基于动态自适应加权Census变换的立体匹配算法

Stereomatching algorithm based on dynamic adaptive weight Census transform

为了解决匹配过程中在弱纹理区域的误匹配问题,提出了在不同搜索窗口下自适应加权的Census变换的立体匹配方法。采用Census变换得出搜索窗口内对中心像素影响不同的特异点,分别将邻域像素与中心像素灰度差值进行处理,得到搜索窗口内邻域特异点对中心像素的影响权重,使用权重值加权求和的方式代替计算汉明距离,得出初始匹配代价。为了增强视差图,采用4路径聚合方式与赢家通吃算法(Winner-take-all,WTA),利用子像素拟合、一致性检查、颜色信息加权中值滤波和视差填充等步骤优化视差。为了验证算法效果,将提出的自适应加权Census算法与ADSG、SGM、LPSM和SGBM2进行比较,在Middlebury 2014数据集上进行了测试实验。验证了所提算法的有效性,提高了匹配的准确率,视差图更加平滑、边缘清晰。

In order to solve the problem of mis-matching in the weak texture region during matching,the stereo matching method is proposed,with adaptive weighted Census transform under different search windows.The Census transform is used to derive the idiosyncratic points in the search window with different influence on the center pixel,and the grayscale difference between the neighboring pixels and the center pixel is processed separately to obtain the influence weight of the neighboring idiosyncratic points on the center pixel in the search window,and the initial matching cost is obtained by using the weighted sum of the weighted values instead of calculating the Hamming distance.In order to enhance the parallax map,the 4-path aggregation approach with the Winner-take-all algorithm WTA is used,and the parallax is further optimized using sub-pixel fitting,consistency checking,color information weighted median filtering,and parallax filling steps.To verify the algorithm effectiveness,the proposed adaptive weighted Census algorithm is compared with ADSG,SGM,LPSM,and SGBM2,and test experiments are conducted on the Middlebury 2014 dataset.The experimental results verify the effectiveness of the proposed algorithm,improve the matching accuracy,and the resulting parallax maps are smoother with clearer edges.