Research On Region Stereo Matching Algorithm Based On Indoor Scene Image

The local stereo matching is a key problem in the research of stereo matching in computer vision.Due to the influence of illumination changes and object occlusion,stereo matching based on indoor scenes has poor matching effect on the edge regions and disparity discontinuous regions.In this paper,The primary research is as follow:1.This paper proposed a illumination robust stereo matching algorithm based on steady state matching probability.First of all,we improve the robustness of the algorithm from two aspects of homomorphic filtering preprocessing and census nonparametric change.Then we get the result disparity map by calculating the steady-state matching probability and the disparity value of filling the disparity points.The numerical simulation results show that the algorithm effectively reduces the effect of illumination change on the stereo matching effect and improves the matching precision of the algorithm2.This paper presents a stereo matching algorithm based on improved census transform.Through the weighted average of the gray value of each pixel in the input image,the dependence on the value of the central pixel in the window is reduced.At the same time,using the Edge-aware Disparity Propagation algorithm to fill the disparity of pixels that fail the left-right consistency check,the accuracy of disparity calculation results is further improved.3.This paper presents a stereo matching algorithm based on three dimensional label iteration.we try to convert the pixel-level matching problem to patch matching.The input image is superpixelsegmented by SLIC and the same superpixel is indicated with the same label.The best label is obtained by iterative optimization of the label value,thus calculating the disparity of each pixel.Experimental results show that this method effectively reduces the time complexity of the algorithm.In this paper,we use Middlebury's indoor scene data set to evaluate the algorithm.Experimental results show that the proposed algorithm can effectively improve the algorithm's illumination robustness and reduce the algorithm's time complexity.