Study On Moving Object Detection And Tracking Algorithm In The Video

Intelligent video surveillance technology is a comprehensive technology which is a combination of image processing technology,pattern recognition technology,electronic sensor technology,and computer data mining analysis technology.The rapid development of community economy has provided a broad application prospects for intelligent video surveillance technology when it is applied in intelligent security,intelligent transportation,intelligent tourism,and urban management and planning.Therefore,the research on intelligent video surveillance technology theories and methods will have great practical significance and scientific value.As for intelligent video surveillance technology,moving target detection and tracking are two major aspects of its research,and are important prerequisite to ensure that the video surveillance system to obtain accurate and effective information.At present,the research has aroused the widespread concern of many domestic and foreign experts and scholars,meanwhile,it has become the current research hotspot.Based on the current development trend of intelligent video,this paper has done some work and research on the theory and algorithm of target detection and target tracking.The paper mainly works and innovations including:1.About the target detection aspects,specific process of the SIFT and SURF algorithm which included feature points extraction,descriptive generation and feature matching are analyzed.The simulation results of MATLAB simulation are used to simulate the effects of illumination,distortion and noise to prove the matching effect of two feature matching algorithms.2.The knowledge of compressed sensing theory is expounded,and the contents consist of sparse signal representation,measurement matrix selection and signal reconstruction algorithm three aspects which are analyzed.3.Improved algorithm based on the compressive tracking algorithm is researched.According to the matching relationship of the SURF feature points in the adjacent frames,the tracking target size change is adjusted and the target template size is adjusted adaptively.The compressive tracking algorithm is still used to track the target and to update the target appearance model.Meanwhile,the mechanism of misinterpretation of the appearance model is added to solve the problem of severe occlusion and distortion.4.A robust tracking algorithm which is optimized for the shortcomings of Spatio-Temporal context(STC)algorithm under occlusion is studied.In the target detection stage,the SIFT algorithm is used to locate the target,reduce the time and search range of the previous search.In the target tracking stage,the idea of STC algorithm is adopted that have completed tracking by the help of the time and space information around the target.For severe occlusion,the STC algorithm is invalid,and the Kalman filter algorithm is utilized to predict the target.