A Split-and-Merge Method for Creating Polygonal Homogeneous-Vegetation Regions from Digitized Terrain Data
Abstract:
Providing a simplified representation of terrain characteristics has applications to optimal-path-planning programs using spatial reasoning. Utilizing computer vision techniques, our program creates polygonal homogeneous- vegetation regions based on map vegetation data from a digitized Defense Mapping Agency Database. Boundary points for regions are identified from the vegetation codes in the database, and then the boundary contours of the regions are traced using a modified look-left boundary tracing algorithm. Each region is then represented by a polyline comprised of line segments that meet a minimum threshold for fit using the linear least-squares criterion. The segments are determined by first recursively splitting the region boundary until all segments meet the fit threshold, and then merging adjacent segments that meet the threshold. Theses.