An Efficient Focused Stereo Matching Method Based on Image Segmentation

Abstract

The invention proposes a high-efficiency focus stereo matching method based on image segmentation. First, use the image segmentation method to divide the reference image into different image blocks, perform Census transformation on the reference image and the image to be matched at the same time to obtain sequence values, then use the disparity of neighboring pixels to dynamically correct the disparity range of the matching pixels, and finally use the matching cost function , using the segmented block as the supporting window, using sequence values ​​for cost calculation, calculating the aggregated cost value in each segmented block by moving the window in the horizontal direction, and the horizontal offset corresponding to the smallest aggregated cost value represents the value of the segmented block Optimal parallax value. The present invention can not only improve the matching accuracy of the depth discontinuity area, but also has high calculation efficiency, and also has good robustness under different lighting environments.

Description

technical field [0001] The invention belongs to the technical field of stereo vision, and in particular relates to a high-efficiency focus stereo matching method based on image segmentation. Background technique [0002] Stereo matching is one of the most important fields in computer vision. Its main purpose is to find corresponding points from stereo image pairs and calculate dense disparity maps. Many algorithms have been proposed in recent decades, which can be divided into global matching algorithm and local matching algorithm based on the difference of cost aggregation process (see literature 1 D.Scharstein and R.Szeliski.A taxonomy and evaluation of dense two-frame stereo correspondence algorithms[J ]. [0003] IJCV, 47 (1 / 2 / 3:7-42, 2002.). The global matching algorithm usually establishes an energy function according to the respective constraints, and then obtains the disparity result through multiple iterations and optimization of the energy function. The disparit...

AI Technical Summary

Problems solved by technology

However, the rectangular fixed window will bring the boundary blur problem, resulting in the mis-matching of the depth discontinuity area

At the same time, for images with illumination changes and noise, the algorithm also has mismatching

[0007] Document 5 Man Li, Jing Han, Yi Zhang, Lianfa Bai: An improved stereo based oneffective cost aggregation.Proc.SPIE9675AOPC 2015: Image Processing and Analysis, 2015 improved on the existing basis, and did not significantly reduce the error of the depth discontinuity area matching rate

Images