DIGITIZING AND CLASSIFYING WOVEN FABRIC DEFECTS


Ikiz Y., Mutlu Ala D. M.

TEKSTIL VE KONFEKSIYON, vol.22, no.4, pp.346-353, 2012 (Journal Indexed in SCI) identifier identifier

  • Publication Type: Article / Article
  • Volume: 22 Issue: 4
  • Publication Date: 2012
  • Title of Journal : TEKSTIL VE KONFEKSIYON
  • Page Numbers: pp.346-353

Abstract

The aim of this research is to digitize certain woven fabric defects from images of woven fabrics, taken by a CCD line scan camera. %100 cotton, plain and twill woven raw fabrics were used in the experiments. Using a lighted fabric quality control board, 2048*4096 pixels BMP format images of the fabrics were generated by a CCD line scan camera. Defected areas of the images were selected and classified by referring the fabrics. Average gray scale values and dimensions of the defected areas (missing pick, irregular pick density, starting mark, double pick, broken pick, broken end, hole-tear, oily spot, oily end, wrong drawing) were measured with the help of Photoshop CS3 program and results were compared with the regular image areas. Results showed that classification of fabric defects requires much more complicated algorithms than simple thresholding for industrial application of automated fabric quality control.

The aim of this research is to digitize certain woven fabric defects from images of woven fabrics, taken by a CCD line scan camera. %100 cotton, plain and twill woven raw fabrics were used in the experiments. Using a lighted fabric quality control board, 2048*4096 pixels BMP format images of the fabrics were generated by a CCD line scan camera. Defected areas of the images were selected and classified by referring the fabrics. Average gray scale values and dimensions of the defected areas (missing pick, irregular pick density, starting mark, double pick, broken pick, broken end, hole-tear, oily spot, oily end, wrong drawing) were measured with the help of Photoshop CS3 program and results were compared with the regular image areas. Results showed that classification of fabric defects requires much more complicated algorithms than simple thresholding for industrial application of automated fabric quality control.