Video Target Detection And Tracking Based On Euclidean Distance Prediction

With the continuous development of artificial intelligence technology,the iterative updating of Internet technology,the maturity of computer vision technology and the diversification of image processing methods,the real-time detection and tracking of moving objects in video based on image processing becomes more and more The more important it is,the more it becomes a research focus in computer vision,pattern recognition and machine vision.Motion information in video is an important means of video analysis.It plays a decisive role in the understanding of video.It contains a wealth of target state information.Therefore,the detection and tracking of moving objects in video become very meaningful.However,due to frequent interference such as background interference,occlusion,color change and other interference information in video,it is difficult for the existing technology to directly detect and track the moving target in the video.This paper studies how to detect and track the moving target when the moving target is disturbed before and after the moving target is blocked in the video.First of all,this paper studies some pretreatment techniques of image processing,including color image grayscale,image filtering,morphological expansion and so on,analyzes the common detection technology of moving objects in video,Three-frame difference extraction method,and the detection of moving objects under various methods.Secondly,in the process of video moving target tracking,a scheme combining tracking and prediction is proposed to construct the tracking model.Cam Shift algorithm is used to track the moving target of video frame effectively and the size of the tracking window is adjusted according to the movement of the target automatically so as to ensure that the moving target is not lost when the size of the target is partially deformed track.Finally,when the moving target occludes,the Cam Shift target tracking method based on the Euclidean distance prediction is proposed by using the adaptive predictive tracking scheme.When the target occludes,Kalman predictor or linear predictor is adaptively selected to predict the possible position of the moving target.Then the Cam Shift algorithm is used to search the target centroid to improve the accuracy of the search.The test results show that the adaptive predictive tracking method proposed in the paper can effectively solve the problem of high target follow-up rate in the original algorithm and can meet the requirements of tracking accuracy and real-time.It has certain theoretical significance and practical value.