Detecting Fabric Defects with Computer Vision and Fuzzy Rule Generation. Part II: Defect Identification by a Fuzzy Expert System

Abstract

A new method for a fabric defect identifying system uses fuzzy inference in multicondition approximate reasoning and is capable of defect identification. The system uses fuzzy inference rules, and the membership function for these rules adopts a neural network approach. Only a small number of fuzzy inference rules are required to make the identifications of nondefect, slub (warp direction), slub (weft direction), nep, and composite defect. One fuzzy inference rule can replace many crisp rules. With this method, we can design a reliable system for identifying fabric defects. Experimental results with this approach have demonstrated an identification ability comparable to that of a human inspector.