SAD Stereo Matching Algorithm Based On Binocular Viewpoint

The three-dimensional reconstruction through the stereo matching method has the advantages of simple equipment,low cost,and convenient maintenance,etc.However,due to the complex and changeable reality,there are still many problems to be solved.This paper focuses on the following two aspects of SAD stereo matching algorithm.1.Aiming at the problem that matching accuracy and matching speed are incompatible,the SAD algorithm is improved,and the edge features are integrated in the calculation process.Starting from the working principle of the SAD stereo matching algorithm,the reason for the low matching accuracy is that the variance of the gray value between the pixels in the support window is small,and highlighting the small edges of the portrayed image can improve the gray level of each pixel in the support window.The variance between the values.After analyzing and summarizing the characteristics of each edge operator,it is concluded that the Canny operator has the accurate feature of positioning small edges.Therefore,the Canny operator is used to extract edge features,and the Edge-Gray algorithm is designed and completed.2.Aiming at the problem of inaccurate matching of solid color and weak texture regions,an image segmentation algorithm is proposed for SAD algorithm.According to the principle of digital image coloring,the image is divided into template calculations.The template calculation process is as follows: first,the single-channel color values are extracted according to R,G,and B;then the extracted color channel values are binarized by the maximum inter-class variance method that can automatically calculate the threshold;and then the obtained Binary results are fused to obtain the segmentation template map.The obtained grayscale image,edge feature image and segmentation template image are used in the parallax calculation and parallax optimization process to calculate the parallax image.Compared with the SAD algorithm,the Edge-Gray algorithm has an average mismatch rate decreased by 5.97%,and the time cost has increased by an average of 4.72%;the stereo matching algorithm based on image segmentation has an average mismatch rate decreased by 14.23 percentage points compared with the SAD algorithm,and the fused image After the segmentation algorithm,the average time cost increases by 18.5%.Experimental results show that both algorithms have improved matching accuracy.