The Design And Implementation Of The Moving Target Detection And Tracking System

With the development of computer technology and computer vision technology,intelligent video surveillance system is gradually penetrated into all aspects ofpeople’s life and production. Intelligent video surveillance system is a monitor systemthat can automatically and intelligently analyze and processing video sequences, thenit’s able to complete moving target’s detection and tracking.The moving targetdetection and tracking technology is a critical and core part of intelligent videosurveillance system. Research institutions and researchers have invested a lot ofmanpower, material and financial resources to this problem for decades, but there isnot a moving target detection and tracking system can apply to all kinds ofenvironments. With the deepening research and the tireless efforts of researchers,there will be a highly robust, high-performance moving target detection and trackingsystem that can widely applied to various industries. This paper, we study and realizeda moving object detection and tracking system and try to solve the problem of movingtarget detection accuracy and the track failed problem when a moving target isoccluded in the scene and there are many interferences are similar to targets in color.the main work include:Analysis and comparison of various moving target detection methods. This paperstudies and analyzes the current three kinds of commonly used moving targetdetection algorithms: optical flow method, frame difference method andbackground subtraction method, and points out the advantages and disadvantagesof several algorithms. The pixel based adaptive segmentation algorithm (PBAS)uses the adaptive segmentation threshold to conduct pixel segmentation andrandom select update mechanism to update the background model. Its detectionresults have high accuracy, wonderful robustness to noise and the good ability toadapt to the environment. The two value image by motion detection usuallycontains multiple moving objects. In order to achieve the tracking of movingtarget, the position of the moving target and other information is needed. With theconnected region detection, the connectivity of moving target is analyzed and themoving target is calibrated.Analysis and comparison of moving target tracking algorithm.This paperintroduces several commonly algorithm of moving target tracking: Kalman filtertracking algorithm, CamShift tracking algorithm and ABCShift algorithm, andpoint out the advantages and disadvantages of these types of algorithms. With depth study of the combination of Kalman filter and ABCShift algorithm, we findit takes advantage of both the spatial information and color information of thetarget. This algorithm solves the track failed problem of interference by the classmoving target and the target color is severely blocked when a moving target isoccluded in the scene and there are many interferences are similar to targets incolor. The main idea of the algorithm is: First of all, the Kalman predicted targetlocation of current frame;Then, the ABCShift algorithm matching the movingtarget; Fanally,update the state of Kalman filter.Achieve rapid and accuratedetection and tracking system of moving targets. This paper describes a movingtarget detection and tracking system based on MFC in the platform of VS2008and OpenCV, the system is divided into four modules: video capture module,moving target detection module, connected region detection module and movingtarget tracking module. And then analyze the overall system design ideas, designideas of each module design ideas, system’s MFC and OpenCV environment, thedetailed detection and tracking module. By using different tests video to test andanalyze the performance of system. Through the test results, we can conclude thatthe system solves the track failed problem when a moving target is occluded inthe scene and there are many interferences are similar to targets in color.