Abstract: Accurately extracting and timely updating the information of road network is vital to road planning and vehicle navigation. Currently, the road network’s extraction method for mining urban trunk roads, based on GPS trajectories, commonly uses floating car or taxi. However, the existing methods ignore the automatic extraction of pathway, which is very important for earthquake relief, community navigation and village tour and other occasions. Therefore, this paper proposes a new method of road network extraction, based on walking GPS trajectories, which consists of three parts: data preprocessing, road centerline generation and road network accuracy evaluation. In this paper, three methods are adopted to generate road centerlines successively, such as trajectories clustering, cluster center segmentation and centerline fitting. Making experiment with self-collected walking GPS data, the results show that the proposed method not only is able to accurately extract the road network, coverage rate can reach 96.21% while error detection rate was 3.26%, but also can extract pathway and update road network.

Key words: road network extraction, walking trajectories, pathway extraction, clustering algorithm, curve fitting