Multiple Moving Targets Detection And Tracking Algorithm

The detection and tracking of moving objects is an important technology in computer vision area. It can reduce staff, increase efficiency and improve system performance.Now it has been widely used in many fields. The aim of this paper is to improve the self adaptability and real-time capability of the detection and tracking algorithm based on the available research.First an improved target detecting method based on self-adapt background model is proposed in this paper. This method resolves the hole problem of the inter difference algorithm and the misdetection of the background difference algorithm. And it can work out the detection and distinguish of multi-target.Two algorithms are used in this paper to track the moving target. It enhances the real-time capability and anti-jamming performance of the classical tracking algorithms.The moving target tracking algorithm based on matching feature corner uses the matching of the feature corner in order to realize the aim of tracking moving target. Edge detection method is used in this paper to improve the disadvantage about processing time of Susan corner detection method. Then Maximum Close Distance (MCD) algorithm is used to track the feature corner.In order to gain better tracking effect and enhance the real-time capability, more research about Mean-Shift tracking algorithm is done in this paper. Kalman prediction algorithm is used to figure out the problems that target blocked and similar objects interference. A self-adapt search window method is proposed to solve the shape changes of the target during the tracking process. At last, some research about how to keep tracking when two targets met is done.On the basis of above research, a software system with good accuracy, real-time capability and anti-jamming performance is implemented in this paper.