Attractors in Action
Attractors may be computed at any level of a QTM hierarchy, and point data allocated to them at to represent map data at different map scales and resolutions. Generalization can be performed by analyzing contents of attractors, as the example below (a portion of Schaffhausen Canton in Switzerland) illustrates:
At QTM level 16 there is about one attractor for each polygon vertex. At coarser levels of resolution, attractors are larger and contain more vertices, enabling them to act as spatial filters.
One linear generalization method, shown below, eliminates all but one of consecutive points that happen to fall in one attractor. The upper left panel illustrate the method, and the other panels step through 5 QTM levels, coarsening the outlines from a resolution of about 150 m to a resolution of 2 km.
