Performance enhancement of Adaptive Support-Weight approach by tuning parameters

Since Yoon and Kweon proposed the Adaptive Support-Weight approach (ASW) based on color similarity and geometric proximity, local stereo matching methods have experienced great improvement, even are better than most of global approaches. In ASW, how to choose proper parameters is a very difficult and ongoing issue. Until now most studies on ASW used empirical parameter values and little work focused on the discussion of the ASW parameter setting. In this paper, we presented a Relational Analysis Method (RAM) to investigate the effects of different parameters on the ASW performance. RAM simultaneously considers the changes of multiple parameters of ASW. Middlebury test platform and four standard test images are applied to conduct experiments. The results show that the different combinations of multiple parameters have significant influences on the ASW performance. The average bad pixels of the four images vary from 45.4% to 8.66%. The ASW performance obtained by using appropriate parameters is much better than several modified ASW methods. Thus through tuning the parameters of ASW, the performance can be greatly improved.

[1]  Andreas Klaus,et al.  Segment-Based Stereo Matching Using Belief Propagation and a Self-Adapting Dissimilarity Measure , 2006, 18th International Conference on Pattern Recognition (ICPR'06).

[2]  Olga Veksler,et al.  Fast approximate energy minimization via graph cuts , 2001, Proceedings of the Seventh IEEE International Conference on Computer Vision.

[3]  Cheng Lei,et al.  Region-Tree Based Stereo Using Dynamic Programming Optimization , 2006, 2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'06).

[4]  Oscar E. Ruiz,et al.  Statistical Tuning of Adaptive-Weight Depth Map Algorithm , 2011, CAIP.

[5]  D. Scharstein,et al.  A Taxonomy and Evaluation of Dense Two-Frame Stereo Correspondence Algorithms , 2001, Proceedings IEEE Workshop on Stereo and Multi-Baseline Vision (SMBV 2001).

[6]  Christopher Joseph Pal,et al.  Learning Conditional Random Fields for Stereo , 2007, 2007 IEEE Conference on Computer Vision and Pattern Recognition.

[7]  Nanning Zheng,et al.  Stereo Matching Using Belief Propagation , 2002, IEEE Trans. Pattern Anal. Mach. Intell..

[8]  Richard Szeliski,et al.  High-accuracy stereo depth maps using structured light , 2003, 2003 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2003. Proceedings..

[9]  Qican Zhang,et al.  Local stereo matching with adaptive support-weight, rank transform and disparity calibration , 2008, Pattern Recognit. Lett..

[10]  In-So Kweon,et al.  Adaptive Support-Weight Approach for Correspondence Search , 2006, IEEE Trans. Pattern Anal. Mach. Intell..

[11]  Federico Tombari,et al.  Segmentation-Based Adaptive Support for Accurate Stereo Correspondence , 2007, PSIVT.