提出了基于混沌优化的模糊聚类方法(COFCM)。COFCM将混沌优化策略与传统的模糊C-均值算法(FCM)相结合,用混沌变量搜索对模糊聚类目标函数进行全局寻优,同时结合梯度算子,使方法能有效收敛到极值点。以六组人工数据集和真实数据集作为实验数据,对聚类目标函数值、聚类有效性函数指标值进行对比实验,其结果表明COFCM能得到比FCM更好的目标函数值,从而有更好的聚类效果。最后将该方法应用于Lena图像进行图像分割,验证了方法的有效性。
Fuzzy Clustering Analytic Methods based on Chaos Optimization (COFCM) was proposed. COFCM combined chaos optimization strategy with conventional Fuzzy C-Mean Algorism (FCM), and it optimized fuzzy clustering objective function through chaos variables searching and gradient operator intended to achieve effective convergence. With six groups of artificial datasets and real datasets as experimental data, a comparative experiment between clustering objective function value and clustering validity function values demonstrates that COFCM can achieve a better objective function value and thus get a better clustering effect. Finally COFCM was applied to Lena image and the segmentation of it proves the effectiveness of this method.