Moving Target Detection, Track And Application On Surveillance System

As the video surveillance system with the constant expansion, the mere use of human to achieve all monitoring points on-line monitoring is already impossible for many surveillance systems. Therefore, intelligent video surveillance system with automatic or semi-automatic monitoring functions is the future development trends of the video surveillance system.This thesis focuses on the intelligent monitoring system moving target detection, tracking, and the practical application on monitoring system.This thesis analyzes the current typical moving target detection algorithm. For the current monitoring system needs, appropriate to improve existing algorithms. By combining the adaptability of the frame difference method and the accurate of background difference method, achieve the rapid detection of moving targets.This thesis analysis and compares the typical moving target tracking algorithms, and then focuses on algorithms based on MeanShift and CamShift Moving Target Tracking. Then this thesis conducted a series of improvements on the shortcomings of the algorithm for the CamShift. First, improved the method of CamShift back projection calculation through a combination of scene information, making a larger difference between target and background for a larger portion of the weight, making back-projection of target and background difference bigger, and making the weights density extreme region overlap with the target region better. At the same time adjusting CamShift search algorithm, by searching the entire region completely. Avoid the unstable calculation time of the iterative operation. Through the optimization search algorithm to obtain even faster computing speed than MeanShift iterative search; while avoiding the problem of fall into local minima leading to the risk of losing track coming along with the MeanShift algorithm. The last part of this thesis show some sequence of video surveillance to track example.Finally, the papers presented in this paper moving target detection, tracking algorithm is applied to the actual control system, camera-based control on the target track, and gives the results, data and analysis of some face tracking example.