Estimation of empirical parameters in matching of linear vector datasets: an optimization approach

The purpose of object matching is to identify corresponding objects in various datasets. This article provides a Geometric-based Matching framework based on the Optimization Approach, called GeMOA, for improvement of linear object matching in the datasets with different scales, sources, and production time. GeMOA performs object matching in different datasets, while considering only the extracted geometric criteria from the objects, removes any initial dependency on common empirical parameters such as threshold for spatial similarity degree, buffer distance, and weights of the criteria. Moreover, The proposed solution considers all existing relations between the objects, including 1:0, 0:1, 1:1, 1:N, N:1, and M:N. For assessment of GeMOA efficiency, three datasets of scales: 1:2000, 1:5000, and 1:25000, with different sources and collections time. The results show that GeMOA, contrary to many previous methods, does not lose its applicability in confronting datasets with different scales and sources, and attaints a better than 90% F-score. © 2017, Springer International Publishing AG.