一种基于Census变换的可变权值立体匹配算法

针对传统基于Census变换立体匹配算法精度不高的问题,提出了一种基于改进Census变换的可变权值立体匹配算法. 在分析传统Census变换缺陷的基础上,提出利用最小均匀度子邻域均值代替中心像素灰度值进行Census变换,可有效增强算法的抗干扰能力. 通过加权区域海明距离均值和标准差作为相似性测度进行立体匹配,减少误匹配,提高匹配精度并通过左右一致性检测和遮挡填充,生成最终视差图. 实验结果表明,该算法鲁棒性得到增强,在深度不连续区域也可以得到准确的视差. 

A Census Transform Based Stereo Matching Algorithm Using Variable Support-Weight

Aiming at solving the problem of low accuracy of matching algorithm by typical Census transform, a stereo matching algorithm using variable support-weight based on modified census transform is proposed. On the basis of analyzing the defects of traditional Census transform, a modified Census transform algorithm was developed using average value of minimum evenness sub-area as a reference instead of the center pixel intensity, which enhanced the robustness of the algorithm. The matching accuracy was improved by weighting the average value and the standard deviation of Hamming distances in a region. By means of left-right checking and occlusion filling, the final disparity map could be acquired at last. Experiment results indicate that the robustness of proposed algorithm is enhanced. Accurate disparity could be obtained even in the depth-discontinuities regions.