三维重建中立体匹配算法的分析-模式识别与智能系统专业论文.docxVIP

哈尔滨工业大学工学硕士学位论文 哈尔滨工业大学工学硕士学位论文 - - II - Abstract In recent years, more and more computer vision researchers pay attention to 3D reconstruction based on images or videos, especially images. Stereo matching is one of the most important problems in the 3D reconstruction. This thesis aims to find better stereo matching algorithms. We focused on template based stereo matching algorithm using Graph cut (TGC), including the implementation and analysis of the typical stereo matching algorithms and some improvements about TGC based on the basic framework of stereo matching algorithms. In the first part, four typical stereo matching algorithms are implemented and the results are analyzed, then we improve them. The algorithms include WTA and real-time which are local stereo matching algorithms, and two global ones, they are Dynamic Programming and Simulated Annealing. Then we find BT algorithm is a useful technique in stereo matching algorithms, and the global ones take too more times, so efficient global stereo matching algorithms are needed, especially for practical applications. Based on the idea of graph cut, we proposed a global stereo matching algorithm-- template based stereo matching algorithm using Graph cut (TGC) which needs three steps to implement: the segmentation of reference image , disparity planar estimation, templates assigning to disparities using graph cut. Then we give the PBM and running time of several algorithms on the Middlebury test set, the proposed (TGC), graph cut, dynamic programming, simulated annealing, winner-take-all, and real-time. Experimental results show TGC can obtain satisfactory stereo matching results. In the last part, we do four experiments to improve TGC based on the basic framework of stereo matching algorithms. The first improvement, BT algorithm is added to global stereo algorithms, according to the above results, we confirmed it is also helpful for the global ones. The second one, four algorithms based on graph cut are compared