针对局部立体匹配算法中边缘区域易造成误匹配的问题,提出一种基于边缘加权的跨尺度局部立体匹配算法。在代价计算阶段,根据边缘点的数量和结构信息提出一种边缘相似度测量方法,并对满足约束条件的点进行两种策略的“奖励”加权,以提高目标图和参考图中对应点的辨识度;在代价聚合阶段,引入多尺度模型,并采用引导滤波进行聚合;通过视差计算、视差精化得到最终视差图。在Middlebury立体视觉测试平台下对4组标准立体图像对和27组扩展立体图像对进行测试。实验结果显示,在未加入任何精化步骤的情况下,非遮挡区域的平均误匹配率为7.88%,说明本文算法有效改善了边缘区域的匹配精度。

To solve the problem of mismatch of the edge region in the local stereo matching algorithm, a cross-scale local stereo matching algorithm based on edge weighting is proposed. In the cost computation stage, an edge similarity measurement method is proposed according to the number and structural information of edge points, and the points satisfying the constraint conditions are weighted by two strategies. In this way, the recognition of corresponding points in the target and reference maps are improved. Cross-scale model is introduced in the cost aggregation stage, and guided filtering is used for aggregation. Finally, the disparity map is obtained by disparity computation and refinement. Four sets of standard stereo image pairs and 27 sets of extended stereo image pairs are tested on the Middlebury benchmark. The average mismatch rate of non-occlusion regions is 7.88% without any refinement steps. Experimental results show that the proposed algorithm effectively improves the matching accuracy of the edge region.