Urban Road Network Extraction Using Very High Resolution Data Based on Road Signatures Detection

摘要

This paper presents a robust method to extract the road network from an extremely high-resolution aerial image where the road signatures such as zebra crossings, cars, road lines can be seen in detail. In this paper, we propose a combining method using SIFT, SURF, and blob features to recognize zebra crossings. In addition, we use the digital surface model (DSM) data to overcome the shadow problem and to extract the road based on local thresholding, region growing and morphological operations. Finally, we construct the network based on road skeleton and b-spline. The experimental result shows that the proposed method works very well.