With the increasing popularity of route guidance systems, it becomes more important for such highly advanced applications as the cross detection to induce road maps with a higher accuracy. In this paper, we present a novel approach to induce higher-precision maps with the help of the GPS data, mainly identify intersections, curves, and bifurcation junctions. Our approach consists of several successive processing steps: Firstly, try to find out areas where the distribution of GPS feature points is very intensive. Then, analyze the characteristic of the intersections and attach some restrictions to filter out the intersections, curves and bifurcation junctions which meet with our conditions. We also propose this method to improve the efficient of our algorithm. Among the new contributions are a creative method that connects both the TargetRegion_Detect algorithm and the AbnormalPointFilter algorithm which helps us identify the three forms of roads mentioned above and induce road maps with a higher accuracy.
