Quadric surface extraction using genetic algorithms

Abstract

This article presents a new surface extraction method based on genetic algorithms (GA). The proposed method is capable of extracting all types of quadric surfaces (including flat surfaces) with a single surface representation. This representation is first fitted to predefined subsets of data points by a least square fitting algorithm. The evolution of the surface representations is achieved by repetitive application of crossover and mutation operations until a termination condition is met. The expression is finally classified into a specific quadric surface according to a classification table. The proposed method can be used for CAD model reconstruction of 3D objects composed of plane and quadric surfaces.