本文主要針對立體視覺(stereo vision)領域中，提出了一套植基於自適性加權視窗之立體匹配演算法。根據不同區域特性給予適合的區塊(block)做立體匹配(stereo vision matching)，以上稱為自適性區塊(adaptive block)之匹配。除此之外，本研究也可建立在任何單一區塊立體匹配演算法上，以便做為未來研究者對視差(disparity)的研究。
首先，本文以絕對差值和(Sum of Absolute Differences, SAD)為主的搜尋法，使用五種不同大小範圍的區塊，取得五份視差索引，並透過視差一致性、影像粗糙度資訊、視差投票法與自適性區塊，最後考慮八鄰近的範圍進行視差加強優化，以成功改善了傳統單一區塊方法所得到的視差值的錯誤率。並達到降低視差錯誤率的目的。
為了量測視差的可靠度，將得到的視差值與真實視差(ground truth)，以Middlebury所提供的標準立體圖庫與標準立體匹配評估法進行測式，分別針對非遮蔽區域(non-occluded region, Non-occ)、具有視差區域(all region, All)和視差不連續區域(depth discontinuity region, Disc)做分析與比較。為了驗證本文的方法精確度與可行性，實驗的部分將與一些研究者做比較分析。
最後，於實驗結果得知，本論文方法，確實能夠有效的降低以及改善錯誤率。於未來期望將其發展成一套即時的系統，套用在機器人甚至是監視器上，使目標物移動到特定距離時，則有利於進行其他相關研究。

This study aimed at stereo vision, proposed stereo matching algorithm based on support weight with adaptive block matching. According to characteristics of different regions to give an adaptive block stereo vision matching, these blocks is called adaptive block matching. Besides, this study can also built on any stereo matching algorithms with single block, such as post-processing steps.
In this study, we use sum of absolute differences mainly of search method. Through different blocks sizes to obtain the disparity index. And explore the consistency of disparity, coarseness of image, adaptive disparity voting optimal, and disparity refinement to improve error rate. Finally, consider the eight way neighboring disparity to enhance the optimizing. And successful to improved the error rates of the traditional single-block method. To achieve the purpose of disparity error rate reduction.
In order to more accurately represent the accuracy of this method. In the experiment this will be use the Middlebury database. According to three different regional characteristics includes non-occ、all and disc to analyze and compare.
Finally, the experimental results revealed that the proposed method can effectively reduce and improve the error rate. In the future expect to develop into real-time systems. And can be applied to the robot or even monitor, and improved computer intelligence capabilities to research.

中文摘要 i
ABSTRACT ii
誌謝 iv
目錄 v
圖目錄 vii
表目錄 ix
第一章 緒論 1
1.1 研究背景 1
1.2 研究動機 2
1.3 研究目的 3
1.4 論文架構 4
第二章 相關研究探討 6
2.1 前言 6
2.2 相機校正(camera calibration) 6
2.3 立體視覺(stereo vision) 7
2.3.1 視差法(disparity) 7
2.4立體視覺匹配演算法(stereo matching algorithms) 9
2.4.1計算匹配成本(matching cost computation) 10
2.4.2 成本聚合(cost aggregation) 12
2.4.3 計算視差與視差最佳化(optimization) 13
2.4.4 視差加強優化(disparity refinement) 14
2.5 絕對差值和與平方差值和的比較 16
2.6 固定區塊對於視差之缺點 17
第三章 自適性(adaptive)區塊演算法 21
3.1 前言 21
3.2 視差一致性(Disparity consistency) 22
3.3 視差粗糙度(The Disparity for Smooth Region) 23
3.4 視差投票最佳化(The Vote Disparity between the 8-Neighbors) 27
3.5 視差自適性區塊(The Uniqueness of Disparity) 28
3.6 視差校正(Disparity calibration) 28
第四章 實驗結果 30
4.1 前言 30
4.2 深度影像之效能評估比較與分析 30
4.2.1 標準比較影像 30
4.2.2 其它測試影像 32
4.2.3 評價方式 33
4.3 實驗比較數據與結果 36
第五章 結論 46
參考文獻 47

[1] R. C. Gonzalez, and R. E. Woods, Digital image processing, 2nd ed. Prentice Hall, 2002.
[2] R. Ben-Ari and N. Sochen,“Stereo matching with mumford-shah regularization and occlusion handing,” IEEE Trans. PAMI, vol. 32, pp.2071-2084, 2010.
[3] V. Kolmogorov and R. Zabih,“Computing visual correspondence with occlusions using graph cuts,”in Proc. 8th International Conference on Computer Vision, Vancouver, Canada, pp.508-515, 2001, July 7-14.
[4] Q. Yang, L. Wang, R. Yang, H. Stewénius, and D. Nistér,“Stereo matching with color-weighted correlation, hierarchical belief propagation and occlusion handling,” IEEE Trans. PAMI, vol. 31, no. 3, pp.492-504, 2009.
[5] D. Min and K. Sohn,“Cost aggregation and occlusion handling with WLS in stereo matching,” IEEE Trans. Image Processing, Vol.17, no.8, pp.1431-1442, 2008.
[6] J. Y. Bouguet, Camera Calibration Toolbox for Matlab, http://www.vision.caltech.edu/bouguetj/calib_doc/htmls/example5.html.
[7] M.Z. Brown, D. Burschka, and G.D. Hager. “Advances in computational stereo,” IEEE Trans. PAMI, vol. 25, pp. 993-1008, 2003.
[8] D. Scharstein and R. Szeliski.“A taxonomy and evaluation of dense two-frame stereo correspondence algorithms,”International Journal of Computer Vision, vol. 47, pp. 7-42, 2002.
[9] P.J.M Aarts and V. Laarhoven, “Simulates annealing and applications: Mathematics and its applications,”D. Reidel Publishing Company, 1987.
[10] M.S. Lew, T.S. Huang, and K.W. Wang,“Learning and feature selection in stereo matching,” IEEE Trans. PAMI, vol. 16, no. 9, pp. 869-881, 1994.
[11] R. Zabih and J. Woodfill, “Non-Parametric Local Transforms for Computing Visual Correspondence,” Proc. Third European Conf. Computer Vision, pp. 150-158, 1994.
[12] K. Wang, “Adaptive stereo matching algorithm based on edge detection,” International Conference on Image Processing, vol. 2, pp. 1345-1348, 2004.
[13] O. Veksler, “Stereo correspondence with compact windows via minimum ratio cycle,” IEEE Trans. PAMI, vol. 24, no. 12, pp. 1654–1660, 2002.
[14] O. Veksler, “Fast variable window for stereo correspondence using integral images,” in Proc. IEEE Conf. Computer Vision and Pattern Recognition, vol. 1, 2003, pp. 556–561.
[15] Z. Zhai, Y. Lu, and Hong Zhao, “Stereo matching with adaptive support-pixel set and adaptive support-weight,” in Proc. International Conference on Computational Intelligence and Security, 2009, pp. 598-602.
[16] Q. Yang, L. Wang, R. Yang, H. Stewenius, and D. Nister, “Stereo matching with color-weighted correlation, hierarchical belief propagation and occlusion handling,” IEEE Trans. PAMI, vol. 31, no. 3, pp. 492-504, 2009.
[17] F. Tombari, S. Mattoccia and L. Di Stefano, “Segmentation-based adaptive support- for accurate stereo correspondence,” PSIVT, pp. 427–438, 2007.
[18] K. J. Yoon and I. S. Kweon, “Adaptive support-weight approach for correspondence search,” IEEE Trans. PAMI, vol. 28, no. 4, pp. 650–656, 2006.
[19] K. J. Yoon and I. S. Kweon, “Locally Adaptive support-Weight Approach for Visual Correspondence Search,” in Proc. IEEE Computer Society Conference on Computer Vision and Pattern Recognition, vol. 2, 2005, pp. 924-931.
[20] I. J. Cox, S. L. Hingorani, S. B. Rao, and B. M. Maggs, “A maximum likelihood stereo algorithm,” Computer Vision and Image Understanding, vol. 63, no. 3, pp. 542-567, 1996.
[21] J. Kim, K. Lee, B. Choi, and S. Lee, “A dense stereo matching using two-pass dynamic programming with generalized ground control points,” in Proc. of Conf. on Computer Vision and Pattern Recognition, vol. 2, San Diego, CA, June 20-25, 2005, pp.1075-1082.
[22] O. Veksler, “Stereo correspondence by dynamic programming on a tree,” in Proc. of Conf. on Computer Vision and Pattern Recognition, vol. 2, San Diego, CA, June 20-25, 2005, pp. 384-390.
[23] Z. Xu, L. Ma, M. Kimachi, and M. Suwa, “Efficient contrast invariant stereo correspondence using dynamic programming with vertical constraint,” Visual Computer, vol. 24, no. 1, pp. 45-55, 2008.
[24] M. Gong and Y. Yang, “Fast unambiguous stereo matching using reliability-based dynamic programming,” IEEE Trans. PAMI, vol. 27, no. 6, pp. 998-1003, 2005.
[25] Y. Boykov, O. Veksler, and R. Zabih, “Fast approximate energy minimization via graph cuts,” IEEE Trans. PAMI, vol. 23, no. 11, pp. 1222-1239, 2001.
[26] V. Kolmogorov and R. Zabih, “Computing visual correspondence with occlusions using graph cuts,” in Proc. 8th International Conference on Computer Vision, Vancouver, Canada, July 7-14, 2001, pp. 508-515.
[27] V. Kolmogorov and R. Zabih, “Multi-camera scene reconstruction via graph cuts,” in Proc. 7th European Conference on Computer Vision, Copenhagen, Denmark, May 28-31, 2002, pp. 82-96.
[28] S. Larsen, P. Mordohai, M. Pollefeys, and H. Fuchs, “Temporally consistent reconstruction from multiple video streams using enhanced belief propagation,” in Proc. 11th International Conference on Computer Vision, Rio de Janeiro, Brazil, Oct. 14-21, 2007, pp.1-8.
[29] Q. Yang, L. Wang, R. Yang, S. Wang, M. Liao, and D. Nistér, “Real-time global stereo matching using hierarchical belief propagation,” in Proc. 17th British Machine Vision Conference, Edinburgh, UK, Sep.4-7, 2006, pp. 989-998.
[30] T. Yu, R. Lin, B. Super and B. Tang, “Efficient message representations for belief propagation,” in Proc. 11th International Conference on Computer Vision, Rio de Janeiro, Brazil, Oct. 14-21, 2007, pp.1-8.
[31] F. Tombari, S. Mattoccia, L. D. Stefano, and E. Addimanda, “Classification and evaluation of cost aggregation methods for stereo correspondence,” in Proc. IEEE Int. Conf. on Computer Vision and Pattern Recognition. CVPR, Florida, USA, Dec. 2008.
[32] M. Gerrits and P. Bekaert, “Local stereo matching with segmentation-based outlier rejection,” In Proc. Conf. Computer and Robot Vision, 2006, pp. 66–66.
[33] H. Hirschmuller, P. Innocent, and J. Garibaldi, “Real-time correlation-based stereo vision with reduced border errors,” IJCV, vol. 47, pp. 1–3, 2002.
[34] C. H. Lin, and Y. J. Syu, “Fast segmentation of porcelain images based on texture features,” Journal of Visual Communication and Image Representation, May, 2010.
[35] H. Tamura, S. Mori, and T. Yamawaki, “Texture features corresponding to visual perception,” IEEE Trans. Systems, Man, and Cybernetics, Vol. 8, pp. 460-473, 1978.
[36]Middlebury Stereo Datasets, “http://vision.middlebury.edu/stereo/data/“
[37] D. Scharstein and R. Szeliski, “High-accuracy stereo depth maps using structured light,” In Proc. IEEE CVPR, vol. 1, Madison, WI, June 2003, pp. 195-202.
[38] D. Scharstein and C. Pal, “Learning conditional random fields for stereo,” In Proc. IEEE CVPR, Minneapolis, MN, June 2007.
[39] H. Hirschmüller and D. Scharstein, “Evaluation of cost functions for stereo matching,” In Proc. IEEE CVPR, Minneapolis, MN, June 2007.