Abstract: The existing matching methods for the multi-source road data of the same scale mainly consider the characteristics of the road itself, while the effect of the neighborhood features on matching process is generally ignored, which may restrict the further improvement of the matching results. This restriction can be more obvious for the matching data in which the location or rotation differences still exist after the system error rectification. A road network matching method that takes the similarity of the roads' neighborhood habitation cluster into consideration is proposed, which draws on the experience of the human spatial cognitive characteristics for the unfamiliar environment. Firstly, the neighborhood habitation cluster of the road is extracted by the urban skeleton line network; Then by calculating the spatial relation similarity and geometry characteristic similarity of the neighborhood habitation cluster, the matching results can be obtained. The advantage of this method is that for the road data which have obvious location or rotation differences, the similarity of their neighborhood habitation clusters can be treated as a proper matching index. Actually, roads' neighborhood habitation cluster can be a constraint of the road matching process and enhance its robustness. The tests and comparison analysis indicate that this method can solve the matching problems of the road data which still have obvious location or rotation differences after system error rectification and improve the matching correctness.

Key words: multi-source road networks matching, urban skeleton-line network, neighborhood environment, habitation cluster, area cluster similarity calculation