Abstract
These days we live in a digital era where most societies rely on applications that depend on geospatial data. In addition, most of the recent road network maps are represented in vector format and they have accurate road points coordinates that form the road segments representing the roads. However, there are data discrepancies between maps for various reasons. Therefore, if there is a matching process between vector-vector road network datasets, the data discrepancy will cause some incorrect matching pairs. This paper presents a novel solution inspired by Hausdorff distance to confirm if the candidate similar roads from different maps are really similar to each other or not. We use local divergence measurements that make sure these candidate roads have approximately the same length and also run in parallel to each other, which preserves the shape between them. Confirming the similarity requires also a global divergence measurement to be met that ensures the candidate roads are for the same road, not different roads that happen to be beside each other having similar length and shape. Moreover, this approach has the capability to identify the similar roads when one of the roads has either missing road segments or extra incorrect road segments. The experiments for our method shows promising results.