Set of Disparity Map tutorial scripts for StereoPi board with CM3/3+ inside.
Brief tutorial: article in our Blog
Was tested in the following environment:
In the latest Raspbian kernels stereoscopic support has been occasionally broken by implementing new AWB algorithm. You can read some details here. Our Raspbian image provided has no such a problem, but you can get it after 'apt-get upgrade' or 'rpi-update'.
Current solution: after boot run once this command before accessing your cameras:
sudo vcdbg set awb_mode 0
This will turn AWB algo to the previous mode, and stereo works again untill next reboot. So run this command after reboot, or add it to your autorun script.
Ready-to-use Raspbian Image:
Download Link 2 (Google Drive)
I'd like to thank tuxlifan user for this information!
WARNING for Linux users: please do NOT use unzip to extract the zip file but use "7z e stereopi-opencv-20190405.zip" instead.
Checksum for stereopi-opencv-20190405.zip file is: sha256: 6c7a5714db03aae07aeb009483f3fcd0fa43356280da071fdc3bad31dfe6f0c4 md5: fa82c9b997f6797c8445602b933b6ae4)
md5sum using unzip (and getting an error message about "Bad CRC"): 3ce7a63479ffa65fc923297d1fa01792 md5sum using 7z: 8dbe549e3a467a583a961bad85e45ca1
Press 'Q' key in any script for exit, as we added key stroke processing. Do not kill it by Ctrl+C :)
1_test.py - starts camera preview, save captured image if 'Q' button pressed.
Video: https://www.youtube.com/watch?v=wllLrNUw3SE
2_chess_cycle.py - takes a series of photos for stereopair calibration, shows countdown timer. You need a printed chessboard with 9x6 parameters (file "pattern.png" included).
Video: https://youtu.be/1XCAlU3k-xs
3_pairs_cut.py - just cuts all captured photos to left and right images.
Video: https://youtu.be/95DWmPECbDc
4_calibration.py - calibrate StereoPi cameas setup using pairs from script 4.
Video: https://youtu.be/vtPhu23tKGo
5_dm_tune.py - script for fine tune of disparity map settings.
Video: https://youtu.be/Z4j3NrMyeGE
6_dm_video.py - builds disparity map in real time.
Video: https://youtu.be/f29arVstfZA