The Moving Target Detection And Tracking System Based On Image Recognition

The moving target detection and tracking system based on image recognition, without manual intervention, identifies and locates the moving target in sequence of images which received by vidicon camera automatically and controls the vidicon camera to do three-dimensional motion to maintain the moving target in the center of the camera view. Video tracking is one of the most active embranchment in the field of computer vision and image processing, it also used widly in military and the development of national economy. Such as, missile guidance, active vision monitoring and so on.In this paper, the principle of the moving target detection and tracking system is analyzed. The system's structure is designed. The methods of image recognition and control are discussed.At the hand of image detection, in order to meet the accuracy and speed of image detection, after compared the common image identification algorithm theory, a new image recognition algorithm, which based on kalman prediction and SIFT, is proposed. The target's position in the next image is predicted by kalman filter, and then, a adaptive identification window in the region of the position is extracted. Identification the moving target in adaptive window which can not only cut down the time of extracting the SIFT feature but also improve the efficiency of image detection.Because of the system exists serious nonlinear, interference and the target's uncertainty, a fuzzy PID controller is designed and it realized by common PID control tools and fuzzy logic under Lab VIEW. The rise time and adjustment time are shortened and overshoot is reduced. It made the camera tracking the target in real-time possible.Finally, the various parts of the system were tested. Test results show that image identification algorithm combined kalman predictor with SIFT can identified target quickly and accurately; with the help of the fuzzy PID controller, the camera can track the target always.