Dear Bubble Counter:
Please observe the prismatic stickers around the perimeters of the polished slabs, (see
Figure 1). We use these stickers as reference points. Please, if possible, do not remove
the stickers. Here is a short description of the purpose of the stickers:
Our microscope is mounted above a motorized stage. When the microscope image is
centered on the cross-over point at the center of a sticker, we record the stage
coordinates. We do this for each sticker. We also record the stage coordinates at each
stop during the manual point count, and record the phase identified at each stop (air,
aggregate, or paste). For automated analyses, we use a flatbed scanner to collect the
image of the polished surface, (instead of a microscope). We locate each sticker in the
scanned image, and record the x,y pixel coordinates of the cross-over points. A
geometric relationship is established between the set of sticker pixel coordinates and the
set of sticker stage coordinates. With the relationship, phases identified at x,y coordinate
stops during the manual point count can be directly compared to phases identified at
corresponding x,y pixel locations in the scanned image.
Figure 1: Prismatic sticker along edge of polished surface, (left) and close-up of same
sticker with cross-over point at center of image (right).
We would like to make the same comparisons with the manual and/or automated analyses
done at your laboratory. If you wish to help, we request the following:
If you are performing a manual point count, we ask, if practical, for you to record the
x,y stage coordinates of the cross-over point at the center of each sticker, along with
the x,y stage coordinates of the stops from your manual point count, (along with the
phase identified at each stop). Use whatever coordinate system you have established.
If you are performing a manual linear traverse, we ask, if practical, for you to record
the x,y stage coordinates of the cross-over point at the center of each sticker, along
with the x,y stage coordinates of the lines describing the traverse intercepts, (along
with the phase identified for each intercept). Use whatever coordinate system you
have established.
If you are performing an automated point count, we ask, if practical, for you to record
a single classified image of the entire area analyzed, (each pixel classified as either
air, aggregate, or paste). Also, we ask you to record the coordinates of the cross-over
point at the center of each sticker, along with the coordinates of the corners of your
classified image. Use whatever coordinate system you have established.
With the information listed above, it will be possible to compare stops or intercepts
recorded during the manual analyses to the corresponding locations in the classified
images from the automated analyses. It will also be possible to compare pixels identified
using one automated method to corresponding pixels identified using another automated
method.
Since the details of the various automated systems and manual systems are unknown to
us, we do not know what kind of data is practical for you to provide. Please provide any
data you feel comfortable with. Along with this letter, we have posted the text files with
all of the data from our manual point counts at this internet address:
http://www.cee.mtu.edu/%7Ekrpeters/public/FHWA_KANSAS/
Each sample has three files associated with it: a file with the coordinates for the stickers,
a file with the point count coordinates and phase identified at each stop, and a file with
the final ASTM C 457 results. We have not done the automated trials yet. We will do
the automated trials later when the samples are returned to us.
Happy bubble counting!
Karl Peterson, Senior Bubble Counter II
Tranportation Materials Research Center
Michigan Technological University
1400 Townsend Drive
Houghton, MI: 49931-1295 USA
email: krpeters@mtu.edu
phone: 1.906.487.2423