Rotation-Based Scale Adaptive Moving Target Tracking Algorithm

Aiming at the problem of poor tracking effect of correlation filters in dealing with in-plane rotation and scale changes, this paper proposed a scale adaptive correlation filter target tracking algorithm with rotation characteristics based on ECO_HC (efficient convolution operators handcraft). Firstly, we train a scale and rotation filter, and then use the phase correlation algorithm to obtain the scale factor and rotation angle. Secondly, we adopt a dynamic adaptive update strategy for rotation and scale updating. Finally, in the position model update stage, we fuse the background information of front frames to enhance the stability of the template. Experimental data shows that our method is not only robust against in-plane rotation and scale changes, but also can meet real-time requirements.