Increasing Performance of the Semi-global Matching Stereo Algorithm
Keywords:
stereo-vision, semi-global matching, optimization, real-time algorithmsAbstract
The article describes aspects and parameters of the semi-global matching stereo algorithm applicable for optimization towards decreasing the amount of calculations to perform and increasing the number of disparity image frames generated by the algorithm per unit time. Two approaches are proposed, their pros and cons explained, results of their implementation applied to the test data are displayed, and comparative analysis of bad pixels is presented.References
Hirschmuller H. Accurate and Efficient Stereo Processing by Semi-Global Matching and Mutual Information. Computer Vision and Pattern Recongnition, 2005, vol. 2, pp. 807–814.
Scharstein D., Szeliski R. Taxonomy and Evaluation of Dense Two-frame Stereo Correspondence Algorithms. International Journal of Computer Vision, 2002, vol. 47, pp. 7–42. Available at: http://vision.middlebury.edu/stereo/taxonomy-IJCV.pdf.
Hirshmuller H. Evaluation of Cost Functions for Stereo Matching. Computer Vision and Pattern Recognition, 2007, vol. 0, pp. 1–8. Available at: http://vision.middlebury.edu/~schar/papers/evalCosts_cvpr07.pdf; http://www.dlr.de/rm/en/PortalData/3/Resources/papers/modeler/cvpr05hh.pdf.
Xiang X., Zhang M., Li G., He Y., Pan Z. Real-time stereo matching based on fast belief propagation. Machine Vision and Applications, 2012, vol. 23, pp. 1219–1227. Available at: http://link.springer.com/article/10.1007%2Fs00138-011-0405-1.
Gehrig S.K., Rabe C. Real-time Semi-Global Matching on the CPU. Computer Vision and Pattern Recognition Workshops, 2010, vol. 17, pp. 85–92. Available at: http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=5543779&url=http%3A%2F%2Fieeexplore.ieee.org%2Fxpls%2Fabs_all.jsp%3Farnumber%3D5543779.
