关键词: 立体匹配, 跨尺度代价聚合, 快速引导滤波, 区域树代价聚合

Abstract: To improve the accuracy and speed of indoor automatic logistics equipment and industrial grabbing equipment,an improved stereo matching algorithm based on cross-scale cost aggregation was proposed.To solve the problem of high error rate in non-texture area and low-texture areas when used traditional stereo matching method based on Cross-Scale Cost Aggregation(CSCA),different cost aggregation algorithm was used for the cost aggregation of different downsampling layers.Different aggregation methods between different downsampling layers could be fused and suppressed.At the same time,in order to solve the problem that the calculation takes a long time in the cross-scale cost aggregation,the Fast Guide Filter(FGF)was used.In the even downsampling layer,FGF was used,and the odd downsampling layer,Segment-Tree Cost Aggregation(STCA)was used,so that the new algorithm could obtain a more accurate disparity map and greatly reduce the computational time.

Key words: stereo matching, cross-scale cost aggregation, fast guided filter, regional tree cost aggregation