Automatic Detection and Classification of Weaving Fabric Defects Based on Digital Image Processing

Abstract

this paper describes the detection and classification of fabric defects based on digital image processing. The work is intended to provide the higher speed and accuracy of defect detection than human vision and to find the source of the defects. At first, we find the size and position of wefts or warps from an image. Then calculate the pattern of weft and warp positions and figure out whether there is a defect or not. The patterns of weft and warp may differ based on the type of fabrics. Sample pattern of good fabric is used to detect and classify the defect of the fabric with same pattern. OpenCV library and python programming language is used for the experiment. Seven kinds of defects on the fabrics model images are detected and five real fabric images are used for the experiment. The experiment shows the result of successful defect detection with 95% rate, and it is 50% faster than human vision in fabrics density calculation.