Design And Implementation Of Video Warning And Tracking System Based On MeanShift Algorithm

The application of video monitoring has been existed for a long time.With the rapid development of computer technology and network communication technology, video monitoring system has been widely used in many industries.However, most of video monitoring systems still remain in the traditional mode, which is playbacking after accidents. It can be said that the traditional mode of monitoring just plays a monitoring function, does not play the role of control situations.To play controlling, it requires that the system can predict and alarm for dangerous behavior of activities of targets, and prevent and control the occurrence of danger befor the accident, so as to really play a security role, before they happen. This kind of monitoring with prediction and alarm, is known as intelligent monitoring by industry. It is based on target detection and target tracking technology.Nowadays, target detection and tracking in video images is an important research topic.At the same time, it is advocated “building safe city” by the government, so it is urgent to research intelligent video monitoring system based on moving object detection and tracking technology, its application is promising with important commercial value and social significance.In this paper, on the basis of previous research achievements, we analyse the principles and steps of MeanShift tracking algorithm, design and implement a video warning tracking system based on MeanShift algorithm. During the design and implement phase, there are some replenishes and improvements for the deficiencies of the MeanShift: At first, this thesis perfected the semi-automatic tracking to the automatic tracking in MeanShift.Secondly,it had added pretreating of moving targets, which reduce the possibility of track loss in complicatied environment, while enhance the stability of target detection. Besides, this system used binocular application pattern,including a panoramic camera for target detection and a PTZ camera for target tracking, which achieved the acurate positioning and the automatic tracking to make realtime detecting and analysing to target path. If some behavours broke the safety rules, the system will send an warning to security personnel by sending SMS alarm and flashing the warning light.This thesis focuses on optimizing and using of Mean Shift tracking algorithm, emphasizes on further analysing and using of target tracking data, committes to achieving two functions of target tracking and risk warning.This system has largely improved the intelligent level of video monitoring.