Efficient Polygon Amalgamation Methods for Spatial OLAP and Spatial Data Mining

The polygon amalgamation operation computes the boundary of the union of a set of polygons. This is an important operation for spatial on-line analytical processing and spatial data mining, where polygons representing different spatial objects often need to be amalgamated by varying criteria when the user wants to aggregate or reclassify these objects. The processing cost of this operation can be very high for a large number of polygons. Based on the observation that not all polygons to be amalgamated contribute to the boundary, we investigate in this paper efficient polygon amalgamation methods by excluding those internal polygons without retrieving them from the database. Two novel algorithms, adjacency-based and occupancy-based, are proposed. While both algorithms can reduce the amalgamation cost significantly, the occupancy-based algorithm is particularly attractive because: 1) it retrieves a smaller amount of data than the adjacency-based algorithm; 2) it is based on a simple extension to a commonly used spatial indexing mechanism; and 3) it can handle fuzzy amalgamation.