SiamCAR-Kal: anti-occlusion tracking algorithm for infrared ground targets based on SiamCAR and Kalman filter

During infrared ground target tracking, to ensure accurate tracking and minimize target loss caused by background occlusion, an anti-occlusion tracking algorithm based on SiamCAR and Kalman filtering, namely SiamCAR-Kal algorithm, is proposed. First, the algorithm uses the response of the classification branch to determine the occlusion state, and then, to address the short-term total occlusion problem, it uses the Kalman filter algorithm to determine the target position based on historical information. To address the problem of low accuracy of long-term total occlusion of the Kalman filter, an extended search strategy is proposed to achieve target recapture. The experimental results show that the proposed algorithm can not only improve the tracking precision, but also effectively solve the problem of target loss caused by occlusion. It is tested on a sequence of infrared ground targets, and compared with the existing typical SiamCAR, the success rate and precision of the proposed algorithm show an improvement of 3.2% and 4.5% under one pass evaluation indexes.