Target tracking algorithm based on multiscale analysis and combinatorial matching.

Aiming at improving the accuracy of current tracking algorithms and tracking the target more quickly, a new target tracking method that is from multiscale analysis is presented, which is combined with combinatorial polygon matching. First, a target's range is estimated by analyzing multiple scales in the reference frame. Then, the pixels are judged as the target pixels within the target range determined in the previous step by calculating the relevant radial distance and judging the similarity of radial distance between target pixels to be detected and its background pixels. At the same time, the background modeling is updated according to the similarity between the incremental code of the background pixels and the corresponding pixels in the observed images. Last, the target trajectory is obtained by dynamic polygon matching. When encountering external interference, the performance of the improved algorithm is relatively stable. Compared with the existing methods, the proposed algorithm in this paper is verified by simulation, which is simple in operation and can track the target quickly and accurately. [ABSTRACT FROM AUTHOR]