Gray Relational Analysis for Recognizing Fabric Defects

Abstract

A fabric defect image is characterized by its prirmitive properties as well as the spatial relationships between them. A gray level co-occurrence can be specified in a matrix of the relative frequencies with which two neighboring pixels separated by a distance occur on the image. By applying the co-occurrence matrix and gray relational analysis of the gray theory, we can extract characteristic values of a fabric defect image and classify defects to recognize common problems, including broken warps, broken wefts, holes, and oil stains. Gray relational analysis is also used to investigate correlations of the analyzed factors among the selected characteristic indicators in a randomized factor sequence through data processing. By justifying the most correlated defects, recognition accuracy can reach 94%.