Abstract：To solve the problem of meeting accuracy and speed requirements simultaneously, a two-step stereo matching method is proposed. According to the local texture features, a transform is adopted to the couple of images, and the reference image is segmented with mean shift algorithm, the first matching based on the support region with arbitrary shape and size was used to form a parallax constraint, then the second matching based on fixed window is adopted to obtain the initial disparity map. Finally, the disparity map is optimized by the different reliabilities classification. The experiment results show that the algorithm has higher matching speed and accuracy.
陈华,张志娟,刘刚,胡春海,王书涛. 基于局部纹理特性和图像分割的分步立体匹配[J]. 计量学报, 2017, 38(1): 73-77.
CHEN Hua,ZHANG Zhi-juan,LIU Gang,HU Chun-hai,WANG Shu-tao. A Two-step Stereo Matching Based on the Local Texture Features and Image Segmentation. Acta Metrologica Sinica, 2017, 38(1): 73-77.
［1］康新,何小元,冯毅.一种新的模板匹配算法及其在三维形貌测量中的应用［J］.计量学报, 2002, 23(2): 90-93.
［2］Scharstein D, Szeliski R. A taxonomy and evaluation of dense two-frame stereo correspondence algorithms［J］. International Journal of Computer Vision, 2002, 47(1-3):7-42.
［3］Gupta R K, Cho S Y. A correlation-based approach for real-time stereo matching［M］. Advances in Visual Computing. Springer Berlin Heidelberg, 2010: 129-138.
［4］Cheng F Y, Zhang H, Yuan D, et al. Stereo matching by using the global edge constraint［J］. Neurocomputing, 2014, 131: 217-226.
［5］Yoon K J, Kweon I S. Adaptive support-weight approach for correspondence search［J］. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2006, 28(4): 650-656.
［6］Nalpantidis L, Gasteratos A. Biologically and psychophysically inspired adaptive support weights algorithm for stereo correspondence［J］. Robotics and Autonomous Systems, 2010, 58(5): 457-464.
［7］Kowalczuk J, Psota E T, Perez L C. Real-time stereo matching on CUDA using an iterative refinement method for adaptive support-weight correspondences［J］. Circuits and Systems for Video Technology, IEEE Transactions on, 2013, 23(1):94-104.
［8］Liu T, Zhang P, Luo L. Dense stereo corresponddence with contrast context histogram, segmentation based two-pass aggregation and occlusion handling［M］.Advanced in Image and Video Technology, Springer Berlin Heidelberg, 2009: 449-461.
［9］de Souza Gazolli K A, Salles E O T. Exploring neighborhood and spatial information for improving scene classification［J］. Pattern Recognition Letters, 2014, 46: 83-88.
［10］Ma L, Li J, Ma J, et al. A modified census transform based on the neighborhood information for stereo- matching algorithm［C］// Image and Graphics(ICIG), 2013 Seventh International Conference on, IEEE, 2013: 533-538.
［11］Tombari F, Mattoccia S, Di Stefano L, et al. Near real-time stereo based on effective cost aggregation［C］ // Pattern Recognition, 2008. ICPR 2008. 19th International Comference on, IEEE, 2008: 1-4.
［12］Wang L, Liu Z, Zhang Z. Feature based stereo matching using two-step expansion［J］. Mathematical Problems in Engineering, 2014.
［13］于英,杨靖宇,张永生,等.一种适应无人机平台的快速立体匹配方法［J］.计量学报, 2014, 35(2): 143-146.
［14］Tombari F, Mattoccia S, Di Stefano L, et al. Classification and evaluation of cost aggregation methods for stereo correspondence［C］// Computer Vision and Pattern Recognition, 2008. CVPR 2008. IEEE Conference on. IEEE, 2008:1-8.
2017, Vol. 38 
