从双视图到多视图的协同优化立体视觉匹配算法

采用基于视图协同优化算法解决视觉匹配问题,通过对双视图立体匹配的研究到多视图匹配的推广,利用视差平面拟合进行多个视差集的合并.利用奇异值分解法解决线性方程,建立排除异常点的判别规则,解决遮挡及稀疏纹理的问题.实验结果表明:该协同优化算法是有效的,提高了匹配的精度.

An Algorithm of Stereo Matching Based on Cooperative Optimization from Two Views to Multiple Views

The algorithm of stereo matching using cooperative optimization is developed to extract disparity map from image pairs. The procedure of algorithm is generally extended to apply to multiple views from two views. The disparity sets are merged into one by the disparity plane estimation, and the linear equation is solved by singular value decomposition(SVD). And the rules are built to filter out outliers, which can effectively handle the problem of occlusion and textureless. The experimental results show that the effectiveness of our proposal methods and the accurateness of stereo matching.