Research On Moving Target Detection And Statistical Algorithm On Water Surface Based On Vision

In the research of moving target detection algorithm,researchers want to get all the information of the detected target,but due to the influence of various external and inherent factors of the target itself,there is always some gap between the detection results and the ideal results,which often affect the accurate processing and analysis of images and videos.In this paper,the detection and statistical methods of moving targets on the surface of the water are studied,with emphasis on the modeling of the detection algorithm of moving targets on the surface of the water and the statistical method of the number of moving targets.Firstly,this paper introduces the research background,significance and present situation of moving target detection and statistics,briefly introduces and compares some mainstream methods,and discusses the problems and challenges faced by current target detection algorithms.In this paper,the moving target detection and statistical algorithm are studied.The main research contents and specific innovations are as follows:A moving target detection algorithm based on sparse model is proposed.A dynamic surface background dictionary is trained by sparse representation theory,and then the background image block can be easily reconstructed by the dictionary.If an image is known and its background image can be estimated,then the moving target can be obtained more accurately.A moving target detection algorithm based on interval background model is proposed.Firstly,the range of background value on each coordinate point in video sequence is learned from a video without moving target,and it is formed into interval background model.When the moving target is detected,the value of the current pixel is compared with the value of the background model.If the value of the pixel is within the range of the background module,the pixel is the background,otherwise it is the moving target.After all pixel detection is completed,it is combined into an initial moving target image.Finally,the initial moving target image is de-interfered,and the final moving target is obtained Image.A moving target detection algorithm based on image block distance model is proposed.By judging the Euclidean distance between two image blocks,the background image block is obtained and then the background model is obtained,so that the moving target can be obtained more conveniently.A new method for calculating the number of moving targets is proposed.the algorithm is based on the target of the video frame to be tested,the centroid of the algorithm is taken as the study point of mass,the Euclidean distance judgment is carried out on the movement of the center of the centroid coordinates to determine whether to be the same target,and a rectangular frame mark of the target to be tested is predicted by using the Kalman filter to predict the center position thereof,And finally,the statistics of the number of the moving targets are finished.