Translated Title
Stereo Matching Based on Image Segmentation and Reliability Propagation
Translated Abstract
Objective: This paper presents a novel stereo matching algorithm based on image segmentation and inner-regions reliability propagation.Method: Firstly, an improved local method is employed to obtain the initial disparity maps;secondly, our algorithm select reliable pixels and calculate the rate of reliable pixels in a region, the regions with high reliable pixel rate are tagged with reliable regions, otherwise, unreliable regions.A plane fitting algorithm is applied to obtain the parameters of disparity plane corresponding to each reliable region;finally, the disparity parameters of unreliable regions are circulatory optimized by a local interregional reliability propagation and minimum matching energy scheme based on disparity plane parameters belonging to neighboring reliable regions.Result: Experimental result demonstrate the performance of our method——it ranks 6th in the Middlebury benchmark.
Translated Keyword
stereo matching
reliability propagation
image segmentation
plane fitting
Corresponding authors email