Detection of Fabric Defects by Auto-Regressive Spectral Analysis and Support Vector Data Description

Abstract

Substituting computer vision for human eyes in the practice of fabric defect detection can enhance the detection efficiency, decrease labor force, reduce labor intensity and further improve product quality, and, consequently, can satisfy the demands of society for high-efficiency production as well as high-quality products. 1 There exist multifarious fabric defects that appear on a wide variety of fabrics in terms of material, weave, structure and density. Because Abstract For the purpose of realizing fast and effective detection of defects in woven fabric, and in consideration of the inherent characteristics of fabric texture, i.e., periodicity and orientation, a new approach for fabric texture analysis, based on the modern spectral analysis of a time series rather than the classical spectral analysis of an image, is proposed in this paper. Traditionally, a power spectral estimated by a two-dimensional Fast Fourier transformation (FFT) is usually employed in the detection of fabric defects, which involves a large computational complexity and a relatively low accuracy of spectral estimation. To this effect, this paper makes a one-dimensional power spectral density (PSD) analysis of the fabric image via a Burg-algorithm-based Auto-Regressive (AR) spectral estimation model, and accordingly extracts features capable of effectively differentiating normal textures from defective ones. A support vector data description is adopted as a detector in order to deal with defect detection, a typical task of one-class classification. Experimental results for the detection of defects from several fabric collections with different texture backgrounds indicate that a low false alarm rate and a low missing rate can be simultaneously obtained with less computational complexity. Comparison of the detection results between the AR model and the FFT method confirms the superiority of the proposed method.

A Survey on Class Imbalance Learning on Automatic Visual Inspection

“…Bu et al proponen en [66] un método para la inspección de textiles con texturas en los que usan un modelo de estimación espectral en 1D para representar los patrones de textura normales. En su método, los autores usan una SVDD para generar un clasificador que aprende las características de las texturas normales y por tanto clasifica como defectuosos los patrones que se desvían del modelo normal aprendido.…”

Fabric defect detection using an integrated model of bottom-up and top-down visual attention

“…In transformation domain, Tsai and Hsieh adopted a one-dimensional Hough transform in the Fourier domain of texture images to identify defective components; (Tsai & Hsieh, 1999) the Fourier transform is insensitive to the minor modifications to the frequency spectrum caused by local fabric defects. (Bu, Huang, Wang, & Chen, 2010;Jayashree & Subbaraman, 2012;Malek, Drean, & Bigue, 2013;Tsai & Hsiao, 2001) Sari-Sarraf and Goddard used the Daubechies filters to detect small defects at a detection rate of 89% (Sari-Sarraf & Goddard, 1999) Tsai and Hsiao presented wavelet reconstruction approach for the inspection of local defects embedded in homogeneous textured surfaces. (Tsai & Hsiao, 2001) However, this method can't suppress the singular information part of defect-free fabric texture.…”

A real-time computer vision-based platform for fabric inspection part 1: algorithm

“…Therefore, most detection algorithms aim at extracting such textural features either in spatial or spectral domain those are able to discriminate defects from normal textures. Those commonly used features include: first-order statistics (Zhang & Bresee, 1995), second-order statistics (Latif-Amet, Ertüzün, & Ercil, 2000;Wen, Chiu, Hsu, & Hsu, 2001), fractal dimensions (Bu, Wang, & Huang, 2009;Sari-Sarraf & Goddard, 1999), Fourier spectral features (Chan, 2000), wavelet transform coefficients (Kim & Kang, 2007;Yang, Pang, & Yung, 2004), Gabor wavelet features (Hou & Parker, 2007) and AutoRegressive spectral features (Bu, Huang, Wang, & Chen, 2010). Earlier comparison studies involving different texture analysis and image processing techniques suggest that those methods produce good results on certain fabric textures or defect types, and are often poor on others (Bodnarova, Bennamoun, & Kubik, 2000;Conci & Proena, 2000).…”

“…To further confirm the effectiveness of the proposed method, a comparison with a typical feature J. Zhou and J. Wang extraction-based method proposed by (Bu et al, (2010)) (codes available and denoted by auto-regression (AR) in Figure 7) has been carried out. Note that our method is obtained with patch size of 26 × 26 pixels, and the 32 × 32 patch size is used for AR method, as the data sequence obtained from a 26 × 26 patch is too short to be estimated by AR model.…”