Semi-Automatic Road Extraction Algorithm from IKONOS Images Using Template Matching

This paper describes the development of a semi-automatic road extraction algorithm. The algorithm works as follows: First, a user provides an initial road seed per each road segment in the image. Next, orientation of the road seed is calculated automatically by applying Burns line extraction algorithm. A road template is then formed around the road seed and a road segment is extracted automatically through template matching. For template matching, an adaptive least square matching algorithm was used. This algorithm puts weights on road central line parts and postulates that the relationship between template and target windows can be modeled by similarity transformation. This algorithm also assumes that template and target windows have only small differences in brightness values. A 1m resolution IKONOS image over Seoul area was used for this algorithm test. The algorithm extracted road central lines in any orientation and with moderate curvature successfully, after road seeds were given from a user. Current limitations are that the algorithm may not work on the road cast by shadow and that a user must select valid road seeds on road central lines since the algorithm itself can not judge the validity of input seeds. These limitations are currently being examined. The contribution of this paper is that it showed template matching, instead of the well known “snake”, could be used for road extraction.