影像的遮蔽偵測為電腦視覺領域的重要研究課題，本論文提出利用匹配代價曲線特徵以進行遮蔽偵測。本論文方案可分成四個步驟，第一部分找出曲線上最低點的左與右有效區域。第二部分以變異數(variance)擷取曲線特徵，再區分為左右變異數較高之T+T區域，及至少有一區為變異數較低之非T+T區域。針對此兩種區域，第三部份各進行不同的遮蔽偵測方案。由於此兩個遮蔽偵測的方法在遮蔽區域誤判率較高，故針對遮蔽區域再使用另一以匹配影像曲線為基礎且結合geometry-based uniqueness constraint (MC-GUC)之遮蔽方案確認，以增強偵測準確率，減少非遮蔽區域之偵測誤判。第四部份，我們結合跟形像學(morphological image processing)改進遮蔽圖，以提高遮蔽偵測準確率。本論文方案相較於MC_GUC，可有效降低非遮蔽區域之遮蔽誤判平均達5.55%，而整體準確率也能提升1.96%。

Occlusion detection is important for applications of computer vision. Thus, this thesis proposes an occlusion detection scheme using features of matching cost curves. The proposed scheme can be divided into four steps. First, we find the left and right effective regions around the disparity that has the lowest cost of the matching cost curve. Secondly, based on the variances of the left and right effective regions, a matching cost curve can be divided into either a T + T region, having large variances in both left and right effective regions, or a non-T + T region, having small variance on at least one effective region. Accordingly, the third part of the proposed scheme applies two different occlusion detection methods on these two kinds of regions. Since the aforementioned occlusion detection methods have high false positive rate in occluded regions, we use another matching cost based occlusion detection algorithm that combines with geometry-based uniqueness constraint (MC-GUC) to enhance detection accuracy and reduce errors on non-occluded regions simultaneously. Finally, we use morphological image processing to improve the accuracy of occlusion map. Compared with MC_GUC, the proposed scheme can effectively reduce error rate around an average of 5.55% on non-occluded regions, while the overall detection accuracy can be improved around 1.96%.

[1]O. Veksler, “Stereo correspondence with compact windows via minimum ratio cycle,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 24, no. 12, pp. 1654-1660, Dec. 2002.
[2]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, Apr. 2006.
[3]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, Aug. 2002.
[4]S.-H. Li, “Asymmetric Occlusion Detection Using Matching Cost for Stereo Matching,” Master thesis, Deptartment of Communication Engineering, National Central University, Taiwan. Jul. 2011.
[5]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.
[6]D. Scharstein, and R. Szeliski, “A taxonomy and evaluation of dense two-frame stereo correspondence algorithms,” International Journal of Computer vision, vol. 47, no. 1-3, pp. 7-42, Apr. 2002.
[7]H. Hirschmuller, “Stereo processing by semiglobal matching and mutual information,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 30, no. 2, pp. 328-341, Feb. 2008.
[8]T. Kanade, and M. Okutomi, “A stereo matching algorithm with an adaptive window: Theory and experiment,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 16, no. 9, pp. 920-932, Sep. 1994.
[9]Y. Boykov, O. Veksler, and R. Zabih, “A variable window approach to early vision,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 20, no. 12, pp. 1283-1294, Dec. 1998.
[10]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, May 1990.
[11]S. T. Barnard, and M. A. Fischler, “Computational stereo,” ACM Computing Surveys, vol. 14, no. 4, pp. 553-572, Dec. 1982.
[12]M. Bleyer, C. Rother, and P. Kohli, “Surface stereo with soft segmentation,” IEEE Conference on Computer Vision and Pattern Recognition, pp. 1570-1577, Jun. 2010.
[13]M. Bleyer, and M. Gelautz, “Graph-cut-based stereo matching using image segmentation with symmetrical treatment of occlusions,” Signal Processing: Image Communication, vol. 22, no. 2, pp. 127-143, Feb. 2007.
[14]Q. Yang, L. Wang, R. Yang et al., “Stereo matching with color-weighted correlation, hierarchical belief propagation, and occlusion handling,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 31, no. 3, pp. 492-504, Mar. 2009.
[15]S. Ince, and J. Konrad, “Geometry-based estimation of occlusions from video frame pairs,” IEEE International Conference on Acoustics, Speech, and Signal Processing, vol. 2, pp. ii/933-ii/936 Vol. 2, Mar. 2005.
[16]C. L. Zitnick, and T. Kanade, “A cooperative algorithm for stereo matching and occlusion detection,” Pattern Analysis and Machine Intelligence, IEEE Transactions on, vol. 22, no. 7, pp. 675-684, Jul. 2000.
[17]W. Xiong, H. S. Chung, and J. Jia, “Fractional stereo matching using expectation-maximization,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 31, no. 3, pp. 428-443, Mar. 2009.
[18]P.-M. Jodoin, C. Rosenberger, and M. Mignotte, “Detecting half-occlusion with a fast region-based fusion procedure,” Proceedings of the British Machine Vision Conference, pp. 417-426, Sep. 2006.
[19]J. D. Oh, S. Ma, and C.-C. Kuo, “Stereo matching via disparity estimation and surface modeling,” IEEE Conference on Computer Vision and Pattern Recognition, pp. 1-8, Jun. 2007.
[20]M. Bleyer, M. Gelautz, C. Rother et al., “A stereo approach that handles the matting problem via image warping,” IEEE Conference on Computer Vision and Pattern Recognition, pp. 501-508, Jun. 2009.
[21]W. Chen, M.-J. Zhang, and Z.-H. Xiong, “Fast semi-global stereo matching via extracting disparity candidates from region boundaries,” IET computer vision, vol. 5, no. 2, pp. 143-150, Mar. 2011.