Abstract:

A lot of infrared target tracking systems currently used have some problems. For example, when the target in a complex background, or the target is very small, or the target is obscured by other objects, the targer would be lost by the tracking system in the tracking process. To solve this problem, a new infrared target tracking algorithm based on matrix S_1/2 norm is proposed in the framework of the particle filter. First, the target particles in the current frame are sampled around the target state of the previous frame. Then, the target particles are filtered, and the filtered particles as a whole are input to the model of the minimization problem which is based on the joint representation of matrix S_1/2 norm and l_1,1 norm, and solves the optimal solution of the model. Finally, the optimal target particle (i.e., the target in the current frame) is determined according to the difference of the coefficients between the target dictionary and the background dictionary. Experimental results show that the proposed algorithm can effectively track the infrared targets with complex background, or weak targets, or the target obscured by other objects, and has stronger robustness and better real-time perfor-mance than the classical similar target tracking algorithms.