【摘要】 通过对模糊C均值聚类算法进行分析评价,提出了一种高效的自适应图像数据聚类方法,该方法采用曲线的多项式拟合技术自动获取随数据分布动态变化的阈值,改进后的算法克服了模糊C-均值聚类对聚类中心的敏感性以及聚类的局部性,并在此基础上建立了相应的索引机制.仿真实验表明,自适应模糊聚类索引大大提高了检索性能.更多还原

【Abstract】 It depends on how to cluster the massive image data to build an image retrieval system in content-based image search engines.This paper introduces a newly adaptable fuzzy clustering algorithm after making an analysis of the fuzzy C-means clustering algorithm.The new algorithm has solved the locality and sensitiveness of the initial condition of the fuzzy C-means clustering,based on which an indexing mechanism was built.Simulations show that the newly adaptable fuzzy clustering performs well in retrieval efficiency.更多还原