The Research Of The Ground Background Infrared Target Tracking Algorithm

Moving object tracking is a challenging problem in the field of computer vision and pattern recognition, which has an important application prospect in the field of civil and defense industry. The infrared imaging target tracking has greater range of application in the military field. Many people did a lot of research on infrared target tracking. However, because of some of the characteristics of infrared tar get, the tracking still has many unresolved issues. This paper does some research on the infrared target tracking algorithm of ground background.To deal with the weak of infrared target and inconvenient to manually select the trace window, a target trace window selection algorithm base on traceability is proposed. It can more easily to initialize the tracking and reduce errors caused by manually selecting the target.To deal with the low contrast and the low signal to noise ratio when tracking targets in infrared image sequences, an infrared target tracking algorithm based on Local Steering Kernel feature is proposed. Founded on the tracking framework of Bayesian, a robust and effective target model is built by using LSK feature to describe the infrared tar get. An online real-time target sample library update method is adopted to enhance the robustness of the algorithm. Experimental results show that the algorithm is effective and robust.To deal with the problem of the infrared target scale variable, a variable scale infrared object tracking algorithm base on on-line Ada Boost is proposed. The algorithm adds a scale calculation function which use sample matching to calculate the target scale changes. The algorithm has the following characteristics: 1) The target tracking algorithm can adapt to the change of target's appearance and partial acclusion base on on-line learning. 2) Using haar features to establish target appearance model is more applicable to the infrared target which lack of texture characteristics. So as to make the target appearance model more robust. 3) The tracking algorithm adopt to the change of target's scale when added to the scale calculation function which base on sample matching. So the infrared target tracking has a better result when the scale changed dramatically. Experimental results show that the algorithm achieves high accuracy in the infrared target tracking underdifferent background and adapts to the scale changes. The algorithm has a high real-time and robustness. It has obvious advantages compared with other algorithm.