Semi automatic Road Extraction from Aerial Images

A semi automatic method for road finding is proposed in this paper. It can incorporate global shape information, grey feature and gradient feature for road finding in aerial image extraction. 'Semi automatic' means that a road is exacted automatically after some seed points have been given by an operator through the activation of a mouse. By this method, road finding is enhanced by both considering the boundary as a whole and using model based global shape information. Road finding is formulated as an optimization problem. Results from the method applied to road extraction from real aerial images are presented.