Detection And Tracking Of Moving Objects Based On Real-time Vision

As the core content of computer vision technology,tracking and detection of moving targets is becoming more and more important,and has been widely used in military guidance,medical diagnosis,human-computer interaction and traffic security.Aiming at the requirement of hockey robot project,this paper mainly makes a deep analysis and Research on the algorithm of moving target tracking and detection.Aiming at the deficiency of traditional algorithm,an improved algorithm is proposed.The experiment shows that the improved algorithm can meet the requirements of hockey target tracking accuracy and real-time.The main contents of this paper are as follows:1.Introduce the commonly used moving object detection algorithms: inter-frame difference method,background difference and ViBe algorithm.Through the comparative validation experiment of moving target detection algorithms,the detection effect of each detection algorithm is analyzed.The results show that the detection effect of inter-frame difference method is better,and it can be used to detect hockey targets.2.The moving target tracking algorithms: Mean-Shift target tracking algorithm,Camshift target tracking algorithm and particle filter target tracking algorithm are studied.The tracking effects of each tracking algorithm on hockey target are verified by experiments.Aiming at the shortcomings of traditional tracking algorithm,an improved Camshift target tracking algorithm is proposed.3.Taking the basketball target tracking experiment as an example,the superiority of the improved Camshift target tracking algorithm over the traditional target tracking algorithm is verified.Then,the experiment verifies whether the improved Camshift target tracking algorithm meets the accuracy and real-time requirements of hockey target tracking.The results show that the improved Camshift target tracking algorithm can track the hockey target well and meet the requirements of hockey target tracking accuracy.4.In order to meet the real-time requirement of high-speed hockey,a vector tracking method combining inter-frame difference and collision algorithm is proposed.This method takes the moving target hockey as the object,through establishing the collision model,using LS-DYNA explicit dynamic analysis software to simulate and analyze the process of ice hockey collision with the table in motion,collects the data of hockey collision with the table in different situations,and uses MATLAB to fit the collected data to get the collision algorithm.The experimental results show that the improved algorithm can quickly track the target position of hockey and meet the accuracy and real-time requirements of high-speed hockey tracking.