Docstoc

lidar_exercise_2010_nov

Document Sample
lidar_exercise_2010_nov Powered By Docstoc
					        Airborne Laser Scanning

                          Exercise

Nusret Demir
demir@geod.baug.ethz.ch
                   Outline
• LIDAR data
• SCOP ++
• Exercise tasks
LIDAR Data
     DTM-AV Raw Data of Swisstopo




     Buildings are shown as voids,
     Vegetated regions have low density
                                 DSM Raw Data of Swisstopo




DSM raw data has all points
so it is not a DSM in reality.
Profile views of the off-terrain objects (trees, buildings)
                        Grid DSM / DTM Data of Swisstopo




Grid DSM (left), and grid DTM(right) with hill shading view
               Software Packages
•   SCOP++ LIDAR
•   Terrascan
•   LIDAR Analyst (for ArcGIS)
•   LP360 (for ArcGIS)
•   ALDPAT(free)
•   etc.
***most of the existing packages can be found at http://www.gim-
   international.com/files/productsurvey_v_pdfdocument_12.pdf
                             SCOP++
INPHO’s solution for LIDAR processing is LIDAR Box software bundle
covering the complete workflow of LIDAR based DTM generation

• SCOP++ LIDAR
   • fast & fully automatic filtering process
   • classification of point clouds into
        • terrain/ground points
        • building points
        • off-terrain points
   • consideration of break lines, if available, for improving DTM quality
• DTMaster (Stereo)
   • developed for quality assurance of DTM data
   • combines LIDAR with photogrammetry
       • stereoscopic measurement for absolute quality control
       • acquisition of breaklines for improving the quality of the DTM
                       Exercise

Tasks:


1.Visual comparison between LIDAR DSM and image based DSM

2. DSM generation and derivation of DTM from raw DSM point cloud

3.Quantitative comparison between result DTM/DSM and reference
  DTM/DSM
                     Exercise
Input: Raw DSM LIDAR point cloud & original LIDARDSM/DTM

ca.2 x 3 km
About 2 million points
Create new project
Import the data (raw DSM)
General processing menu - View the data




                            There might be
                            some voids in the
                            data
Task 1 : Visual comparison between LIDAR DSM and image based DSM
Hill shade DSM
                 In ArcGIS
                 Conversion menu :Convert asc to raster
                 3D Analyst-Raster surface menu:Hillshade
LIDAR DSM in hillshade view




                              Selected
                              Area for
                              the
                              exercise
Task 2 : Derivation of DTM and DSM generation from raw DSM LIDAR
point cloud
                       Theory in the background
  STAGE 1: Elimination of buildings


• A. Edge detection
   – Point cloud is tiled into small raster cells
   – Computation of gradient of each raster cell
   – Detection of cells with steep gradients
• B. Region growing to detect area of each object
• C. Points belonging to buildings are excluded from further processing
    STAGE 2: Hierarchic robust filtering
•    Generation of a low resolution data
     pyramid
     using the original data (xyz-
     coordinates of the lowest points in a
     grid)

•    Computation of a low resolution DTM
     using robust interpolation along with
     blunder detection


•    Elimination of LIDAR points outside
     a predefined tolerance band


•    Computation of a DTM with full
     resolution
     using robust interpolation along with
     blunder detection
                      Data Processing

LIDAR filtering workflow


A. Basic settings

   1) Grid width :
        e.g. 1m for LIDAR Data with 1 point / sqm
   1) Mean Accuracy:
        e.g. 0.050 m; depends on the flying height and surface type
   1) Filtering factor:
        controls the weight function and tolerances
   1)   (Computation time: less significant; default value recommended )
                        Data Processing

LIDAR filtering workflow

B. Defining filter
strategy
  Several predefined, effective
  strategies are available:

 For classifying points
 • Lidar Default :  three output files : ground points, building points , other off-terrain
   points,
 • Lidar Default Strong:  more points classified as off-terrain points
                 Strong
 • Lidar Default Weak :  more points classified as terrain points

 For computing DTM or DSM
 • Lidar DSM :  result is digital surface model (DSM)
 • Lidar DTM Default:  result is digital terrain model (DTM)
             Default
                  Data Processing
What is the best strategy for my project?




 • The default strategy will give the best result if you need a very
   precise DTM. (Case for the exercise)
 • In areas with little vegetation, the weak strategy delivers better
   results.
    Manual Process Setting for Reduction of
    DSM to DTM with robust filtering
 • Main Steps
 EliminateBuildings
 ThinOut:
 The input : a set of points
 the output : the reduced set of
     points.
 SortOut :
 comparison of points to a DTM.
 If the points are within a certain
     distance from the DTM they
     are accepted, otherwise they
     are rejected.
 Filter
 Interpolate
 Classify (not necessary for the
     exercise)
Choosing the strategy already defines the parameters,
but they can always be modified by the user
Derivation of DTM from raw DSM LIDAR point cloud
EliminateBuildings
   •The input file is sorted into small quadratic cells of
   size Cell size (m).

   •The slope (percentage) between
   neighboring cells are computed.

   •Groups of cells with low slopes surrounded by cells
   with slope >= Minimal slope are assumed to be a
   building,
   if the group of cells form an area of size >= Minimal
   area(m2).
                                          ThinOut
                                             the parameter has to be
                                             twice the size of the average point distance (m).




                                                Recommended
                                                If flying height is too high
                                                 because in this case, there are more
                                                blunders in the data.

*If the data is too big and the memory is not sufficient, “cell size” parameter can be
tuned bigger. (NOT NECESSARY FOR THE EXERCISE)
     Filter &Interpolation




                                                                            The size of the computing unit
                                                                            is 200m by 200m (=(21-1)*10).

Points below the averaging surface
:ground points
Points which are more than 1,5m above
the averaging surface :off-terrain points
                                              For the exercise, It is not necessary to change the parameters,
penetration rate:The penetration portion of
ground points under tree canopy. It is        if the final result is not sufficient, try to change the strategy first
specified in percent
                                              (weak, strong, etc.)
                                              if the result still is not sufficient,
                                              in case , there are still non-filtered objects in the final DTM result,
                                              decrease the upper tolerance in the upper brunch menu
                            SortOut




the original laser points are compared to the DTM of the previous step. The
units are meter.
Classify

*it is not necessary for the exercise
Exports the ground,vegetation,building
points in the separate files
- Export DTM file
DTM visualization - Import DTM file to ArcGIS

                  Conversion from asc to raster
DTM result
DSM generation from raw point cloud
DSM interpolation - one level robust interpolation
DSM interpolation - one level robust interpolation
DSM interpolation - one level robust interpolation
Interpolated DSM




It only eliminates blunders, gross errors.
Changing the parameters is not necessary for the exercise
DSM interpolation - Export DSM file
DSM visualization - Import DSM file to ArcGIS

                  Conversion from asc to raster
Interpolated DSM




                   the voids can be filled
Interpolated DSM




                   select the position after
                   the filter step (here 2)
final DSM
Task 3 : Quantitative comparison between result DTM/DSM and reference
DTM/DSM
       Import DTM to ArcGIS
(apply same steps for DSM as well)



      Conversion from ASCII to raster
                                           DTM from
                                           SCOP++




                                          Give a
                                          name to
                                          your DTM
                                          result


                                        Should be
                                        set in
                                        FLOAT
                   Analysis of the DTM /DSM

Comparison of z values
• 3D Analyst Tools- Raster Math - Minus
                                               Result




                                                Reference




                                              Give a
                                              file name
                                         4.Click
               2.Select       3.Choose
               “Classified”   a color-
1.Click here                  bar
           Residual
           statistics




1.select range          2.Click
values                  OK
Select Layout view
Add Title and Legend
1                            2




3.After 2 more clicks, you
have a legend
                                 5.Follow these menus:
4. On the legend, right
                                 Legend items-style-
click-properties
                                 properties-General
Select
“show
labels”
Residual map




               Residuals
• Thank you for your attention

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:8
posted:8/20/2012
language:Unknown
pages:54