Registration of vector maps based on multiple geometric features in topological aspects

Abstract

Maps are one of the most intuitive tools for recording and representing geographic information. Digital maps composed of vectors are mostly used for preserving cartographic data in a compact and efficient manner. However, the registration of multiple vector maps is an essential step when geographic information from various sources is used in an integrated analysis. Typically, control points are used as observables to solve the transformation parameters of a registration model, while check points are used for quality assessment. As a consequence, poor distribution or an insufficient number of control or check points might lead to a biased registration solution. In this study, linear features and projective invariant points were both used as observables in a registration analysis between multiple vector maps. The goal was to provide an improved geometric constraint when solving the transformation parameters. Furthermore, two numerical indices, namely the absolute consistency and relative geometric similarity, were used to evaluate the quality of the registration solution. Based on the results of a case study, the proposed approach has been shown to be capable of better making use of the geometric connotation inherent in a vector map and providing a comprehensive quality evaluation for the obtained transformation solution. Consequently, a more reliable and robust registration analysis for digital vector maps can be achieved when the proposed approach is implemented.