Research On Target Tracking Technology Of Intelligent Video Surveillance System

With the rapid development of computer vision field,the intelligent technology of video surveillance system has also made some progress.Research on target tracking method for intelligent video surveillance system has become a hotspot in the research of computer vision.The core function of the intelligent video surveillance system is the basis of the target recognition and behavior analysis of the intelligent target monitoring system,which has important research value.This paper focuses on the target detection method and target tracking method of the intelligent video surveillance system.Aiming at the defects in the existing algorithms,a multi-target tracking method for intelligent video surveillance system which can realize automatic target recognition is proposed.In this paper,we first study the image processing and machine learning methods involved in target detection and target tracking,including noise filtering,binary image morphological processing,random forest classifier and P-N machine learning.Secondly,according to the application requirement of intelligent video system,the traditional target detection algorithm is researched and analyzed in detail,which is frame difference method,Gaussian background modeling and optical flow method.The advantages and disadvantages of each algorithm are analyzed.In this paper,a Gaussian background modeling method based on LBP operator is proposed to overcome the shortcomings of anti-shadow interference in Gaussian background modeling method.The background model is described by LBP operator.In this paper,Mean Shift algorithm is studied and verified experimentally.For the defects of the Mean Shift target tracking algorithm which can not overcome the obstruction interference,this paper proposes an improved Mean Shift target tracking algorithm based on machine learning,and achieves the filter for occlusion interference In addition,the improved method is applied to realize the multi-target tracking expansion and real-time tracking for multiple targets.At last,the target detection algorithm and target tracking algorithm are used to develop the intelligent video surveillance target tracking simulation system,which is implemented by using OpenCv computer vision library and Qt image library.The algorithm is implemented in this paper.