Target Tracking Algorithm Based on Local Region Restoration

Abstract

The traditional target tracking algorithm, in the case where the target is blocked by smoke, will fail. In order to solve this problem, this paper presents a new target tracking algorithm which is called that the target algorithm based on local region restoration. Firstly, we restore the region blocked by smoke in the area to be matched of the real image by using the gradient feature of the refer image, which is smokeless. Then calculating the similarity between the refer image and the area to be matched, the location of the target in the real image is the region with the highest similarity. In this algorithm, the dynamic and static characteristics of the smoke are used to determine the area blocked by smoke, and the local area recovery is a method based on dynamic gradient. The experimental results indicate that the tracking algorithm based on local region recovery can achieve better effect when the target is seriously blocked by smoke.