Error measures in greedy insertion simplification methods

Abstract

In this work we study algorithms for obtaining simplified representations of terrain models. Terrain data is difficult to deal with, due to its complexity and size. Simplification methods are some of the techniques used to reduce its natural complexity. Among these methods, greedy insertion with vertical local error measure is considered to be the one that yields best results. In spite of its qualities, the method presents shortcomings, especially when applied to terrain with different height variation regions. We investigate the existence of better error measures to deal with this situation and propose the adoption of greedy insertion methods in which the local vertical error is modified according to the variability of the surface in a neighborhood of the point under consideration. The proposed error measure used with the greedy algorithms yields a significant improvement in the visualization process.