Open Collections

Abstract

Terrain surface simplification or approximation is very important in many terrainrelated applications. Due to a variety of reasons such as the data acquisition method and more detailed features information, terrain models are extremely, and not necessarily, large nowadays. A fast algorithm that can produce compact yet highly accurate model would be desirable. In this thesis, a hybrid algorithm integrating greedy insertion and quadric error based approximation is given. This algorithm is designed to provide a highquality approximation of original model, yet reasonably fast. Experiments show quality improvements of the approximated models over previous methods.