The Research And Application Of The Method Of Event Detection And Target Tracting Base On Platform Of Monitoring

Along with the rapid development of intelligent technology, network intelligent video monitoring system also has been widely used. Video surveillance system also experience process of the developement from simulate to digital, networked, intellectualized. Moving target detection and tracking technology is an important part of computer vision, is the basis for the realization of intelligent monitoring and control system, it has become a research hotspot in recent years because of its wide prospect of application. At the same time, with the development of embedded technology mature and all kinds of high performance multimedia processor appeared, it also set off a wave of embedded intelligent monitoring system of research and development. Moving target detection and tracking system based on image recognition, image sequences through the camera to automatically identify and locate scheduled moving targets without the need for human intervention. It can do three dimensional motion and control the camera to keep due to moving objects near the camera view center position. Movement target tracking is a very active branch in the field of computer vision and image processing.It has been widely used in the field of military and national economy and so many ways, such as missile guidance, road monitoring, active visual monitoring and so on.In this paper, we analyze and study the moving target detection and tracking system based on the principle of image recognition and expounds on the recognition and tracking of moving target and the camera controling method.In the part of moving target detection we introduces three kinds of existing motion target recognite technology and an image recognition method based on SIFT features firstly. After analyzes the advantages and disadvantages of SIFT image recognition method, we make an improve to the image recognition method base on SIFT feature with the help of kalman predictor and finally point out the method to realize the moving target image recognition module.In the part of target tracking we have learn and analysis the traditional tracking algorithm called MeanShift and pointed out the MeanShift algorithm shortcoming:the tracking of the target feature extraction by the lack of spatial information and the bandwidth is not adaptive.For this defect,in this paper,we made some improvements which describe the target characteristic with the help of weighted histogram to make up for the target space information the target lacked.Because pap coefficient will vary according to the bandwidth, it can help system realize adaptive tracking. At the same time, we have studied the arget occlusion problem in this paper. We combine track prediction algorithm and kalman predictor to make up for the defect of single use kalman predictor, and devised a switching algorithm with target gray scale, texture and motion estimation, it enables the system stronger robustness.Finally in camera control part,we expounds the technological principle of the traditional PID control and fuzzy control and their respective defect.we combining with the advantages of both fuzzy PID control method is designed and developed to control camera and lead it to the3d activities to track moving targets and keep the target stay in the center of the video view.