HUMAN TARGET TRACKING METHOD BASED ON KERNELIZED CORRELATION FILTER AND INTERACTIVE MULTIPLE MODEL

Through-the-wall radar (TWR) is widely used to locate and track human targets behind walls. However, it is difficult to obtain the prior information such as the position, speed and direction of the human targets, and target images often occur defocusing and deformation with the movement, resulting in difficulty in human targets location and tracking. Aiming at solving these problems, a target tracking algorithm combining kernelized correlation filter (KCF) and interactive multiple model (IMM) algorithm is proposed. In the proposed method, KCF is used to locate the target area on the radar image and obtain the target position measurement. Then, the obtained target position is taken as the input of IMM algorithm, which outputs the final tracking trajectory. Numerical simulation and experimental results show the effectiveness of the proposed algorithm.