GLCM-based chi-square histogram distance for automatic detection of defects on patterned textures

TL;DR

This paper introduces a novel machine vision algorithm that uses GLCM-based chi-square histogram distance and hierarchical clustering to automatically detect defects in patterned textures, inspired by human visual second-order statistics.

Contribution

The paper proposes a new defect detection method combining GLCM-based chi-square distance with hierarchical clustering for patterned textures.

Findings

Effective defect detection on wallpaper fabric images

Reduced computation time through gray level quantization

Successful differentiation of defective and defect-free blocks

Abstract

Chi-square histogram distance is one of the distance measures that can be used to find dissimilarity between two histograms. Motivated by the fact that texture discrimination by human vision system is based on second-order statistics, we make use of histogram of gray-level co-occurrence matrix (GLCM) that is based on second-order statistics and propose a new machine vision algorithm for automatic defect detection on patterned textures. Input defective images are split into several periodic blocks and GLCMs are computed after quantizing the gray levels from 0-255 to 0-63 to keep the size of GLCM compact and to reduce computation time. Dissimilarity matrix derived from chi-square distances of the GLCMs is subjected to hierarchical clustering to automatically identify defective and defect-free blocks. Effectiveness of the proposed method is demonstrated through experiments on defective real-fabric images of 2 major wallpaper groups (pmm and p4m groups).