Research On Moving Target Recognition And Tracking Method

Research on moving target recognition and tracking methods has always been a hot topic in computer vision research.It is widely used in video surveillance,medical surgery diagnosis and treatment,and intelligent transportation.Therefore,the research of this subject has a strong theoretical significance.Practical significance.This paper mainly studies the moving target recognition and tracking methods in static background.The problem of target recognition in complex background is not obvious and the tracking failure occurs when occlusion occurs.It is improved on the basis of analyzing traditional recognition and tracking algorithms.Finally,the research results obtained in this paper are as follows:Research on the method of moving target recognition.Firstly,the traditional target recognition algorithms are analyzed and compared,and according to their respective advantages and disadvantages,in order to solve the problem that the recognition effect is not obvious,a new improved algorithm is proposed based on the frame difference and background difference method.The algorithm combines the three-frame difference method of edge detection with the background difference method based on Surendra algorithm,and tested in five different environments,such as single-environment,large-scale and small-scale target scenes,fast target scenes,multiple moving target environments and lighting environments.The results show that the proposed algorithm can extract the moving targets from the image sequence by comparing the results in a single background and various complex backgrounds completely and accurately.It is suitable for different scenarios and meets the requirements of real-time performance.The environment is robust,which makes up for the shortcomings of the frame difference method and the background difference method,and is feasible and effective.Research on moving target tracking methods.The classification of commonly used tracking technologies is described,and then the basic principles of the Camshift algorithm are elaborated and experimental demonstrations are carried out.When the target is occluded,it will lead to tracking failure.To solve this problem,the predictor-to-tracking process is introduced.The predictor is used to estimate the expected position of the target in the next frame image to solve the occlusion problem.In this thesis,a method based on linear prediction is proposed,which is combined with the Camshift algorithm and compared with the common Kalman+Camshift algorithm.The method compares experiments in four aspects in three environments,and the four aspects are tracking effect,tracking error,number of iterations and time spent.The results show that the method can not only solve the tracking failure caused by occlusion,but also meet the requirements of real-time and tracking in different environments.The prediction is accurate and the method is feasible and effective.The improved algorithm proposed above can solve the problem that the target recognition effect is not obvious and the tracking failure caused by the occlusion,and the new algorithm has a significant improvement in performance compared to the pre-improvement.