Employing united Delaunay triangulation in contour lines generalization

Abstract

Geomorphologic elements such as valley, ridge are structural information to represent natural geomorphology in spatial extension and distribution. The united Delaunay triangulation network methodology makes an improvement over traditional triangulation method, with its capability of detecting human recognizable bends. Although this united Delaunay triangulation network is more suitable for terrain structure extraction, it hasn't yet been applied in contour line generalization. This article aims at exploring its application in contour lines generalization. First, the curves of contours lines are extracted, and then at the curves local terrain structure lines are linked together. Later, a tree structure of these structure lines is built, representing their relations. In generalization, the square root model is chosen for specifying the number of valley selection. And based on the built triangulation network several important aspects to evaluate the importance of valley are taken as quantitative indexes for determining which valley needs kept and which removed. The general valley importance index calculated by this evaluation method affects the degree of generalization. In result, the generalized map generally shows no intersection and keeps the original terrain characteristic, and, thus exemplifying this united Delaunay triangulation model's great potential in contour line generalization.