F3E-Limbs Prepping High & Low Stereo Directories
Before each subtest regarding the various limb images can be performed a high stereo and low stereo directory must be made. The high and low stereo directories will contain approach, PS, and four DS polycam images. As its name suggests, the high stereo directory's DS polycam images will have a high stereo angle and vice versa for the low stereo angle directory. Both of these directories are supposed to have a 52 by 52 meter area centered on the tag site tiled down to 5cm. Once the tag site is tiled down to 5cm Geometry will be run to update the spacecraft position when the images were taken. Then a 20 by 20 m area of the tag site will be evaluated by recording the RMS value (comparison between truth and effort). The updated spacecraft position and pointing will be recorded as well.
Important Note
The entirety of this test will take place within the /SPC_Test/F3E-Limbs directory on ormacsrv1. When ever an action is performed during the test, commands will be logged for future reference. To do this type the following commands
script date pwd ls -l
Then continue with the procedural step. When done with a procedural step cancel the script and place it into the /SPC_Test/F3E-Limbs/log directory. With script still enabled do the following to end it while in the test directory
exit mv typescript ../log/<name/path>
Procedures
Note that making the high and low stereo directories follow the same preocdures with the exception of different DS polycam images being used.
1. Setting up the test directory
Basically the orig directory is being copied over to the test directory with the test directory being emptied first.
Test F3E Test Directory Set-Up
2. Symbolically Linking the Correct List of Images to PICTLIST.TXT
While in the test directory for high stereo:
rm PICTLIST.TXT ln -s ../lsupport/PICTLISTF3Ehigh.txt PICTLIST.TXT
While in the test directory for low stereo:
rm PICTLIST.TXT ln -s ../lsupport/PICTLISTF3Elow.txt PICTLIST.TXT
3. Register High Stereo or Low Stereo Images
For the high stereo directory (test directory):
imgListName=highRegistering.txt
For the low stereo directory (test directory):
imgListName=highRegistering.txt
4. Tile at 20 cm
bigmapName=TILE20 GSD=0.00020
5. Iterate and Look for correlation errors
Test F3E Iterate -- Use the bigmap scenario
find_nofitP support/logEval.sh
6. Tile at 10 cm
bigmapName=TILE10 GSD=0.00010
7. Iterate and Look for correlation errors
Test F3E Iterate -- Use the bigmap scenario
find_nofitP support/logEval.sh
8. Check basic stability and run Geometry
Geometry is run here to update the spacecraft position so that the calculations with residuals is more accurate. Running Geometry here will also decrease the chance of issues when tiling down to 5 cm.
While executing this procedural block, there were many images that contained 3 landmarks or less. All of the images were afflicted by this issue were not the DS images added however. Nonetheless, all images with less than three images but had some of the landmarks on an image lit limb were fixed with autoregister (adding more landmarks to image). This was done to improve the accuracy of the updated SC position after running Geometry.
This is accomplished by performing the following inputs after RESIDUALS is run:
# makes a list of images to auto register
awk '/\*/ {print $1, $8, $9}' PICINFO.TXT | grep -v " 0" | cut -c -12 > ImagesToAutoAlign.txt
# puts the list in the correct format
vi ImagesToAutoAlign.txt
:%s/P/ P/g
ZZ
echo END >> ImagesToAutoAlign.txt
# sets up auto register
cp ImagesToAutoAlign.txt make_script.in
cp scripts/make_scriptA.seed .
make_scriptAP
nohup sh run.sh &
9. Tile at 5cm
bigmapName=TILE05 GSD=0.00005
Test F3E Bigmap Tiling a Use the bigmap scenario
10. Iterate and Look for correlation errors
Test F3E Iterate -- Use the bigmap scenario
find_nofitP support/logEval.sh
11. Run Geometry for the Final Time
Note that if images contain three landmarks or less then resolve the issue the same way that was used at 10 cm.
12. Stat Collection
- Landmark Center Point Cloud Evaluation:
The goal here is to get central vector for each 5 cm maplet located in a 20m by 20m area around the tag site.
First enter the following:
bigmap < support/EVAL20.in
This will output a list of the landarks within the evaluation region of the tag site. Copy and paste this list into LMKFILES/LMRKEvalList.txt
While in the test directory input the following commands.
grep "0.0500" MAPINFO.TXT | awk '{ print $1 }' >> LMKFILES/LMRK05.txt # Makes a text file of all 5cm landmark names
cd /SPC_Test/F3E-Limbs/test/LMKFILES
while read line; do f=`grep $line LMRKEvalList.txt | wc -l`; if [ $f -gt 0 ]; then echo $line >> EvalLMRK05List.txt; fi; done < LMRK05.txt # Makes a list of 5 cm landmarks in 20 by 20 meter area around tagsites
while read line; do grep "VLM" $line* | awk '{ print $1, $2, $3 }' >> PC05.txt; done < EvalLMRK05List.txt # Makes the point cloud
cp PC05.txt LMRK05.txt ../../results/<subtest name>/PC05.txt # Copy point cloud files to results
cp PC05.txt ~/send/ # Copy point cloud to send directory so it can be evaluated on personal machineNow, these are the exact commands I enter to evaluate the point cloud on my machine.
cp send/PC05.txt send/CompareOBJfiles/ cd /Users/kdrozd/Dropbox\ \(OREx\)/SPC-ORex/eval_tools/altwg-2016.01.25-macosx-x64/bin vi ~/send/ComparOBJfiles/PC05.txt :%s/D/E/g ZZ ./ComparePointCloudToOBJ --compute-optimal-translation-and-rotation ~/send/CompareOBJfiles/PC05.txt truthTAG1-5cm-1.1K.obj
The Copy the output of the last command into a file in the the subtest directory located in the /SPC_Test/F3E-Limbs/results directory.
- Collecting SCOBJ and Pointing Vectors:
Follow the procedural block below.