GitHub - science-enthusiast/loopyBP at 91e3e127afdc66d57b772d2ebd89aadaaf6c6296

science-enthusiast / loopyBP Public

Notifications You must be signed in to change notification settings
Fork 1
Star 13

Loopy belief propagation for stereo depth estimation. Based on sum product algorithm.

Notifications

Name		Name	Last commit message	Last commit date
Latest commit History 4 Commits
BP.hpp		BP.hpp
Matrix.hpp		Matrix.hpp
README		README
birchTomasDisp.cpp		birchTomasDisp.cpp
birchTomasDisp.hpp		birchTomasDisp.hpp
interpPix.hpp		interpPix.hpp
main.cpp		main.cpp
pottsModel.hpp		pottsModel.hpp

Repository files navigation

This implements sum product belief propagation for stereo depth estimation.

It expects as input three arguments: the left image, the right image (in that order) and maximum disparity in pixels. It is easy to figure out maximum disparity using a tool like gimp and viewing the images as layers.

It works better with higher resolution images. Try images from the middlebury dataset without downsampling.

The code is not yet optimized. Expect slow speed.