StereoMatching

implementation of some basical stereo matching mathods:

Stereo match cost computation

1. absolute differences
2. Squared difference
3. truncated absolute differences
4. truncated squared differences
5. normalized cross-correlation
6. hierachical mutual information

cost aggregation

1. fixed window based
2. shiftable window based
3. segmentation based
4. bilateral filtering based

disparity optimization

1. graph cut based
2. scanline optimization
3. semi-global matching

references

[1] Scharstein D, Szeliski R. A taxonomy and evaluation of dense two-frame stereo correspondence algorithms[J]. International journal of computer vision, 2002, 47(1-3): 7-42.

[2] Hirschmuller H, Scharstein D. Evaluation of cost functions for stereo matching[C]//Computer Vision and Pattern Recognition, 2007. CVPR'07. IEEE Conference on. IEEE, 2007: 1-8.

[3] Hirschmuller H. Stereo processing by semiglobal matching and mutual information[J]. IEEE Transactions on pattern analysis and machine intelligence, 2008, 30(2): 328-341.

[4] Stefano Mattoccia, Stereo Vision:Algorithms and Applications. http://www.vision.deis.unibo.it/smatt

[5] Boykov Y, Veksler O, Zabih R. Fast approximate energy minimization via graph cuts[J]. IEEE Transactions on pattern analysis and machine intelligence, 2001, 23(11): 1222-1239.