Efficient Decimation of Polygonal Models Using Normal Field Deviation

Abstract

A simple and robust greedy algorithm has been proposed for efficient and quality decimation of polygonal models. The performance of a simplification algorithm depends on how the local geometric deviation caused by a local decimation operation is measured. As normal field of a surface plays key role in its visual appearance, exploiting the local normal field deviation in a novel way, a new measure of geometric fidelity has been introduced. This measure has the potential to identify and preserve the salient features of a surface model automatically. The resulting algorithm is simple to implement, produces approximations of better quality and is efficient in running time. Subjective and objective comparisons validate the assertion. It is suitable for applications where the focus is better speed-quality trade-off, and simplification is used as a processing step in other algorithms.