Road tracking by circular template matching from high resolution remotely sensed imagery

Template matching has been brought effective results in road tracking. Actually, the shape of template plays a very important role. In this paper, a novel circular template is utilized to search the center points of the road. The circular template is not necessary to rotate for the isotropy of template when searching the optimal road axis point, which means less computation. At the same time, least square matching method is employed to trace the road trajectory. The main purpose of the paper is to propose a circular template that improves the performance value of tracking road axis. Experiments show that the circular template matching method (CTM) is more efficient than the rectangular template matching (RTM).