A robust object tracking algorithm based on SURF

Abstract

In this paper, we propose a robust object tracking algorithm based on SURF. First, we adopt a two-stage matching method to improve the accuracy of SURF matching points. Then a template update method is used to deal with the problem of object appearance change. Next we use matching points between new template and candidate region to locate the initial position of object. As template update and object occlusion will cause the accumulation of tracking errors, therefore, at last fixed template is used to correct object's position. For those frames which have little matching points, we use Meanshift instead of SURF to track object. The experiments demonstrate that our work is robust and can track object accurately in complex environments.