基于相对隶属度的织物风格模糊聚类分析

分析比较了统计数学、模糊数学和灰色数学在织物风格分析中的缺陷.特别针对模糊数学需人为确定隶属函数的缺陷提出采用基于相对隶属度的织物风格模糊聚类分析方法.以相对隶属度矩阵和模糊聚类中心矩阵的迭代求解为依据,对苎麻等十种织物的风格测试参数进行了聚类分析.通过聚类计算得到了织物相对类别特征值的聚类结果.经验证认为该结果符合客观实际和人们的感知经验.

The Fuzzy Cluster Analysis of Fabric Style Based on Relative Membership Degree

In fabric style study the often used analysis approaches are statistics,fuzzy and grey mathematics.In this article an overview among these analysis methods has been given.Comparison study has been carried out to identify the inaccuarcy of the methods.In particular,it was demonstrated that inaccuracy of using fuzzy approach in the fabric style analysis was caused by the subjective membership function.As an alternate,it was proposed to adopt the fuzzy cluster analysis method in fabric style study,which is under the relative membership degree.By using this method an iteration process was involved to solve relative grade of membership matrix and fuzzy cluster central matrix.This new approach of cluster analysis was then performed on the style testing parameters over ten types of fabric,such as ramie,etc.The eigen value of relative classification was obtained.The results were verifed.