基于灰度共生矩阵的板材纹理模糊分类器设计

摘要

We designed a classifier based on fuzzy logic for straight and curved texture.We extracted energy and correlation from the features of gray level co-occurrence matrix ( GLCM) by analyzing its scatter, and then designed the fuzzy domain and the membership function for the extracted features and texture patterns .We constructed fuzzy reasoning and the lookup ta-ble by the extracted fuzzy rules and Mamdani gravity method.The experiments were conducted with 100 straight and curved textures with the classification rate of 90.5%.%针对板材表面的纹理特征，设计了直纹、弯纹二维模糊逻辑分类器。首先对灰度共生矩阵提取的板材特征量进行散点分布分析，提取出用于纹理分类的能量和相关性基本特征；设计能量、相关性与纹理的模糊论域及隶属度函数，完成了特征量与纹理类别的模糊化；通过提取样本的模糊规则和Mamdani重心法，完成模糊推理，并得到纹理分类查询表。应用设计的纹理模糊分类器对样本库中100幅直纹和弯纹样本进行分类实验，分类总体正确率达到90．5％。