Research Of Target Tracking Based On Infrared Video Image

Video target tracking is a core issue of computer vision area, it has become the focus of national attention. In this paper, target tracking of infrared video image is studied, which is used for vehicle infrared video tracking system. As the complexity of the background in the video and infrared target make tracking more difficult.An effective infrared tracking algorithm is designed and implemented based on the existing tracking algorithm. Considering accuracy and real-time of the target tarcking, optimized template matching algorithm and wavelet pyramid hierarchical search algorithm is presented. Optimal template matching algorithm is optimization of the traditional template matching algorithm, a new similarity measurement method was proposed. The algorithm of tracking confidence is obtained and the adaptive adjustment strategy of template updating thresholds is proposed. Also, the adaptive template updating algorithm which is based on the confidence of tracking is achieved. Considering the changes of target size in the tracking process, the adjust strategy of template size is proposed, in order to reduce the impact of the fixed template size on the tracking accuracy. In addition, kalman filtering is used to predict the position of the target, which resolves the temporary occlusion problem of the moving target, ensuring continuous tracking. Based on the research of the principle of wavelet pyramid, wavelet pyramid hierarchical search algorithm is designed and implemented. Wavelet pyramid improve the efficiency of the algorithm without reducing the accuracy of the tracking.Experiments show that the adaptability of the proposed algorithm to the environment and its stability of tracking are greatly improved, which has a certain real-time and can meet practical demands.