Academy & Industry Research Collaboration Center (AIRCC)

Volume 9, Number 18, December 2019

A Progressive Stereo Matching Algorithm based on Texture Extension


Yingjiang Li1, Yuzhong Zhongb2 and Maoning Wang2, 1Chongqing university of technology, China and 2Sichuan University, China


It is difficult to get accurate stereo matching for the occluded and smooth areas of an image. In this paper, a stereo matching algorithm is proposed based initially on texture region and then on smooth region. First, the reasonable texture points in the left and right image pairs are obtained, and the texture point disparity values of the left and right images are calculated respectively using the adaptive-weight cost aggregation method. Then, starting from the texture points, the disparity values of the smooth areas are calculated by expanding the disparity maps of the left and right textures. Finally, the expanded left and right disparity maps are fused to obtain the final one.In the matching process, the algorithm effectively avoids the occluded area and adds constraint methods to improve the matching accuracy and speed. Experiments show that the proposed algorithm is fast and has higher accuracy disparity between discontinuous and smooth regions.


stereo matching, adaptive weights, texture extraction, disparity expansion