Research On Methods Of Infrared Target Tracking Based On Particle Filter

Infrared imaging guidance is a kind of intelligent precise-guidance technology, which uses the difference of infrared radiation between target and background to obtain infrared image that is processed real-time by the high speed processor, making infrared image guidance an efficiency guidance method. The tracking of infrared target is one of the pivotal tache in infrared image guidance system. With modern high-tech battlefield becoming increasingly fierce confrontation, in complex natural background and interference conditions, accurate tracking of infrared target is one of the major challenges of infrared guided weapons.To solve the problems above, this paper mainly studies the particle filter tracking algorithm and based on its improved IR target tracking method. Firstly, introduces pretreatment before target tracking, mainly including the image enhancement, noise reduction and target segmentation, analyze to experiment on real infrared data for image preprocessing and segmentation. Secondly, introducing the particle filter and mean shift algorithm, to deal with the problem that tracking precision of particle filter is influence by particle count, an improved infrared target tracking algorithm based on particle filter combine with mean shift is proposed. It uses mean shift to optimal estimated value of particle filter, compared with the traditional combining ways, the proposed method reduce the computation time, without decreasing the tracking accuracy. Meanwhile, the introduction of anti-occlude strategy, to ensure the stability of target tracking when the target is occluded. Finally, in view of the particle degradation problems, use quasi-Monte Carlo Gaussian particle filter to track the target. It omits the resampling step, not only avoid particle degradation, but also reduce the amount of computation. To further enhance the describe ability of target, import Sigma Set and propose an IR target tracking algorithm based on Sigma Set QMC-GPF. Experimental results show that the algorithm is more precise than the traditional particle filter algorithm, and has strong anti-interference ability.