未确知均值聚类

利用未确知系统理论分析特征对样本分类所作贡献,定义特征的分类权重,并作为启发性知识用于确定样本与各类间的加权距离及样本属于各类的隶属度,建立未确知均值聚类算法。IRIS数据检验表明,未确知均值聚类算法误判样本数少、收敛速度快、鲁棒性好,是一种实用、有效的无监督聚类算法。

Uncertain means clustering

A new semi-fragile watermarking algorithm for image content authentication based on chaotic mapping is presented.Chaotic sequence generated from the block itself by means of Logistic mapping is used to generate watermarking embedding;then the generated watermarking information is embedded into LSB plane of another image block with Torus Automorphism mapping,which the corresponding relation is determined by the secret key to establish the correlation among the image blocks.The results show that the method could localize the tampered watermarked image by using the high sensitivity on initial value of the chaotic mapping without host image in watermark extraction.Moreover,watermarked images obtained have good subjective quality,precise localization of tampered areas,and the algorithm is simple and safe.