应用遗传算法的平面曲线最优多边形拟合

平面曲线的多边形拟合是模式识别和计算机视觉研究中非常重要的一类算法。该文提出了一种利用遗传算法（GA）的最优多边形拟合算法。曲线的分段拟合操作被方便地编码为基因串。一种基于面积和模型简单度的全局拟合度量用作GA的适值函数。该文算法可以在无需事先给定分段数目及最大允许偏洋的条件下自动确定这些参量进行多边形拟合。文中给出的实验结果表明了算法的可行性和有效性。

Optimal Polygonal Approximation of Planar Curve Using Genetic Algorithm

Polygonal approximation of a planar curve is an effective representation for shape analysis and patternrecognition. This paper provides a new method for optimal polygonal approximation that taking usage of geneticalgorithms(GAs). A global measure of the closeness between the polygonal to the eurve acts as the fitness function of GAs, By a simple and direct coding of chromosome, the optimal solution is then obtained, where no initial arguments(such as the number of vert.ices or ermr tolerance) need to be given. Some experimental results thatdemonstrate the practicable of our method are also shown in this paper.