Object Surface Representation Via NURBS and Genetic Algorithms with SBX

摘要

A technique to represent object surface via NURBS and genetic algorithms is presented. In this technique, the surface is generated based on control points. Then, the control points and the weights are optimized via genetic algorithms to find the NURBS, which represents the object surface. The genetic algorithm is constructed through an objective function, which is deduced from the NURBS surface. This objective function is minimized by using the simulated binary crossover. The proposed genetic algorithm improves accuracy and speed of the NURBS surface representation. The contribution of the proposed method is elucidated by an evaluation based on model accuracy and speed of traditional genetic NURBS surface representation.