Matching Road Network Combining Hierarchical Strokes and Probabilistic Relaxation Method

Abstract

Aiming at the complex multisource road network matching modes, this paper proposed a road network match- ing method based on hierarchical stable strokes. The progress of the matching method adopts recursive method and gives consideration to both global consistency and local similarity of homonymous road features. One layer of hierarchical strokes generating from road spatial structure are selected at a match and the matched parts are used as a stable reference for the next layer. The detailed matching is implemented by an iterative probability relaxation method. In the course of coarse matching three similarity indicators, including the geometry shape similarity of strokes, the relative position of nodes and strokes and the topological structure similarity of nodes, are introduced into the initial matching probability calculation. After that, an iterative calculation is carried out for updating matching probability. Finally, a self-designed se- lecting principle is applied to trace the exact matching results. Matching road centerline from different producers of the same geographical area shows that our method finds 1:1, 1:N and M:N matching modes and gets a satisfactory matching result, which means hierarchical stable reference and the stroke structure can provide an effective way to identify different matching relations and avoid the non-rigid deviation of the homonymous features.