Map Retrieval Based on Geometric Spatial Relations

TL;DR: This paper generates a graph structure for representing the relationships among geographical objects using the three relations and defines the geometric spatial relation which is composed of an overlapping relation, an adjoining relation, and a neighboring relation.

Abstract: Until now, many researchers have proposed spatial relations between geographical objects in the field of geographical information systems and spatial databases. However, these proposals have focused on binary relations to represent the features of geographical objects, but have not concentrated on the complexity among geographical objects. In this paper, we aim to model the complex structure among geographical objects. To represent the structure among geographical objects, we define the geometric spatial relation which is composed of an overlapping relation, an adjoining relation, and a neighboring relation. Additionally, we discuss a retrieval function which finds out similar locations by using a sketch as a query. In order to find out locations, we generate a graph structure for representing the relationships among geographical objects using the three relations. We find a similar structure by comparing two graph structures. To decide a similar structure, we define the similarity which is calculated by correspondence among relations. The correspondence among relations is the value which shows the difference between a query and data. By using this similarity, we can find not only sub-graphs but also similarity graphs.