立體視覺匹配(stereo matching)藉由尋找兩視角影像之對應點，逕而估算視差資訊，但可能因為物件遮蔽、深度不連續，同質區域、光線而造成對應不準，無法獲得準確的視差資訊，雖然近年來已發展許多全域和局部的最佳化方法，改進對應不準的問題，但在物件遮蔽區域的大區塊的對應錯誤仍無法以最佳化方法解決，所以遮蔽區域偵測與視差修正變得相當重要。
因此，本論文提出基於匹配代價之非對稱式立體匹配遮蔽偵測演算法，其中包含三項重點，第一，我們利用適應性支持權重演算法(adaptive support-weight algorithm)計算初始匹配代價，以獲得較為準確的視差資訊，第二，我們以非對稱遮蔽偵測為主架構提出匹配代價之遮蔽偵測，並且結合geometry-based uniqueness constraint使遮蔽偵測能以較少計算時間達到一定的準確率。最後，我們在多階層式信息傳遞演算法以加重訊息傳遞的權重於遮蔽改善。實驗結果顯示，我們所提出的基於匹配代價之非對稱式立體匹配遮蔽偵測，相較於left/right checking failure可達到有較低的false negative rate (FNR)，且相較於geometry-based uniqueness constraint可達到有較低的false positive rate (FPR)，而結合多階層式可信度傳遞演算法，的確能有效改善遮蔽區域的視差資訊。

Stereo matching uses two images from different viewpoints to find corresponding points to estimate disparity (depth). However, stereo matching may lead to mismatching due to object occlusion, depth discontinuity, homogonous region and light effect. Although many local and global optimization methods have been proposed to solve the mismatching problem, none of them can solve the mismatching error. Thus, occlusion detection and handling of a wide range of errors are important issues in stereo matching.
Therefore, this paper proposes an asymmetric occlusion detection algorithm using matching cost for stereo matching, which includes three main points. First, we use adaptive support-weight algorithm to compute the initial matching cost, which improves the accuracy of disparity map. Secondly, we propose asymmetric occlusion detection using matching cost, and combine the geometry-based uniqueness constraint to reduce computation time and to achieve accurate detection. Finally, we further handle occlusion by increasing the weight of message propagation in hierarchical belief propagation. Our experimental results show that our proposed method obtains lower false positive rate (FPR) than left/right checking failure, and lower false positive rate (FPR) than geometry-based uniqueness constraint. Moreover, we adopt the hierarchical belief propagation algorithm to refine disparities in occluded regions.

[1] M. Tanimoto and M. Wildeboer, “Frameworks for FTV coding,” in Proceedings of Picture Coding Symposium, pp.1– 4, May 2009.
[2] Y. Boykov, O. Veksler, and R. Zabih, “Fast approximate energy minimization via graph cuts,” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol.23, No.11, pp.1222–1239, Jan. 2001.
[3] J. Sun, H. Y. Shum, and N. N. Zheng , “Stereo matching using belief propagation,” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 25, No.7, pp.787 –800, Nov. 2002.
[4] G. Egnal and R. P. Wildes, “Detecting binocular half-occlusions: empirical comparisons of five approaches,” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol.24, No.8, pp.1127 – 1133, Nov. 2002.
[5] K.J. Yoon and I.S. Kweon, “Adaptive support-weight approach for correspondence search,” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 28, No.4, pp. 650-656, May 2006.
[6] D. A. Forsyth and J.Ponce, Computer vision: a modern approach, Cambridge University Press, London, England, Aug. 2002.
[7] C. Loop and Z. Zhang, “Computing rectifying homographies for stereo vision,” in Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, Vol. 1, pp. 23-25, Jun. 1999
[8] D. Scharstein and R. Szeliski, “A taxonomy and evaluation of dense two-frame stereo correspondence algorithms,” IEEE Workshop on Stereo and Multi-Baseline Vision, pp. 131-140, Dec. 2001.
[9] M. Z. Brown, D. Burschka, and G. D. Hager, “Advances in computational stereo,” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 25, No. 8, pp. 993-1008, Aug. 2003
[10] P. F. Felzenszwalb and D. P. Huttenlocher, “Efficient belief propagation for early vision,” in Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, Vol. 1,pp.261 - 268, Jul. 2004.
[11] Q. Yang, L. Wang, R. Yang, S. Wang, M. Liao, and D. Nister, “ Real-time global stereo matching using hierarchical belief propagation,” in Proceedings of the British Machine Vision Conference, pp. 989-998, Sept. 2006.
[12]C.-K. Liang, C.-C. Cheng, Y.-C. Lai, L.-G. Chen, and H. H. Chen, “Hardware-efficient belief propagation,” in Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, pp. 80–87, Jun. 2009.
[13] W. T. Freeman and E. C. Pasztor, “Learning low-level vision,” in Proceedings of International Conference on Computer Vision, Vol. 2, pp. 1182-1189, Sept. 1999.
[14] Y. Weiss and W. Freeman, “On the optimality of solutions of the max-product belief propagation algorithm in arbitrary graphs,” IEEE Transactions on Information Theory, Vol. 47, No. 2, pp. 732-735, Feb. 2001.
[15] Q.Yang, L.Wang, R.Yang, H.Stewenius, and D.Nister, “Stereo matching with color-weighted correlation, hierarchical belief propagation and occlusion handling,” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 31, No. 3, Mar 2009.
[16] K. Nakayama and S. Shimojo, “Da Vinci stereopsis: Depth and subjective occluding contours from unpaired image points,” Vision Research, Vol.30, No.11, pp.1811–1825, 1990.
[17] S. Barnard and M. Fischler, “Computational stereo,” ACM Computing Surveys, Vol. 14, No. 4, pp. 553–572, Dec. 1982.
[18] C. L. Zitnick and T. Kanade, “ A cooperative algorithm for stereo matching and occlusion dtetection,” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 22, No. 7, pp.675–684, Jul. 2000.
[19] G. Egnal and R.P. Wildes, “Detecting binocular half-occlusions: empirical comparisons of four approaches,” in Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, Vol. 2, pp. 466-473, Aug. 2000.
[20] L. D. Stefano, M. Marchionni, S. Mattoccia, and G. Neri, “Dense stereo based on the uniqueness constraint,” in Proceedings of IEEE International Conference on Pattern Recognition, Vol.3, pp. 657 – 661, Dec. 2002.
[21] F. Boughorbel, “Adaptive filters for depth from stereo and occlusion detection,” in Proceedings of SPIE, Vol. 6803, pp.68030J-1 – 68030J-10, Feb. 2008.
[22] A. S. Ogale and Y. Aloimonos, “Shape and the stereo correspondence problem,” International Journal of Computer Vision, Vol. 65, No. 3, pp. 147-162, Dec. 2005.
[23] S. Ince and J. Konrad, “Geometry-based estimation of occlusions from video frame pairs,” in Proceedings of IEEE International Conference on Acoustics, Speech, and Signal Processing , Vol. 2, pp. 933 – 936, Mar. 2005.
[24] P. M. Jodoin, C. Rosenberger, and M. Mignotte, “Detecting half-occlusion with a fast region-based fusion procedure,” in Proceedings of the British Machine Vision Conference, No.1, pp.417–426, Sept. 2006.
[25] J. Sun, Y. Li, S.B. Kang, and H.-Y. Shum, “Symmetric stereo matching for occlusion handling,” in Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, Vol. 2, pp. 399 – 406, Jun. 2005.
[26] D. Min and K. Sohn, “Stereo matching with asymmetric occlusion handling in weighted least square framework,” in Proceedings of IEEE International Conference on Conference on Acoustics, Speech, and Signal Processing, pp.1061– 1064 , Apr. 2008.
[27] D. Min and K. Sohn, “An asymmetric post-processing for correspondence problem,” Signal Processing:Image Communication, Vol. 25, No. 2, pp.130-142, Feb. 2010.
[28] http://vision.middlebury.edu/
[29] Y. Wei and L. Quan, “Asymmetrical occlusion handling using graph cut for multi-view stereo,” in Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, Vol. 2, pp. 902-909, Jun. 2005.
[30] F. K. Yang, An improved stereo matching algorithm using belief propagation for homogeneous regions, MS thesis, National Central University, Jun. 2009.
[31] S. Srivastava , S. H. Lee , N. I. Cho , S.U. Lee, and J. Park, “Stereo matching using adaptive belief propagation along ambiguity gradient,” in Processing of IEEE International Conference on Image Processing, pp.2085-2088, Nov. 2009.