ABSTRACT One way to assess the quality of building polygons on OpenStreetMap (OSM) is to find matches to buildings in authoritative data. A common approach to finding matches checks the similarity of geometric characteristics between polygons representative of the same objects. However, building on OSM, like all crowdsourced data, involves the characteristics of nonrigid deviation, shape homogenization, etc., which complicate the matching process. We propose an approach to match OSM building polygons to authoritative data to determine the local optimal geometric similarity in two main steps. First, we extract street blocks from OSM road networks and use street blocks as the unit of analysis to find matches. Second, to account for nonrigid deviations and shape homogenization of the building polygons in OSM, we extract local optimal geometric similarity factors and their weights in each street block to match the building polygons between the OSM dataset and the authoritative dataset. The experiment shows that the proposed method obtains the local optimal geometric similarity factors and their weights are obtained automatically without any manual intervention, and it can achieve a precision higher than 95% and the F‐measure () better than 95% for the combined measure of precision and recall.
Yu et al. (Mon,) studied this question.