Structural Pattern Recognition And Simplification Of Road Junctions

With the wide application of GIS, it is getting clearer and clearer that the multi-representation database is needed in sophisticated GIS application. The current pactice of multi-repsentation database is that of multiple but isolated representations in the database. Such method instroduces difficulties in data updating and using. It is desired that multiple representations of various levels of detail can be automatically derived from a single detailed representation. That raises the need of automatic map generalization in GIS.This research is focused on road junction model generalization, i.e. deriving simplified representations of complex road junctions from its detailed representation. The data used in the experiments are that of the road network of Hong Kong and Chengdu. A process of object description, identification, and generalization is adopted in the study. Firstly, typical junctions and instance data are represented as directed attributed relational graph (DARG). A node in the DARG represents a road segment constituting a road junction. An edge in the DARG represents connection relationship between two road segments. Then, junction patterns are searched in the instance data DARG That is the process of subgraph matching. Ullman's algorithm for subgraph matching is adopted in this study. Once a junction is recognized, it can be simplified according to some predefined method. The selections of road junction are based on structural analysis of typical junctions and the result of perceptual grouping. The simplification of a road junction involves both geometric and topological simplification. Topological simplification is based on graph theory and can maintain road network connectivity. Geometric simplification is guided by topological simplification to produce consistent results.In this research, an experimental system of road junction generalization is implemented with Visual Basic .Net on the basis of ArcGIS Engine. The experimental results show that, the suggested method is successful in identifying and simplifying the road junctions. The proposed method is considered to be useful in road network generalization and the generalization of other geographic features as well.