Research On Object Tracking Algorithm In Intelligent Surveillance

Object tracking is a fundamental topic in computer vision field and is widely applied in intelligent robotics, Medical diagnosis, especially in surveillance. As the intelligent high speed dome camera can provide a broader view and more details, it has been gradually widely used in surveillance. However, most of the current products based on high speed dome monitor what is happening in the field just by stuff, which is labor-intensive. Therefore, the automatic object tracking technology in surveillance becomes a hot issue.The current tracking algorithms are not applicable when used in surveillance based on high-speed dome because none of them can meet the speed and occlusion requirements by themselves. This paper researched these two aspects, details are as follows:1. Firstly this paper built the hardware platform and proposed the implementation plan of the monitoring system.2. This paper presents an object tracking algorithm which combined particle filter and Mean Shift. This approach has the advantages of particle filter and Mean Shift, which can meet the speed and occlusion requirements of the surveillance system.3. This paper proposed a variable speed control method, which can reasonably regulate the speed of high speed dome according to the target position. When the position of the target near the center of the field reduce the speed of the high speed dome, on the contrary speed the high speed dome. In this way the target can stay in the centre region of the field stable.4. This paper conducted a lot of validation and comparative experiments, which provided a factual basis for the design of tracking algorithm and verified the accuracy and real-time of the designed method.Experiments show that the object tracking method designed for surveillance system can track the targets when they are fast-moving or in occlusion in the large field.