Improvement of Kernel Correlation Filtering Algorithm Based on Kalman Filter

Abstract

Kernel correlation filtering algorithm is based on the concept of correlation filtering. By introducing methods such as kernel functions and cyclic matrices, the target tracking speed is improved to a new level while ensuring high tracking accuracy. However, its tracking effect is not good when the target changes at multiple scales and the target is blocked. In view of the above problems, this paper proposes a multi-scale calculation method based on depth information optimization to solve the problem of multi-scale changes in the target. Based on the introduction of occlusion detection, a kernel-correlation filtering algorithm based on Kalman filtering is proposed to solve the problems of occlusion and jitter problem. The improved algorithm is tested on the TOB-50 data set and compared with the original algorithm. The results show that the improved algorithm proposed in this paper has better performance than the original algorithm when the target scale changes, occlusion, and screen jitter.