采用小波框架的纺织品缺陷分类方法  

Textile Defect Classification Based on Discriminative Wavelet Frames

作　　者：杨学志[1] 沈晶[1] 

机构地区：[1]合肥工业大学计算机与信息学院,安徽合肥230009

出　　处：《工程图学学报》2009年第5期108-112,共5页Journal of Engineering Graphics

基　　金：国家自然科学基金资助项目(60672120)

摘　　要：针对纺织品缺陷分类,提出了一种基于判别小波框架的分类方法。该方法采用小波框架来描述纺织品图像的多尺度纹理特性,并设计与纺织品缺陷纹理相对应的小波框架函数来替代标准小波函数,来更有效地描述各类缺陷纹理的内在结构差异。在判别特征提取训练方法框架下,通过将小波框架函数和分类器两者的设计相联合,来实现缺陷分类错误概率的最小化。对8类纺织品缺陷的466个样本,以及434个无缺陷样本进行了分类实验,获得了95.8%的分类准确率。A new method based on discriminative wavelet frames is proposed for the classification of textile defects. Multiscale texture properties of textile image are characterized by its wavelet frames representation. Instead of using the standard wavelet functions, wavelet frame function adapted to textile textures is designed for a more efficient characterization of structural differences between different defective textures. Under the framework of discriminative feature extraction （DFE） training, the wavelet frame function and a classifier are simultaneously designed with the common objective of minimizing classification errors. The proposed method has been evaluated on the classification of 466 defect samples containing eight classes of textile defects, and 434 nondefect samples, and 95.8% classification accuracy is achieved.

关 键 词：计算机应用 纺织品检测 小波框架 自适应小波 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]