A new approach for urban roads detection using laser data and aerial digital images

Abstract

The automatic feature extraction from digital aerial images is not a trivial task mainly due to occlusion problems, shadows and different viewpoints. To obtain an improved feature extraction we used laser data, which have additional information such as height and material type of the surface. In this paper we performed the combination of digital image and laser data in order to improve the results of automatic extraction of urban roads. Initially, the urban roads were detected from the response of laser information; in the sequence we applied two different approaches to connect the disconnected road segments. The results were very promising, with sensitivity rate of 92%.