Research On Algorithms Of Detection & Tracking For Moving Object In Video Serial Images

With the rapid development of computer technology, video has become the mainstream carrier of the world's information exchange. Moving object detection and tracking in video serial images has become a hot point. Sport video is an important one. During the volleyball competition, defence judgment has become an important part, which affects the game process in a certain extent. So this paper has studied the key technologies of the moving object detection and tracking in the volleyball match to improve the accuracy and the real-time performance, which has applied value and practical significance.The main contents of the paper are pretreatment, moving object detection and tracking. Each part has been described in detail and the impoved algorithms have been gived, meanwhile, experimental results have been gived and analyzed.In the dissertation, the main study contents are listed as follows:For moving object pretreatment, the thesis analyzes noise suppression and image enhancement technology, and discusses the mean filtering, median filtering and histogram equalization. Finally it gives simulation results of various methods.For moving object detection, the traditional background estimation method based on block is impoved, which adds the average luminance and time similarity measurement. It can detect moving object fast and accuratly.For moving object tracking, a fast tracking algorithm based on Kalman filter and multi-resolution is gived. It uses the Kalman filter to predict the center of the moving object and then uses multi-resolution to find the best matching position. And for the radiation problem, it gives the self-adapting updating memory filter algorithm. The new algorithm can track moving object real-timely and accurate.