一种基于自适应权值的立体匹配方法

提出了一种新的自适应权值的立体匹配方法,在匹配中无需逐像素确定其支持窗口的尺寸。首先根据像素间的相似性和邻近性对匹配窗口内每一像素的支持权值进行调整,使与待匹配点位于同一区域的像素权值增大,然后在匹配的代价函数中引入视差平滑性约束项,从而获得最终视差。在Middlebury提供的标准图像上进行了测试。实验结果表明,该方法可以获得良好的视差图。

An algorithm of stereo matching based on adaptive weight

A new stereo matching algorithm based on adaptive weight is proposed. There is no definition of support-window size for each pixel needed, and the support-weight of each pixel in a given window is adjusted according to the similarity and proximity between them. A pixel from the same homogeneous regions as the pixel under consideration has larger weight, the disparity smoothness constraint term is introduced in matching cost function, and the disparity map is acquired. The proposed method is tested with standard image of Middlebury, and the experimental results show that the method can get good disparity map.