![]() |
CMP Stereo Algorithm Evaluation |
Center for Machine Perception Czech Technical University Prague http://cmp.felk.cvut.cz/ |
Jana Kostliva Radim Sara Jan Cech |
Introduction | Results | Download | Upload |
This page presents results of tested methods. The main page (this page) shows the basic characteristic Efficiency of an algorithm which evaluates overall algorithm performance in range [0,1], where 1 is the best, ideal, algorithm. "SFB" in the table represents Stereo Feasibility Boundary, the boundary reached by all tested algorithms together, i.e. the best possible performance over all tested algorithms.
Detailed evaluation (plots and additional tables with other characteristics) on one scene only will be shown after clicking on the scene name. Clicking on numbers in the table will show results of selected algorithm only on selected scene.
Algorithm | |||||||
SFB | 0.9673 | 0.9892 | 0.9939 | 0.9769 | 0.9574 | 0.9016 | 0.9951 |
SGM [7] | 0.9603 | 0.9848 | 0.9843 | 0.9767 | 0.9511 | 0.8862 | 0.9948 |
WTA [6] | 0.8664 | 0.9211 | 0.9842 | 0.8827 | 0.7902 | 0.7517 | 0.8660 |
SDM [1] | 0.8450 | 0.9124 | 0.9609 | 0.7431 | 0.8517 | 0.7137 | 0.8958 |
CSM [2] | 0.8252 | 0.9406 | 0.9621 | 0.6541 | 0.8223 | 0.7842 | 0.8172 |
DP [3] | 0.6364 | 0.8914 | 0.6091 | 0.2689 | 0.8054 | 0.7597 | 0.8701 |
BP [5] | 0.4516 | 0.5521 | 0.5740 | 0.0903 | 0.2660 | 0.7124 | 0.7723 |
GC [4] | 0.4503 | 0.5293 | 0.4118 | 0.2425 | 0.2409 | 0.7887 | 0.6867 |
[1] | J.Kostkova, R. Sara: Stratified Dense Matching for Stereopsis in Complex Scenes. In BMVC2003, pp.339-348, 2003. [implementation] |
[2] | R. Sara: Finding the Largest Unambiguous component of Stereo Matching. In ECCV2002, pp.900-914, 2002. |
[3] | I. Cox et al.: A Maximum Likelihood Stereo Algorithm. CVIU, 63(3) |
[4] | V. Kolmogorov, R. Zabih: Computing Visual Correspondence with Occlusions Using Graph Cuts. In ICCV2001, pp.508-515,2001. [implementation] |
[5] | M. Tappen, W. Freeman: Comparison of Graph Cuts with Belief Propagation for Stereo, Using Identical MRF Parameters. In ICCV2003, pp.900-907,2003. [implementation] |
[6] | : Winner-takes-all with correlation (implemented by JKostliva). Research Report CTU-CMP-2007-08 Appendix A, 2007. [implementation] |
[7] | H. Hirschmueller: Stereo Processing by Semi-Global Matching and Mutual Information. In IEEE Transactions on Pattern Analysis and Machine Intelligence, Volume 30(2), February 2008, pp. 328-341. |
Go to the cmp homepage! | comments and suggestions to Authors |