Abstract
This paper proposes a method to detect corresponding vertex pairs between planar tessellation datasets. Applying an agglomerative hierarchical co-clustering, the method finds geometrically corresponding cell-set pairs from which corresponding vertex pairs are detected. Then, the map transformation is performed with the vertex pairs. Since these pairs are independently detected for each corresponding cell-set pairs, the method presents improved matching performance regardless of locally uneven positional discrepancies between dataset. The proposed method was applied to complicated synthetic cell datasets assumed as a cadastral map and a topographical map, and showed an improved result with the F-measures of 0.84 comparing to a previous matching method with the F-measure of 0.48.
Highlights
Map conflation of spatial datasets from different mapping agencies usually encounters locally uneven positional discrepancies between corresponding objects of the datasets
Final corresponding vertex pairs (CVPs) are obtained as nearest vertex pairs within a tolerance distance
Because the datasets are gradually aligned though the iteration, final CVPs are obtained as nearest vertex pairs within a tolerance distance
Summary
Map conflation of spatial datasets from different mapping agencies usually encounters locally uneven positional discrepancies between corresponding objects of the datasets. Given a point in one dataset, several candidate points in another dataset within a distance threshold are evaluated with similarity measures such as distance, and a single point with the highest similarity is chosen as the corresponding point [1] These similarities are affected by the aforementioned discrepancies. To find the above object pairs, intersection analysis has been applied which works well when the objects to be matched are sufficiently large and isolated each other within each dataset such as building objects This is because the positional discrepancies do not significantly affect the objects’ intersection relations. This is because the cells are mutually exclusive and collectively exhaustive, a cell in one dataset can significantly co-intersect cells in another dataset which represent different real-
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have