Abstract:Road curvature and lane number information (hereinafter to be referred as road information) play an important and necessary role in transportation research. In order to avoid the drawbacks, i.e., high cost and poor time effectiveness, of traditional ways existing to obtain the road information, it has become a hot topic to mine road information through GPS data. In this paper, we propose a novel approach for mining road information from low precision GPS data, which including: a) Analyze and test the distribution of real world GPS data; b) Propose the weighted approximation least squares method (WALSM) to mine the curvature information of the road so as to establish the optimal center line of the road; c) Establish the GPS Data Distribution Variance - Road Width Discrete Model (DV-RWDM) so as to get the road lane number information. We demonstrate our approach using real world low precision GPS data and the results show that we can mine the road information with high accuracy. This provide us a lower cost and high timeliness guideline for the processing and application of low precision GPS data.

A Novel Approach for Mining Road Information from Low Precision GPS Data