Local Stereo Matching Algorithm Based on CIELAB Color Space and Census Non-parametric Transformation
Juan Du, Huan Zhao, Sheng Xu · 2019
In order to meet the demand of stereo matching in low-texture areas and improve the matching accuracy, a stereo matching algorithm based on CIElab color space and Census non-parametric transformation is proposed in this paper. The algorithm is able to improve the traditional adaptive algorithm by selecting the absolute gray level difference in CIELab color space instead of in RGB space. Census non-parametric transformation introduced the similarity measurement method by replacing the gray value of the central pixel point with the pixel relationship between the central pixel and its neighborhood. Finally, we performed disparity post-processing for left-right consistency detection, sub-pixel enhancement and median filter. Experimental results show that the proposed algorithm can effectively improve the matching accuracy, enhance the robustness of stereo matching, and improve the blocking effect of traditional algorithms in low-texture areas.