A disparity map estimation algorithm for the three-dimensional swept volume display: Optical adaptive direction Normalized Cross-Correlation window

摘要

This paper presents a disparity map estimation algorithm for the three-dimensional (3D) swept volume display based on multi-view stereo. In this paper, we first segment the interested object from row-aligned images by optical flow. Then the Normalized Cross-Correlation function and adaptive direction window are integrated for stereo correspondence, which means to use the direction window to fix boundary problems, statistical parameters to sparse texture issues, precise initial matching principle to repetitive texture troubles, image pyramids to the improvement of computation speed difficulties. The results show the robustness and efficiency of this algorithm.