporting
files, described below.

2.4.2 File Transfer Protocol
Scanned images occupy very large files. Extremely large files. Moving them
between computers can be much like getting an oil tanker to do a three-point

Morgan M. Robertson                     35                       Creating a DEM
turn in Sydney Harbour. Fortunately, there is a very easy way.

In File Transfer Protocol (FTP), you establish a relationship between a local
directory in your computer and a remote directory in, not surprisingly, a remote
computer. Once this connection is established, you can get files (transport them
from the remote directory to the local one) or put files (transport them from the
local directory to the remote one).

The following commands will be needed:
  To enter FTP mode:
 Type ftp (computername) from a PC’s C:\NET or C:\NET\NCSA
  directory to open an FTP connection with a remote computer (the remote
  computer must have an active net connection). You will then be in FTP

    Within FTP mode:
   bin: enters the binary transfer mode.
   pwd: tells you what the remote directory is.
   lcd (directory): allows you to change the local directory.
   cd (directory): allows you to change the remote directory. For example,
    typing cd .. will move up one directory in the remote computer (note the
    space between cd and ..).
   get (filename): retrieves a file from the remote directory and puts it in the
    local directory.
   put (filename): moves a file from the local directory into the remote
   quit exits the FTP mode.

All computers hooked up to the network have identification names; the network
computers are named TAMMAR, TROFIS and DINGO, the scanning computer is
named KOOKABURRA, and the DVP computer is named HUNTSMAN (although
these last two may change in the future; check to be sure).

So, to open an FTP connection between the scanning computer and the DVP
computer, you would type ftp huntsman from the C:\NET\NCSA directory of
the scanning computer. To open the FTP relationship between two computers,

Morgan M. Robertson                      36                        Creating a DEM
both computers need to be in actively connected to the network. Therefore, the
DVP computer would have to be on, and logged in to one of the three network
computers for the “ftp huntsman” command to have any effect (see below).

2.4.3 Moving image file to the DVP computer from the scanning computer
Since both computers must have an open net connection, and you can’t be in
two places at once, you must prepare the DVP computer and leave it on while
you go back to the scanning computer and execute the FTP commands.
 Turn on the DVP computer.
 Select pcnfs from the Startup menu. The DVP software and the network
  software use the same interrupt and cannot be run on the same computer
  unless one of the two is disabled during startup. If you selected normal, you
  would have been able to run the DVP software, but would not have been able
  to access the network.
 Enter the C:\NET directory.
 Open a net connection by typing telnet tammar.
 Leave it in this condition while you go back to the local computer and
  execute the FTP commands.

On KOOKABURRA (the scanning computer):
 Go to the C:\NET\NCSA directory and type ftp huntsman.
 You will be asked to log on, and then you are presented with an ftp> prompt.
 Type bin to enter binary transfer mode.
 Select the appropriate directories on both computers with the cd and lcd
 put the image from KOOKABURRA to TAMMAR.
 Type quit to exit FTP mode.
 The image file can then be deleted from the scanning computer to make
  room for more images.

    Hint: The transfer can also be done from HUNTSMAN’s C:\NET directory by
    typing ftp kookaburra. The relationship is reversed; you will be getting the file
    rather than putting it, and the local and remote directories are reversed, but
    otherwise the procedure is the same.

Which directory in HUNTSMAN you place it in is up to you. In the past, each
study area has had its own subdirectory within the DVP directory (e.g.

Morgan M. Robertson                        37                        Creating a DEM
2.4.4 Moving image files from the scanning computer to a net computer
If you want to store the image files in one of the net computers’ directories, be
sure and ask the UNIX operator which directory would be appropriate.
    To transfer:
   From the scanning computer’s C:\NET\NCSA directory, type ftp tammar.
   Set the transfer mode to binary.
   Set the local and remote directories.
   put the image file in the net computer’s directory.
   Type quit to exit FTP mode.
   The image file can then be deleted from the scanning computer to make
    room for more images.

Later, when you need to use them, you can type ftp tammar from the DVP
computer’s C:\NET directory, and get the image files into a directory in the
DVP computer.

2.4.5 Storing the image files on magnetic tape
Only files on a network computer can be stored on magnetic tape, because only
these computers have the appropriate hardware attachments. Therefore the first
thing that must be done is to transfer the image file from the scanning computer
to either a network computer or the DVP computer (see above).

The magnetic tape hardware can only be operated by the UNIX operator, so
once the files are on a network computer, the UNIX operator will store them on
tape for you. Let him/her know which files are to be stored, in which network
directory you have put them.

When you need to retrieve the files from tape storage, request the UNIX
operator to transfer them from tape to a network drive. You may then use FTP
to transfer them from the network directory to the DVP computer.

Morgan M. Robertson                     38                       Creating a DEM
                                     Chapter 3


Before starting this chapter, you should have transported two correctly oriented
overlapping images in a directory in the DVP computer. You are ready to begin
constructing a three-dimensional digital model.

This chapter will lead you through the process of bringing the two images that
you will use for your model into correct orientation so that they create a three-
dimensional image when viewed through the stereoscope attachment, and so
that they can accurately display the real-world coordinates of the features they
contain. This is a necessary step before you can begin to capture data from the
model. Since there are several coordinate systems involved (those of the
photograph, the screen, the real world, etc.), this process used to take
photogrammetrists days of painstaking work. The ease of orienting
photographs is one of the primary advantages of using the DVP.

While accomplished DVP users can orient a model in half an hour, learners
should allocate at least a day or two to learn the Orientation process.

Two “bugs” (program errors) have been found in the Orientation process,
neither of which are fatal. The Leica representative has suggested ignoring the
bugs. Procedures for getting around the bugs are given as they occur. One is
described in the next section, and the other occurs in the Internal Orientation
section (section 3.5). These are the only bugs which have been found in the
DVP program.

Morgan M. Robertson                     39                        Creating a DEM
3.1 Structure of the DVP
Time: 5 min.

To begin work on the DVP, you must turn on the PC, the two monitors, and the
digitizing tablet. Make sure that the dongle is inserted into the printer port
of the PC. Without it, the DVP will not function.

The PC will present you with a Startup Menu. You can choose to execute a
“normal” start, or a “pcnfs” start. If you plan to work on the DVP, choose
normal. If you plan to access the network, choose pcnfs. The reason for the
choice is that the DVP program and the PCNFS program both use the same
interrupt, and cannot be run on the same computer, unless you disable one or
the other during the computer startup. If you use the network and then want to
work on the DVP, you have to press Ctrl+Alt+Delete together, a key
combination which will reboot the computer and bring up the Startup Menu

Enter the DVP program by typing dvp from any directory. The introductory
screen shows the main menu. Occasionally, the program may “hang” (freeze
up) before reaching the introductory screen. In these cases, press (together)

The DVP has two main functional modes, which are accessible from the main
menu: Orientation mode and Vectorization mode. These modes cannot be run
simultaneously; to get to Orientation mode from Vectorization mode you need
to exit to the main menu, and vice versa. Orientation mode allows you to
correctly orient the coordinate system of the image with the coordinate systems
of the screen, of the overlapping photo, and of the world. Vectorization mode
allows you to capture data by marking, on the screen, the locations of image

Morgan M. Robertson                    40                       Creating a DEM
Think of these two modes as two paths leading away from your starting point,
and you will have comprehended the basic structure of the software. Once you
have gone down one path, you cannot reach the destination offered by the other
unless you return to the starting point (the Main Menu).

                 Figure 3.1: basic structure of the DVP menu system.

3.2 Steps to Orientation
Time: 5 min.

The goal of Orientation is to “line up” the five coordinate systems mentioned in
the steps below as accurately as possible so that the three-dimensional model
corresponds to the actual terrain as closely as possible. The five coordinate
systems “translate” between the real world and the screen of the DVP

There are five steps towards orienting your photo:
1. Naming or choosing the model

2. Photo Referencing: orienting the digitizing tablet’s coordinate system with
   the photographs’ coordinate system.

3. Internal Orientation: orienting the photographs’ coordinate system with the
   scanned image’s coordinate system (matching millimetres to pixels).

Morgan M. Robertson                      41                        Creating a DEM
4. Relative Orientation: orienting the scanned image’s coordinate system with
   the adjacent overlapping scanned image’s coordinate system (matching pixel
   1a with pixel 1b).

5. Absolute Orientation: orienting the scanned images’ coordinate system with
   the world’s coordinate system (matching pixels to Australian Map Grid

Once you have performed all these orientations, the computer will be able to
translate between pixels on the screen and meters on the ground.

3.2.1 The reference photograph
Select one of the two photographs that compose the model and lay it under the
protective surface of the digitizing tablet, so that it covers the blank space in the
upper right. Lay it in the correct orientation. This is your reference
photograph. It is your physical guide to the digital model and will assist you in
moving around, but keep in mind that the model covers only about 60% of the
reference photo (the area of overlap).

3.2.2 Getting around in the DVP
You can make selections from menus on the DVP in a number of ways.
 You can move the digitizing puck and press the yellow button to select, but
   the puck accompanying this machine is temperamental and the buttons are
   occasionally over- or under-sensitive. You may have to press three or four
   times to get a response; alternately one light press of the button may send you
   whizzing through the next two menus.
 You can also use the arrow keys to move and the Enter key to select.
 Finally, you can press the key which is capitalised or highlighted in the menu
   selection; e.g. Press E for “corrElate”. Macros
Furthermore, if you examine the digitizing tablet you will notice that there are

Morgan M. Robertson                      42                        Creating a DEM
boxed areas with functions written within. These are macros. If you press the
green button on the digitizing puck, the DVP will enter a mode in which it
responds to commands from the puck (see section The menu for
commands executable by the puck appears in the lower section of the monitor.

In this mode, you can move the puck over a box on the digitizing tablet, press
the green button again, and the DVP will execute the command written within
the box. While this manual generally gives the keyboard procedure for
executing various commands, they can often be executed at the click of a puck
button using a macro.

There are many empty boxes left on the digitizing tablet. In these you can
define your own macros. Please refer to section 3.70 in the Leica manual for
instructions on how to accomplish this.

3.3 Naming or choosing a model
Time: 5 min.

You now know enough about the structure and function of the DVP to begin
the Orientation process. You should still be in the Main Menu.

First you must tell the DVP the name of the model and where to find the images
that it will consist of.
To name a model:

1. Select option Model from the menu. The prompt Model name: will appear.

2. Select a name which includes the numbers of both photographs, and perhaps
   the study site as well; e.g. 023-024 or bald2324. If you are returning to work
   on a model you have already made, type its name now.

3. You will then be asked that kind of model it is: DVP, SPOT image, or
   rectified orthogonal image. CSIRO’s system can only work with DVP
   models, so select DVP.

Morgan M. Robertson                    43                        Creating a DEM
4. The software asks for a file containing scanner corrections. At the time of
   writing, there is no such file for the CSIRO A4 scanner. If no file exists,
   press Enter.

5. At the Model description: prompt, you can type a short (68 character)
   description of the model. This is not necessary, but simply for your

6. If you are choosing a model you have worked on before, you will return to
   the Main Menu at this point. If you are making a new model, the software
   will confirm with you that it is a new model by asking New model (y/n):. A
   no answer will return you to your starting point in the Main Menu. A yes
   will cause the software to ask you where to find the images.

7. At the Pathname of the left image: prompt, you should type the complete
   pathname, starting from C:\, of the left image (the DVP will assume that the
   right image is in the same directory as the left image, and when it prompts
   you for the right image pathname, you will only have to type in the

This completed, the DVP returns you to the Main Menu.

3.4 Photo reference
Time: 10-15 min.

In this section you will let the software know the precise location of the
reference photograph on the digitizing tablet by using the Reference function.
To match the tablet coordinates with the photograph coordinates, you will need
to match two points on the photograph to two points on the tablet. When you
mark identical points on objects with two radically different coordinate
systems, in this case the reference photograph and the image file in the
computer, the computer can then easily impose the same coordinate system on
them both. This is the concept behind the entire orientation process.

To perform a Photo Reference:
 Select Ref. from the Main Menu.
 If the model is new, the DVP will ask which photo is being used as the
  reference photo, the left one or the right one.

Morgan M. Robertson                     44                       Creating a DEM
 If you are returning to work on a model you have already started, the DVP
  will also include the option Same, in case you have not moved the reference
  photograph from the digitizing tablet since the last time you worked, and the
  reference coordinates are still the same.
 If you choose Same, the DVP will return you to the Main Menu and you are
  ready for the next step.
 The DVP will then display the upper left corner of the reference photograph
  on the screen.
   Hint: This is your first glance at the image as displayed on the DVP; if it looks
   too dark or too light, you may want to re-scan the image before proceeding.

 The DVP now tells you to select a point on the reference photo (on the
  tablet) that is visible on the screen. Do this as precisely as possible. As the
  fiducial marks are usually visible in the photo referencing process, they
  should be used.
      1. Move the puck to the fiducial mark in the upper left corner of the
          photo, so that the crosshairs meet at the tiny dot in the centre of the
      2. Click the yellow button. You have marked the point on the reference
          photograph that you will match on the image. A red cross will then
          appear on the screen.
      3. Move the cross, using the puck, to the centre of the fiducial mark
          displayed on the screen, so that its crosshairs meet at the small dot.
      4. Click the yellow button again.
      5. The DVP will then “drive” you (move the display to a different part
          of the model) to the lower right corner of the photo and prompt you to
          repeat the process with the lower right fiducial mark.
 Occasionally, the fiducial marks are not visible. Simply use whatever easily
  identifiable point is visible on the screen.

Once this is completed, the DVP will return you to the Main Menu.

3.5 Internal Orientation
Time: 15 min.

The Internal Orientation function brings the photographs and the images into
the same coordinate system by matching points on the reference photo with
points on the scanned image. The fiducial marks are used to link the image’s
coordinate system (in pixels) to the coordinate system of the reference
photograph (in millimetres).

Morgan M. Robertson                       45                        Creating a DEM
3.5.1 Getting around a bug
Select the Orient. option from the Main Menu, and the DVP will enter the
Orientation mode after a bit of hard drive activity.
   Hint: You can always simply select Orient. or Vectorize upon entering the
   DVP without selecting Model or Ref.; the software will ask you the relevant
   questions pertaining to model selection or photo reference in the course of
   entering the Vectorizing or Orientation mode.

What you should see is a split screen, with a small portion of each photo
displayed on each half, and a red floating mark at the center of each half. But it
is very likely that this is not what you will see.

The software has a bug which makes it fail to display the images in the
Orientation mode the first time around. You will probably see the menu bar at
the bottom of a dark screen. The Leica representative in Sydney, Brian
Nichols, has been unable to explain this bug, but since is it not fatal to the
operation of the DVP he has advised us to ignore it. To get around the bug, you
must exit the Orientation mode by selecting Menu from the Orientation menu;
this will get you back out to the Main Menu. Then, re-enter the Orientation
mode, and the proper display should appear.

3.5.2 Making a camera file
Once successfully in the Orientation mode, select Inter. from the Orientation
menu. The purpose of the Internal Orientation is to match the coordinate
system of the reference photograph, in millimetres, with the coordinate system
of the image, in pixels. The fiducial marks are there to provide the reference
photograph with a coordinate system. They have a known coordinate location
relative to the exact center of the photograph, and their coordinates are recorded
on the camera calibration certificate (see section 1.2). The camera file into

Morgan M. Robertson                       46                       Creating a DEM
which you will enter these coordinates allows the DVP to impose a coordinate
system on the reference photograph.

Examine the camera calibration certificate. In the “Fiducial Marks” section of
the certificate, you will find both the coordinates of the marks, and a diagram
telling you which fiducial mark is #1, which is #2, and so on.

To give these coordinates to the DVP, you need to create a Camera File:
1. Select cAmera.

2. The software will ask you for a filename. If a camera file already exists, type
   its name and you will be moved on to the rest of the orientation.

3. If there is no file, enter a name of your own choosing.

4. The DVP will confirm with you that it is a new camera file.

5. The DVP will prompt you for the focal length of the camera. This can be
   found both on the certificate and on the photographs themselves, in the
   camera information box on Side C.

6. The DVP will then ask you for the x,y coordinates of the fiducial marks.
   However, you may have to alter the coordinates. The diagram on the
   certificate shows the photograph and information bar, and the coordinate
   origin. Perhaps Side D is at the top; but if your image is oriented with Side
   B at the top, your fiducial marks won’t have the same coordinates as are
   given in the certificate. On the certificate, mark #1 might be in the lower
   right and have a location of 106.000,-105.999. But on your image, mark #1
   might be in the upper left, in which case its coordinates are -106.000,
   105.999. You must alter the (+) and (-) signs of the certificate’s coordinates
   to suit the rotation of your image. Otherwise the computer will think fiducial
   mark #1 is mark #3, etc.

7. Enter the adjusted coordinates without brackets, separated by commas; e.g. -

8. After you have entered the last fiducial coordinate, the DVP will ask you for
   a fifth coordinate. Simply press Enter.

You will be able to use the same camera file for (at least) all the photographs in

Morgan M. Robertson                     47                       Creating a DEM
the same run (assuming the pilot of the aeroplane didn’t change cameras in
mid-flight!). Furthermore, you will be able to use the same camera file for
every photograph in that entire area, if they were all taken on the same flight.

3.5.3 Recording internal reference points
Now you will pinpoint, for the DVP, the location on the two images of the
coordinates you have just entered.

When you finish entering fiducial coordinates, the program drives you to the
general area of the first fiducial mark. On the split screen will be displayed the
first fiducial marks of the two photographs which comprise the model.
   Hint: Throughout the rest of the Orientation, the split screen will display
   roughly the same area of the two photos which comprise the model. However,
   until you get the two photographs into perfect orientation, the two displayed
   portions will not be of exactly the same areas.

To record the position of the mark, you will need to move each floating mark to
the centre of the fiducial marks on both photographs. There should be a small
(sometimes invisible!) dot at the centre of each fiducial mark (see Figure 3.2);
aim for this.

                          Figure 3.2: the third fiducial mark. Moving the floating mark
You will use the digitizing puck, not the mouse to move the floating mark.
You’ll notice that the floating marks move in tandem, when the puck is moved.
This can be annoying when you need to move the floating marks in opposite

Morgan M. Robertson                       48                      Creating a DEM
directions to arrive at a fiducial mark. The following are some tricks to
manipulating the floating mark which will be useful throughout the orientation
process and beyond.
 To freeze the left mark and move the right, press number 2 on the keyboard
  (not the number pad).
 To freeze the right mark and move the left, press number 1 on the keyboard.
 To release both marks, press number 3.
 To magnify the area around the floating mark, press F8.
 The white button (#2) on the puck toggles the floating mark between 1-pixel
  movement mode ( in which it looks like this: ) and 10-pixel movement
  mode ( ). If you want to move across the screen quickly, 10-pixel jumps
  can be quite useful.

Use these aids to position the marks correctly. Click the yellow button to
record your first internal reference point and the DVP will drive you to the next
fiducial mark.

3.5.4 Photo base
After recording all four fiducial marks in the same manner, the Photo base
prompt will appear, with a default of 90 mm. If you examined the two photos
under a mirror stereoscope, the “photo base” would be the distance between
identical points on the two photos once they were correctly oriented. The
software uses this figure as the starting point for an iterative process, so it
doesn’t need to be precise. For most photographs, 90 mm is an appropriate

3.5.5 Calculation of Internal Orientation
After entering the photo base figure, the DVP will calculate the accuracy of
your placement of the fiducial marks. What it is calculating is how closely your
placement of the fiducial marks with the digitizing puck matched the
dimensions of the coordinates you entered. There will be two sets of
calculations: an orthogonal transformation and an affine transformation. You

Morgan M. Robertson                      49                         Creating a DEM
are interested only in the affine transformation numbers. The measurements are
quite sensitive, and an error of a few pixels away from the exact centre of a
fiducial mark can show up as a large vx or vy (measurements of error).

According to Leica’s representative, the vx and vy figures should be below 15
microns (0.015). However, one of the things that can influence the accuracy of
the Internal Orientation (aside from carelessness on the part of the user!) is the
deviation between the fresh-from-the-camera negative and the photograph that
you actually scanned. Using photographs on paper, that have been shipped
back and forth, taken out in the field, bent, rained on, sneezed on, marked up
with china graph pencils, buried in soft peat for nine months and recycled as
firelighters, and finally scanned on a fairly low-quality scanner, one must
expect the image to have been corrupted somewhat from the original negative.
An error of 30 microns (0.030) is fairly common.

3.5.6 Modifying the Internal Orientation
If your numbers are acceptable, select Save, and the DVP will accept these
numbers for the Internal Orientation. It also asks you if you desire to print out
the results.

If the numbers are not so good, you can do several things.
1)     If the numbers are universally bad, you can select Remove immediately
       after the calculations. The software will then return you to the
       Orientation menu, and you will have to place the fiducial marks again.

2)     If you suspect that only one of the measurements is bad (one very bad
       measurement can throw off all the results):
             Save the results.
             From the Orientation menu, select Intern. again, and then
              Modify. The modification menu will be displayed, along with the
              coordinates for the first point.
             Select Next until the mismeasured point is displayed.
             Select Reject.
             To remeasure it, first choose cEntering to drive to the point.

Morgan M. Robertson                     50                        Creating a DEM
               Press Esc twice to enter the mode in which you can position the
                floating mark.
               Position it accurately, and click the yellow button. The DVP will
                ask you which fiducial mark you have just re-recorded.

3)       If the results are all bad, you can Save them and then select redO from
         the Internal orientation menu. This erases everything, and you have to
         go back to the first step, which was to select a camera file (section 3.5.2)

When you have accepted the results of an Internal Orientation calculation, you
are ready to perform the Relative Orientation.
     Hint: At the bottom right of your screen are a series of letters spelling IRA.
     These stand for Internal, Relative, and Absolute, for the three orientations you
     must perform. As you finish each orientation, that letter will become highlighted.
     If you are not sure whether or not the computer accepted the orientation you just
     performed, check to see if the appropriate letter is highlighted. When all three
     are highlighted, you will know that you have successfully completed Orientation,
     and it is safe to move on to the Vectorizing mode.

3.6 Relative Orientation
Time: 1 hour

3.6.1 Selecting the method of pointing
Performing a relative orientation brings the coordinate systems of the two
images into alignment by identifying at least six identical points on both
images. Point X on image 1 is matched to Point X on image 2; Point Y1 is
matched to Point Y2; etc.

These “control points” should be as widely distributed as possible. If the points
are all clustered in one area, the rest of the model is subject to greater and
greater inaccuracy, the farther away one moves from the cluster of points. The
ideal distribution is thus:

Morgan M. Robertson                         51                         Creating a DEM
               Figure 3.3: proper distribution of relative control points.
The DVP was designed for work in urban areas, where the chance of finding
reliable clear features that can be seen on both images at each of these six
points is extremely good. Therefore, it will helpfully drive you automatically to
areas of the photograph which will maximise the dispersal of the control points,
and therefore the accuracy of the orientation. However, since you are working
with images with few outstanding features, you will want to choose for yourself
where to place your control points. So the first thing you must do is to tell this
to the computer:
1. Select Displ., and the menu for Display Options appears.
2. Select the Manual (Pointing) option.
3. The DVP then gives you the option of using six or nine locations, and of
   recording one or two controls at each location. For images of this scale and
   resolution, the A(1pt) option is sufficient (it asks for one control points at
   each of six locations).

3.6.2 Selecting relative control points Getting around
The DVP now lets you select your six points. This section will help you learn
how to use the three modes of controlling the DVP, so that you may move the

Morgan M. Robertson                        52                         Creating a DEM
control marks to identical points, and record each point.

From this stage on, there are three modes of controlling the DVP; the names for
each mode have been invented for convenience’s sake, and will not be found in
the Leica manual:
 In the Menu Mode the menus are active, and you can execute menu
  commands. In this mode you can select menu options using the keyboard or

 The Image Mode allows you to move the floating mark about the image on
  the upper part of the screen. In this mode the menus are inactive, but you
  can move the floating mark about the two images and mark control points on
  the image.

 The Puck Mode activates the Puck menu, which is displayed in the menu bar
  of the monitor (this mode was discussed earlier in section The puck
  menu allows you to execute options by pressing the four buttons of the puck.
  This mode allows you to display different areas of the model on the split
  screen by moving the puck around the reference photograph and clicking the
  green button (the Window option on the puck menu). You can also use the
  green button to execute macro commands.

To “get around” in the DVP, you must be able to move comfortably between
these three modes. The diagram below shows how this is done.

MENU: Accepts menu commands; responds to keyboard.
 Esc /\
  |   |
  \/ F9
IMAGE: Allows floating marks to be moved about image; responds
Btn. 4 /\        to puck and mouse in tandem.
  |     |
 \/   Esc
PUCK: Allows driving to other parts of image, and puck menu commands to
               be executed; responds to puck.

Remember this structure! More about Puck Mode

Morgan M. Robertson                    53                      Creating a DEM
The two grey squares at the bottom right of the screen represent the two images
which combine to make the model. The part of each image currently displayed
on the split screen is indicated by a small white rectangle. When you enter
Puck Mode, you will see two small dark grey rectangles appear, which mark the
location of the puck on each image as it moves about the reference photograph.
As you move the puck around the reference photograph, so move the dark grey
rectangles. As you move around the reference photograph in Puck Mode, keep
an eye on the location of the dark rectangles on the two grey squares to make
sure you don’t leave the boundaries of the model.

The grey squares are also useful in visualising the angle of view of each half of
the split screen. If the white rectangles are nearly in the middle of the left grey
square, you know that you are looking at the features of the image from nearly a
90 degree angle. Driving to the control points
You are ready to select your control points. Look at your two real photographs
and try to find a feature that is distinct on both. You can then drive there in the
“Model” mode, thusly:
1.   Press Esc to leave the Menu mode.
2.   Press Button 4 once to access the puck menu.
3.   Move the puck over the spot you want to go to.
4.   Press button 4 again to select Window.

Once you are near the point, you must try to bring the two floating marks to the
same point. You can do this monocularly or in stereo vision. Monocular pointing
Using the same techniques as you used in Internal Orientation (see section
3.5.3), bring the points together. This time, however, you are moving the marks
in three dimensions. The puck controls the x and y location of the marks, while

Morgan M. Robertson                      54                       Creating a DEM
the mouse (on your left) controls the z location. You’ll notice that the mouse
only moves the marks along one axis; it appears to be moving along the x axis,
but it is actually moving towards and away from you, in three-dimensional
coordinate terms

Along with number keys 1, 2 and 3, and F8 (Zoom) another very useful key is
F5 (Centering). Press this key to center the images on the locations of the

In marking your point, keep in mind that the two images are taken from
different perspectives, and tree tops (for example) can look very different from
different angles (see Figure 3.4).
   Hint: Practice the two methods of monocular pointing.
   1. Center one of the marks on your feature, then move the mouse to bring the
   two marks together. Move them back towards your feature with the puck
   (adjusting the x,y coordinate position). One may still be off-target, so move the
   marks together again with the mouse (adjusting the height). When they are
   together on your feature, click the yellow puck button to record that point.
      Hint, cont’d:
   2. Center one of the marks on your feature. Freeze it using 1 or 2. Move the
   other mark to that feature. Click the yellow puck button to record your point.
        The second method cannot be used in the vectorizing mode, so it might be a
   good idea to practice using the first method.
        If you have the marks resting on the same feature in both images, and after
   moving the puck they are no longer resting on identical features, this indicates
   that the elevation has changed between the first puck position and the second.

Morgan M. Robertson                       55                        Creating a DEM
   Figure 3.4: The same tree from two perspectives, with identical points A and B.

The DVP was designed for urban work, where easily-identifiable features are
common and precise (building corners, road paint stripes, etc.). In forested
terrain, using 1:25000 photos, the texture is unavoidably rather fuzzy, and
precise features are hard to come by. Roads work very well, as do isolated
trees. Each pixel is important, so be sure that you are on precisely the same
point on each image. Each pixel of discrepancy equals 0.8 metres of error on
the ground. While this may make some urban surveyors faint with distress, it’s
not that bad of an error at this scale. Still, mind your pixels. Stereoscopic pointing
Stereo viewing allows you to compile the two images into one in your brain and
end up with a three-dimensional result (in the same way that your brain takes
two images with different perspectives from your two eyes, and lets you see in
three dimensions). Instead of looking back and forth between two images to
make sure you have selected the same point, you can see the two images as one
three-dimensional image, and when the floating marks have reached the same

Morgan M. Robertson                      56                        Creating a DEM
point, there will appear to be only one mark.
   Hint: If you are having trouble bringing the marks together, you are probably
   not using the mouse properly. Only the mouse can bring the floating marks
   together at a single point.

To use stereoscopic pointing:

1. Loosen the upper bolt on the stereoviewer arm, and lower it so that you can
   look through it comfortably, then retighten the bolt.

2. Look through it. Notice that the 3D image your brain compiles also contains
   the floating mark.

3. Manipulate the floating marks using the mouse and puck, until they arrive at
   identical locations

4. Press F5 to center the two images (this is a very important step!). Without
   centering both images on their respective floating marks you will not be able
   to view in stereo.
         If you focus on the features of the image and there appear to be two
          floating marks, then the two marks are not on identical points.
         If you focus on the marks (bringing them together to see just one
          mark) and it appears to be in front of or behind the image, then the
          marks are, again, not on identical points (see Figure 3.5).
         Only when there appears to be only one mark when you are training
          your focus on the forest canopy (or whatever surface is shown), are
          the marks lined up. The mark will appear to have settled on the
          surface of the feature. Problems of stereo pointing
The trick is to learn to look “beyond” the computer screen. You must trick your
eyes into believing they are looking down from a great height onto a forest;
then you can use the puck and mouse in conjunction to bring them together onto
your feature.

Morgan M. Robertson                     57                        Creating a DEM
                      Figure 3.5: parallax in stereoscopic pointing.
Play around with it; each person has different capacities for stereo vision, and
you may find it particularly easy to place your point in three dimensions using
the viewer. But be warned... it is not nearly as easy as using a traditional
stereoscope and looking at two entire photos. The stereo image has repeatedly
proven to be difficult to achieve, even by experienced photogrammetrists, and
even more difficult to use reliably as a source of measurement. Factors
contributing to this are the low resolution of the image (800 dpi) and the fact
that only a very small portion of the image is shown at a time. Monocular or Stereo pointing?
Which method you use is a matter of individual preference, but keep in mind
that the great majority of the time you spend working on the DVP will be spent
pointing using one method or the other. Find your preferred method, and stick
with it, for consistency’s sake.

One of the disadvantages of using monocular pointing is that it is difficult,
given the rather fuzzy resolution of 1:25000 photographs of forested areas, to
be sure that you have pinpointed the very same part of the very same tree on

Morgan M. Robertson                        58                          Creating a DEM
both photographs. Stereo pointing adds the quality of depth to help you ensure
that you have brought the two floating marks to the same point. However,
stereo pointing is slower and generally more difficult. Marking the relative control points
Whatever method you use, the basic procedure is much like that of the Internal
1. Drive to your first point in Puck Mode.

2. In Image Mode, bring the marks together at the same point using the mouse
   and puck.

3. Press F9 to enter Menu mode.

4. Before recording the point, you may want to correlate it to see how closely
   your placement of the marks matches the software’s idea of what should be
   going on. To do this, select corrEl., and the DVP will present you with a
   percentage of correlation. If this is above 50%, your placement is accepted.
   You can decide for yourself how good you want your correlations to be, i.e.
   the computer may think a 60% correlation is fine, but you may decide you
   want a higher level of accuracy.

5. Select Record from the Relative Orientation menu to mark the control point.

6. You will be asked for the point number. This is simply a reference for you;
   if you just press Enter the software will number the points by itself, starting
   from #1, by placing a small yellow dot on the grey boxes in the lower right
   to mark the first relative control point.

7. After recording the point you will be back in Image Mode, to drive to the
   area of your next control point.

3.6.3 Computing the Relative Orientation
Once you have recorded six control points (and they should be well spread-out,
at least approximating the          shape), you can select Compute. The screen of
computations will appear. It is the py (error) figures to the right of the
coordinates that you will want to pay attention to. Accuracies of within 3-5
microns (0.003-0.005) have been achieved even with high-scale photos of

Morgan M. Robertson                           59                  Creating a DEM
forested areas, so errors of over 15 microns should probably be looked at again.
The procedure for redoing one or all points in the Relative Orientation is the
same as that described for the Interior Orientation (see section 3.5.6).

3.7 Absolute Orientation
Time: if you have already acquired points in the field, 1 hour.

When you finish the Relative Orientation, you have constructed your three-
dimensional model. Congratulations! You should now be able to drive to any
point on the model and (once you have corrected elevation with the mouse, then
pressing F5 to center) view it in stereo. But the model is just floating there in
space, with no connection between its coordinates and the coordinates of what
it represents (that is, the Earth).

Just as you “pinned” one image to the other with six identical points, you will
now “pin” the model to the Earth at points in the model for which you know the
terrestrial coordinates. This is the most difficult orientation to perform, as the
largest opportunity for error lies here. You will have to have already taken a
bare minimum of four GPS and altimeter readings at sites well-distributed
around the model (see section 1.3). Unfortunately, it will still be difficult to
achieve a good Absolute Orientation; the accuracy of a GPS receiver in
isolation is not that great, they often cannot pick up a signal under even a
moderate canopy, and the chances of finding six or seven well spread-out points
in each model that you can physically go to and take readings at are sometimes

When you have accepted the computations for Relative orientation, the DVP
moves you back out into the main Orientation menu, with cursor highlight
already helpfully placed on Absol..

3.7.1 Importing absolute control points

Morgan M. Robertson                                 60            Creating a DEM
       If you already have constructed adjacent models (e.g. you have
completed vectorizing model 024-023, and are now working on model 023-
022), you can use the vectorized graphic elements of the previous (overlapping)
model as absolute control points. To do this:
1. Select Superimp. from the Absolute Orientation menu.
2. Select the FileNameXyz option.
3. The DVP will ask for the *.XYZ file from the previous model which
   contains all the vectorized graphic elements.

See section in the DVP manual for more information on this
procedure. Do this only if you trust your skill at vectorizing! If you have
carelessly vectorized your last model, the absolute orientation will suffer in this
model. It’s like reincarnation that way.

3.7.2 Recording absolute control points
To achieve Absolute Orientation independently (i.e., without importing
previously marked points), you will use many of the same procedures that you
used in achieving Relative Orientation.
 You start out in the Menu mode in the Absolute Orientation menu
 Move the floating marks in the Image mode to the locations at which you
  have taken GPS and altimeter readings.
 Record them in the Menu mode.
 Drive to the next point in the Puck mode.

Unlike relative orientation, you cannot correlate your marks. You also cannot
freeze one mark and move the other around on the x and y axis (although you
can still move it independently on the z axis). This is because you have already
fixed the two images in relation to each other, and moving the marks
independently would violate the relative orientation. If you can’t get the
floating marks to match up exactly because of this restriction, you will know
that your relative orientation wasn’t as accurate as it could have been. Point Numbers and Ground Files

Morgan M. Robertson                       61                      Creating a DEM
When you record your first mark, the software will ask you to name a file to
keep the coordinates of the absolute control points in (a “groundfile”).
   Hint: It is wise to give all the various files connected to your model the same
   name. Since they all have different tags, their 8-character names can all be that
   of the model, i.e. 208-207.IOR, 208-207.XYZ, 208-207.VAR. The ground
   control file has an *.XYZ tag. A list of all the files associated with a completed
   model, and their tags, can be found on page 6-6 of the DVP manual.

The DVP will then ask you to give this point a number, and finally it will ask
you for the coordinates with the following prompt: (X,Y,Z/X,Y/Z/*):. This
prompt indicates that you can enter the full three-dimensional coordinates of the
point, or just the x and y position, or simply use it as an altimetric control point
(z). When entering the z coordinate, remember to add the height of the canopy
if you are marking a point on the canopy instead of the ground beneath it (see

When you press Enter the software will confirm with you that you want the
point saved in the currently selected groundfile. Continue to mark the spots of
GPS readings that were taken in the field. Keep in mind that, although the
DVP will perform an Absolute Orientation with a minimum of four control
points, you should have at least six or seven well-distributed points.

3.5.3 Computation of Absolute Orientation
When you have marked the sites of at least four GPS/altimeter readings on the
screen (and if you’re only going to use four they had better be very well-
distributed), you are ready to Compute. The results can be quite horrendous.
Unless you are very precise in your placement of the marks, and you have
perhaps 8 evenly-distributed points, it is likely that you will get errors of up to
150.000 or 200.000. This is, of course, totally unacceptable. The dX, dY and
dZ figures are in metres, so this is an error of 150 to 200 metres. Most good
photogrammetric work done by hand has a maximum acceptable error of 20-25

Morgan M. Robertson                       62                         Creating a DEM
Urban surveyors, using triangulated GPS receivers and points that have been
surveyed hundreds of times, can get errors as low as a few centimetres. But
with this material, equipment and subject matter, the best that can be achieved
is accuracy to within five metres (dX, dY and dZ values of 5).

It may take some experimenting with accepting and rejecting various data
points. On one occasion, after a day of rigorously precise GPS point collection,
the points were entered on the DVP only to find that some error figures were
still around 200 metres. However, after two points (out of nine) were rejected,
all of the error figures fell to around 4.5 metres. It can be done.

When you achieve acceptable levels of error, and Save the computation, the
computer will return you to the Orientation menu. From here you must exit to
the Main menu using the Menu command, and you have completed the
orientation of your model. You are now ready to begin capturing data from it.

Morgan M. Robertson                      63                       Creating a DEM
                                       Chapter 4

                      VECTORIZING ON THE DVP

4.1 What is Vectorizing?
Time: 5 min.

In this chapter you will, at last, learn to utilise the basic function of the DVP,
which is to extract data about an image and record that data as coordinates in a
“vector file”. To do this, you will draw colored lines, dots, arcs, rectangles,
circles and polygons over the features you wish to record, and store the three-
dimensional location of those shapes in a file. The concept is very similar to a
“paintbrush” program or similar art software, except that these lines are known
as “vectors”. Although to produce a digital elevation model (DEM) requires
only the use of individual points, the use of lines and polygons will be
discussed as well, since the location of roads and vegetation types might also be
easily captured using this software.

To understand vectorizing, suppose you are an urban planner. You are assigned
the task of creating a three-dimensional digital map of the city’s street system.
You have two oriented overlapping images of the city on the DVP. In
Vectorizing mode you can trace the outlines of streets, and draw arcs around
corners, until you have traced out the entire street pattern of the area. If you
have kept the floating marks at identical points on both photos while drawing
the lines, the street pattern you create should be in three dimensions.

The coordinates represented by these lines are stored in a vector file, which can
be formatted for export to several types of GIS (Geographic Information
System) software.

If you, the urban planner, are then asked to add the outlines of all buildings to
your street map, you can use a different colour code to represent houses, and yet

Morgan M. Robertson                       64                       Creating a DEM
another one to represent commercial buildings. In this way you can represent
any number of types of features on the same map, and store the information in
the same file.

Your real-life task of creating the DEM is much the same. To do this, you will
use the same technique you have been using since the Relative Orientation.
You will find a feature identifiable on both images, and bring the floating
marks together on that feature. You will then proceed to mark that point with a
small colored dot, the three-dimensional coordinates of which will be stored in
a vector file. If you mark dots all over the surface of the image, a map of the
elevation of the entire area can be produced. You can then draw lines and
polygons in different colours to mark off roads, vegetation types, and bodies of

4.2 Modes and Codes
Time: 30 min.

As you enter the Vectorizing mode, you will be asked to name two files for the
storing of certain kinds of information. You must name a vector file, where the
coordinates of the features you trace will be stored, and you must name a code
file, where the DVP keeps the information which differentiates between classes
of data (building data, street data, vegetation type data, etc.) It stores the
different codes you use to capture data from a model.

   Hint: You can always simply select Vectorize or Orient. upon entering the
   DVP without selecting Model or Ref.; the software will ask you the relevant
   questions pertaining to model selection or photo reference in the course of
   entering the Vectorizing or Orientation mode.

To enter the vectorizing mode of the DVP:

   Entering the Vectorizing mode:
1. From the Main menu you must select Vect.. You will then be in the main
   Vectorization menu.

Morgan M. Robertson                      65                         Creating a DEM
2. Select Vect. again.

   Selecting a vector file:
3. The DVP will ask for the name of a vector file (which it calls an “.XYZ”
   file). Enter an appropriate name, or if you are returning to work on a model
   you have already done some vectorizing on, enter the name of the vector file
   you have already created.

   Selecting a code file:
4. Then the DVP will ask for the name of a codefile. This is the file where the
   DVP keeps the information which differentiates between building vectors,
   street vectors, vegetation type vectors, etc. When you select a code (below),
   you will enter a number. The DVP will go to the codefile to find out what
   kind of features will be represented by that code. Naturally, you can use the
   same codefile for several models, if you are capturing the same kinds of data
   from all of them.
   Hint: You can use the same codefile for several models, if you are capturing the
   same kinds of data from them all.
5. If the codefile is a new one, the DVP confirms this with you.

6. You are then asked to select the code you wish to use first. At this point, you
   should simply press Enter if you are beginning a new model or codefile. If
   you are returning to vectorizing work, select the number of the code you
   wish to capture data with.

7. The DVP will enter Image mode (movement of cursor on image), where you
   will begin the vectorizing process.

   Hint: You have either just created or activated a vector file (step 3 above).
   Several vector files can be used with the same model, or one vector file may
   contain all the information for many models. Since, in creating a DEM, you will
   be collecting only a few kinds of data, it is suggested that you use only one
   vector file per model. Alternatively, you could use one vector file for an entire
   run. The Leica representative recommends against using one vector file for an
   entire gradsect, due to size considerations. The vector files will eventually be
   exported and analysed on other software, and it is best to export it in relatively
   “small bites”.
        If you use one vector file for several models, due to a small amount (10%)
   of model overlap, when you start a new model 10% of it will already be
   vectorized for you!

4.2.1 Getting around in Vectorization
The menu structure of the Vectorization mode is particularly hard to
conceptualise. There is a laminated schematic diagram of the menu structure in

Morgan M. Robertson                       66                         Creating a DEM
the front of the Leica manual which may help.

As in Orientation, there are three modes of operation in Vectorization: Image,
Menu and Puck. They operate in exactly the same way, except that there are
several levels of menus to deal with. They are structured thusly:
 The Main Vectorization menu allows you to quit, return to the Main Menu,
  and enter the Vectorization menu.
 The Vectorization menu allows you to change codes, edit vectors, choose
  vector types, etc. Many of these commands have subordinate menus. By
  pressing F10 (Main Menu) you can move up to the Main Vectorization
 The various subordinate menus will leave you in Image mode when you
  finish executing their function. Press F9 (Current menu) to return to the
  Vectorization menu.

In Vectorization, as in Orientation, the Image Mode allows you to move the
floating mark about the image. However, it also allows you to enter commands
at the COMMAND: prompt below the split screen. Explanations of the
commands which may be used are found between pages 3-27 and 3-39 in the
Leica manual. The most useful command is \SCRN2, which transfers cursor
control to the second screen (the large monitor to one side of the DVP
computer). The Second Screen
Many of the commands acceptable at the COMMAND: prompt have to do with
the second screen. The major utility of the second screen is that it can display
either image in its entirety, with all vector elements also displayed in their
proper colour. Think of it as an accessory to your reference photograph. When
cursor control has been transferred to the second screen, you can move about
the image, zoom in or out, and store certain parts of the image in a buffer for
quick retrieval. It is not a necessary tool in the construction of the DEM, but it
has certain uses. Chapter 8 of the Leica manual describes the use of the second
screen; see also pages 3-38 and 3-39.

Morgan M. Robertson                      67                        Creating a DEM
4.2.2 Selecting a code
You are left in the Image mode when you finish selecting a vector file and a
code file, so press F9 to bring up the Vectorization menu. Before you can draw
vectors on the screen, you must select a code, and select a colour which will
represent that code.

Select Code from the menu. The DVP will ask for a code number, and then for
a code name. Each code has a number, so the data class “roads” might be Code
3. If you are beginning with a new code file, you might as well type “1” .
The DVP will then comment on the fact that this is a new code number, and ask
you for its name (i.e., “canopy elevation”, “roads”, “type 2 veg.”, etc.). If you
have more than a few codes, you might write down a list like this to help you
remember the various codes you make:

Code    Name
1       Roads
2       Canopy Elevation
3       Type 1a vegetation
4       Type 2 vegetation
5       Water

        ... and so on.

Then you must assign a colour to your new code. In the Vectorization menu,
select Var (for “variables”). In the Var Menu, select coLor. There are 32
colours available. Pressing 0 at the prompt will produce a screen displaying the
32 colour options, and allow you to select one of them.
   Hint: Some colours show up better than others Furthermore, pale yellow is the
   default colour for a vector in the process of being drawn (when it is finished, it
   takes on the colour of the code to which it belongs), and so it is best not to assign
   any of the yellow colours to any vectors. If you are assigning the colour for the
   elevation model points, make sure they stand out brightly, because the points are
   quite small. Bright red is a good colour for elevation points.

When you have selected a code and a colour, the DVP returns you to the Image

Morgan M. Robertson                         68                         Creating a DEM
4.2.3 Selecting a mode
You can now draw vectors on the screen, but you have not selected what kind
of vectors you will draw. You can use points, two kinds of lines, three kinds of
rectangles, two kinds of arcs, and two kinds of circles in the process of drawing
vectors on the screen. These are all known as “vector modes”. To select which
one of these you will use, you must select Mode from the Vectorization menu.
   Hint: You can use the same code in many different vector modes. If you are
   tracing a road in “line” mode, and you come to a roundabout, you can switch
   modes to “circle” and draw the roundabout on the screen, then switch back to
   “line” and continue with the road.

For the purposes of constructing the DEM, you will be using only “point” mode
and two kinds of “line” mode. Instruction in using these two modes are found
in the next section. Please refer to Chapter 7 of the Leica manual for discussion
of other modes.

When you have finished selecting the mode and the code, the status displays at
the bottom of the screen should indicate several things:
 What vector file the vector coordinates are being stored in.
 What code you are using
 What colour is being used for that code (the letters that spell out the name of
  the code should be in that colour).
 What vector mode you are using.

4.3 Placing an element
Time: 45 min. to learn.

After you select the vector mode, the DVP reverts to Image Mode. You are
now ready to start placing vector elements on the screen. At this point, it
becomes vital to understand the difference between “vector” and “vector
element”. A “vector” is the line joining two points that you have marked with
the digitizing puck. A “vector element” is a sequential collection of vectors,

Morgan M. Robertson                     69                       Creating a DEM
which is completed when you stop vectorizing a feature and move onto another.

Think of vectors as letters of the alphabet, and vector elements as the words
constructed from those letters. While you can edit and delete individual
vectors, the smallest unit recognised by the DVP is the vector element. It
simply classifies the points you mark as part of one vector element, until you
select a command such as Clse or Opn, which will be discussed below, in the
same way as we use the space bar to separate words when typing.

4.3.1 Lines
With images of forested areas at 1:25000 scale, all features besides elevation
are best captured using “line” mode. If you have selected “line” mode, you will
be describing either an open or closed polygon, with either a “point-by-point
line” or a “stream line”. Figure 4.1 illustrates the options available.

                      Figure 4.1 diagrams of the two line modes Open and closed polygons with point-by-point lines
For almost all features you will encounter, point-by-point line mode is preferred
over stream line mode. At the scale at which you will be working (1:25000),
the angularity of point-by-point lines will not be noticed even when tracing very
non-geometric features.
 Bring the floating marks together at the starting point or corner of your
 Press the yellow button to mark that point, and a small pale yellow dot will
 Move the floating marks to the next corner, make sure they are both on the
  same spot.
 Click the yellow button again. A pale yellow line will appear between Point
  #1 and Point #2 (this is a vector).

Morgan M. Robertson                         70                    Creating a DEM
 Go on to describe the rest of your feature in this fashion, as if playing
  “connect the dots”.
 If you make a mistake, you can delete as many segments as you want by
  pressing the Delete key on your keyboard. Creating closed polygons
If you are tracing around the edge of a feature (i.e., creating a closed polygon),
do not try to manually join the line to its starting point. Rather, after you have
marked the last corner, like so:

                       Figure 4.2: vectorizing a closed polygon.
press F9 to bring up the Vectorization menu, and select cmD. From this
Command menu, select Clse, and a vector will be drawn between the last point
you marked and the starting point of the figure. When you select Clse, you
have created one vector element, and it will take on the colour of the code you
have selected. You may then move on to begin vectorizing a new feature,
select a new code, etc. Creating open polygons
You may instead be tracing a feature that does not close upon itself, such as a
road or shoreline (i.e., creating an open polygon). In this case, when you reach
the end of the feature, or you want to stop tracing it for a while, select cmD
from the Vectorization menu, and then select Opn. The line will take on the
colour of the selected code, and you can move on to other operations.

You can describe a single feature with one element or many, depending on its

Morgan M. Robertson                       71                       Creating a DEM
length and the nature of the features. If the feature you are working on is a road
that extends for the length of the model, it may be very inconvenient to record it
all in one vector element. You may want to vectorize it in many elements so
that if you have to correct your placement of the road, the vector you edit is of
negotiable size. Stream lines
If you are tracing a meandering feature such as a creek bed or track, you can
use the “stream” function. By pressing the Caps Lock key before marking
your starting point, you can capture the feature simply by moving the floating
marks along its path. A meandering line in pale yellow is left in the path of the
floating marks. To finish this line, select the same Opn command that you used
to finish an open polygon.

This option comes poorly recommended for two reasons:
 It is very hard to keep the floating marks together while tracing along a
  meandering feature if the elevation is changing. You will often end up
  creating small “squiggles” in the line as you try to get a fix on the elevation
  by adjusting the marks.

 At the scale at which you will probably be working (1:25000), the increased
  accuracy made possible by using meandering rather than straight lines is of
  no use whatsoever. The joining up of enough straight segments, at such a
  scale, appears just as meandering as the use of the stream function.

You can increase or decrease the sensitivity of the “stream” line to the
movements of the puck by pressing F4. It is hard to keep your hand completely
steady in tracing a meandering line, and if the sensitivity is set to a very high
level, then unintended hand movements can be recorded in the vector that you

4.3.2 Points
To record a point, select Point from the Mode submenu of the Vectorization
menu. Then, in Image Mode, bring the floating marks together on a feature and

Morgan M. Robertson                      72                        Creating a DEM
click the yellow button.

4.4 Editing an element
Time: 20 min. to learn.

Once a vector element takes on the colour of its code, you have finished with it,
and you move on to other vector elements. But what if you want to go back to
delete it, or add on to it?

4.4.1 Snapping
If you want to add on to an open polygon line that you have been tracing you
must perform what the DVP rather inexplicably calls “snapping” to a point.
This means to go to a point that you have previously marked, and “activate” it
so that you can draw more vectors away from it.
1. Position the floating marks as close as possible to the point on the vector
   element that you want to extend from (it doesn’t have to be the endpoint; you
   can “snap” to any point that you have marked, be it in the middle or end of
   the vector element, and use that as the beginning point of a new vector

2. From the Vectorization menu, select cmD.

3. Select the Snap option from the menu that appears.

4. If you did not place the floating marks close enough to the point, the DVP
   will exclaim “Snap seeking: Element not found !!!”. Try to place the marks
   closer, and make sure that you have the elevation adjusted correctly.

5. If you placed the floating marks close enough to a point, the DVP will
   highlight that element and ask (badly-translated from the Quebeçois) “Is this
   the good element?”, and present you with several options.
         The Next option rejects this element in favour of the next closest
         The End option accepts the selected element, and “activates” the
          point at the end of the element nearest the floating marks the
          beginning of the new vector element.
         The Vertex option is only used with circles or arcs.
         The neaRest option accepts the selected element, and “activates” the
          point nearest to the floating marks.

Morgan M. Robertson                    73                       Creating a DEM
Once you have “snapped” to a point, you can use it as the starting point of a
new vector element.

4.4.2 Deleting elements
To delete an element that you have already completed:
1.   Place the floating marks very near it just as in the “snapping” procedure.
2.   From the Vectorization menu, select Edit.
3.   From the Edit menu, select Delete.
4.   The DVP will search for the “good element” in the same manner as the
     “snap” search.
          If it highlights an element, it will ask if it has found the right element,
            and you can either choose the Next element, or Accept the one the
            DVP has found.
          If you choose Accept, the entire element disappears, and its
            coordinates are deleted from the vector file.

4.5 Other Features
Time: 2-3 hours to learn.

Many vectorizing features are available as macro commands on the digitising
tablet. Experiment with them.
Among the most useful are:
         Line: enters line mode.
         Points: enters point mode.
         Open: executes Opn command (described in section
         Close: executes Clse command (described in section
         Code: allows you to change the code you are vectorizing in.
         Image off/on: makes the screen show only vectors, with no image
         behind them.
         Vecteur off/on: makes the screen show only the image, hiding the
         Code 1: changes the code to code 1.
         Code 2: changes the code to code 2.
         Screen 2: gives puck control to the cursor on the larger screen.

4.5.1 Making codes visible and invisible
Occasionally you may find that the display of several codes at once on the
screen is distracting. For example, if you are working on vectorizing a forestry

Morgan M. Robertson                        74                        Creating a DEM
track, you may not want to see the numerous tiny dots marking elevation
scattered all over the model. It is possible to display only a selected set of the
codes at a time.
1. Select Var from the Vectorization menu.
2. Select Code.
        To make a group of codes invisible, select Invisible, and enter the
          upper and lower limits of the code group to be made invisible,
          separated by a space or a return (e.g., entering 3 5 at the “Enter code
          boundaries” prompt makes codes 3, 4 and 5 invisible).
        To make a group of codes visible, select Visible and follow the same

4.5.2 Vector weight and line type
On the DVP screen, all vectorized lines are the same thickness, and all are
solid. However, for output purposes these characteristics can be varied. If you
wish to use a heavier line to vectorize a certain class of data, you can use the
“weight” function to ensure that it appears thicker on whatever editing or
plotting software you analyse the data on (e.g., AutoCAD). Furthermore, you
can use up to eight different types of lines, which are shown on page 7-16 of
the Leica manual. Again, this will not make a line appear different on the DVP
screen, only on the software you proceed to analyse the vector file on.
To change the weight of a line, select Weight from the Var submenu of the
Vectorization menu. The DVP will ask you for a number in meters. For a line
to be 1mm wide on an image at a scale of 1:25000, the weight value should be
25. For 2mm, the weight value should be 50, and so on.

To change the line type, select Type from the Var submenu of the
Vectorization menu. Enter the number corresponding to the line type you want,
as shown in figure 7-19 of the Leica manual.

4.5.3 Text
You can enter text to accompany your digital map, and place it next to features
or vector elements you wish to describe.

Morgan M. Robertson                      75                        Creating a DEM
First you must set the parameters for the text (angle, height, etc.). Enter the
Var submenu of the Vectorization menu, and select tExt. The various
parameters are displayed for you to manipulate.

To enter the text itself:
1.   Enter the Mode submenu of the Vectorization menu.
2.   Select tExt.
3.   Position the floating marks near the desired location.
4.   Press the yellow button on the puck.
5.   Type text.
6.   When you press Enter, the text-entry mode will terminate.

Editing text is much like editing vectors. To edit text:
1. Place the floating marks very close to the text.
2. Select Edit from the Vectorization menu.
3. Select Info from the Edit submenu.
4. If you have placed the floating marks near enough to the text, the DVP will
   highlight the text and ask “Is this the good element?”.
5. Select ModTxt to make changes to the text.

4.5.4 Importing other vector files
Since adjacent models overlap, you can save yourself some time and trouble by
importing the vector elements from an overlapping model that you have already
vectorized, and continuing on from where you left off on the previous model.
This avoids redundancy, thus avoiding redundancy. Although the overlap is
seldom more than 10%, it is definitely worth it in terms of time expenditure.

The imported files are known as “secondary files” or “reference files”, and their
vectors can be displayed on the screen along with the vectors of the active
vector file (all in one colour, to distinguish them from vectors saved in the
currently active vector file). They are an immense help in showing you what
you have already done. You can edit them and “snap” to them just as you
would do with normal vectors, so that features that cross several images can be

Morgan M. Robertson                     76                        Creating a DEM
smoothly vectorized. Up to ten vector files can be displayed in reference.

Unfortunately, the accuracy of imported vectors is only as good as the similarity
of the orientation of the model from which the imported vectors came to the
orientation of the model on which you are currently working. That is, if the
Absolute Orientation isn’t precisely the same for both models, the imported
vectors will be inaccurately placed on the current model.

To display vector files in reference:
1. Select File from the Vectorizing menu.

2. Select SecFile from the File menu.

3. The Secondary File menu displays several options. The DVP has searched
   the active directory for *.XYZ files, and displays the first one it comes
   across below the menu, along with the status of variables such as “visibility”
   and “colour”.
        If this is not the vector file you want to import, select Next or Prev. to
           select a different vector file in the directory.
        If the reference file shown is already activated and you wish to close
           it, select cLose.
        To toggle the status of the reference file between “visible” and
           “invisible”, select Vis/Inv.
        To assign a colour to the selected reference file, select Colour.
        To activate the selected reference file, select Add.
        To enable or disable your ability to edit or delete the vectors of the
           reference files, select acT.

4.6 Creating the DEM
Time: 1-2 hours to learn.

A digital elevation model is created by recording the height of points all over
the model. If you record enough points, the vector file will eventually contain
enough coordinates to reconstruct the topography of the area. Instead of using
the relatively inefficient DEM function provided by the DVP, you will prepare
a vector file for use with a cartographic software known as ANUDEM.

Morgan M. Robertson                     77                        Creating a DEM
4.6.1 The DEM function on the DVP
While the DVP has a DEM function, it has proven to be ineffective in dealing
with relatively large areas such as CSIRO gradsects. The approach the DVP
uses is to ask you to select a grid value x, and then overlays the entire model
with points x meters apart, prompting you to fix the elevation of each point. If
x=10 meters and it took 10 seconds to fix each point, it would take
approximately 67 days (535 hours) of work to complete just one model. At
nine or ten models per gradsect, this is obviously unrealistic.

Furthermore, there is the concern that there may not be a feature that is
recognisable on both images every x meters; while this isn’t normally a problem
in urban environs, it is a very real concern for images of heavily vegetated areas
at low resolution. Accuracy would suffer greatly if a point had to be placed at
regular intervals, whether there was a feature there or not.
Placing points at regular intervals of greater distance is not an option.
Topographic maps already exist with a 360-meter grain, and to produce any
meaningful advance in accuracy the DVP user should produce a DEM that is an
order of magnitude better than anything that currently exists. This means that
the grain should be 36 meters, so ideally there would be a point every 18

Fortunately, a solution has been found; one that is considerably less labor-
intensive than the method described above.

4.6.1 Building a better DEM
The DVP’s DEM system operates on the premise that to produce an elevation
map in the form of a grid with a point every x meters, you must physically place
a point every x meters. So, to capture elevation data from the hillside
diagrammed below, you must place points as shown in Figure 4.3:

Morgan M. Robertson                      78                       Creating a DEM
              Figure 4.3: placement of points using the DEM function.
But what if you could get away with placing many fewer points? The
alternative system is based on the premise that elevation control points do not
have to fill the dual role of grid intersection points. A computer is quite
capable of making a grid of units x meters wide; all you should have to do is to
fix the elevation at critical control points. If you place elevation control points
only where the slope changes, you would produce a map such as Figure 4.4:

         Figure 4.4: the two systems of placing points for a DEM, compared.
...and a program known as ANUDEM can lay a grid over these points when the

Morgan M. Robertson                      79                        Creating a DEM
vector file is exported from the DVP. The key is that you need to place points
only where the slope changes, not the elevation (in other words, you are
recording the second derivative, not the first).

Since accuracy is a function of how well you record changes in slope, try to
avoid the vectorization shown in Figure 4.5, where slices of hillside have been
excluded from the DEM.

Once the changes in slope are recorded, ANUDEM will overlay a grid, using
the points as elevation control points without requiring them to be grid
intersection points as well.

       Figure 4.5: example of bad vectorization in the construction of a DEM.

4.6.2 Marking elevation control points
Do not use the Dem option in the Var submenu of the Vectorization menu.
Instead, simply enter point mode (select Point from the Mode submenu of the
Vectorization menu) and bring the floating marks together at features
recognisable on both images. Click the yellow puck button to mark the point.

There are a few simple rules to abide by when placing elevation control points:
 Use a different code for placing canopy elevation points than for placing
  ground elevation points. When the vector file is being examined with
  cartographic software, and the image is no longer around to refer back to, it

Morgan M. Robertson                      80                       Creating a DEM
   will be very useful to know whether a point was placed on the ground or in
   the canopy. Don’t hesitate to record whatever features are available and
   identifiable on both images, however, whether on the ground or in the

 Under no circumstances does the distance between two points need to be less
  than 15m. Only place dots as close as 15-20m when digitizing areas of
  extremely sharp relief. You can get distance information from the
  coordinates displayed in purple below the split screen. They give the
  position of the floating mark in AMG coordinates.

 If there are strong ridge/ravine landforms in an area, concentrate most of
  your elevation control points near the tops of the ridges and the bottoms of
  the ravines. It may be hard to distinguish these landforms on the split
  screen; for help, use the reference photograph, or even examine the stereo
  pair underneath a mirror stereoscope.

 In areas of less strong topography, place points more evenly and more
  thickly. While on ridge/ravine systems you may need to concentrate only on
  the extremes, undulating topography without any clear trends requires a more
  thorough coverage of control points. These areas are common and,
  unfortunately, more labor-intensive.

The UNIX operator is very likely to be the person who will operate the
ANUDEM software. Ask him/her for advice on recording the elevation of
“tricky” areas.

Keep in mind that, to the left and/or right of your model, you will overlap with
the next or preceding model. You may wish to leave these areas undigitised if
the images are of poor quality; other images may show the same area in better
quality. By the same token, the tops and bottoms of your models may overlap
with images from successive or preceding runs.
   Hint: Remember that you are recording canopy tops, not ground elevation.
   Unavoidably, some ravines and aspects will be hidden from the DVP; keep this
   in mind when analysing the DEM using cartographic software.

Morgan M. Robertson                    81                       Creating a DEM
4.7 Output
Time: 10 min.

This section is concerned with exporting the vector file from the DVP computer
so that it can be analysed with cartographic software.

The ultimate product of this entire process, from scanning through to
vectorizing, is the vector file, which in a sense is the DEM itself. The last task
to be accomplished before you can begin incorporating the elevation model into
GIS applications is to export it in an appropriate format.

4.7.1 Creating an export file in ASCII
You cannot simply send the vector file as-is to the computer on which
ANUDEM is loaded; the file needs to be converted to the ASCII format. To
execute this, do the following:
1. Select File from the Vectorization menu.
2. Select Imp/Exp.
3. Select Export.
4. Select Ascii. The encoding of the various types of vector information is
   given on page 5-63 of the Leica manual, and an example is shown on the
   following page.
5. You will then be required to give a name to the newly-formatted file. It is all
   right to use the name of the model for the filename, but you must not give it
   the “.XYZ” tag, as this will over-write the original vector file. A good
   choice might be “023-024.ASC”.

4.7.2 Transporting the vector file
You must send the ASCII file to the computer which holds the ANUDEM
software. Review the section on FTP procedure (section 2.5.1). Check with
the UNIX technician about which computer hosts the ANUDEM software, and
in which directory the vector file should be put. Then, open an FTP connection
between the DVP computer and the ANUDEM computer and send the file.

You have now finished using the DVP. Any further work on the DEM will be
performed using other cartographic software.

Morgan M. Robertson                      82                       Creating a DEM

To top