Human Body Movement Analysis Research Based On Binocular Stereo Vision

In the rapidly developing information age,computer vision has become a very promising and challenging research field.Binocular stereo vision is an important branch of computer vision.It allows computers to understand the world by mimicking the principle of human eyes.In the binocular vision system,the key part is the stereo matching technology.Due to the influence of various noises in the matching process,the stereo matching technology has not been able to achieve the effect of real-time performance and accuracy.Therefore,it has always been one of the hot topics studied by scholars at home and abroad.In this paper,the binocular stereo matching algorithm is deeply studied.By locating and matching the key points of the target area in the image,the disparity values of the key points are obtained,so that their position relationship in the real scene can be obtained and the analysis.of human action can be realized.In order to achieve stereoscopic matching more accurately and quickly,this paper proposes an average background subtraction method based on adaptive threshold.The background is modeled using a single Gaussian distribution,and then the threshold of each pixel is calculated using the model parameters.Finally,the target region is obtained by differential operation.This method locks the matching range and reduces the complexity of the algorithm.Then,in order to analyze the movement of the human body more directly and effectively,this paper proposes a stereo matching method based on skeleton key points.In order to get the key point,the body is first extracted from the skeleton and contours,and then endpoint detection is performed on the skeleton.Finally,the point with the largest curvature on the contour curve in the neighborhood of each endpoint is used as the key point,and then the extreme line constraint and the normalization are used.The matching rules of the correlations are completed to complete the matching of the key points,and the positional relationship of the key points on the human body in the three-dimensional space can be obtained through the disparity value of the key points.In the end,what this paper implements is a human motion analysis based on SIFT feature point matching.The SIFT algorithm is used to locate,describe and match the feature points in the target area.Finally,the depth relationship is obtained by using the disparity values.The above two methods are all based on the target feature point matching method.Due to the uniqueness and characterization of each feature point,the accuracy and real-time performance of the matching result is very high,which provides a more accurate and effective analysis basis for human motion analysis.It has certain practical value.