Automated defect detection in uniform and structured fabrics using Gabor filters and PCA

• We present an algorithm for automated texture defect detection in fabrics.
• Detection and classification are based on a complex Gabor filter bank and PCA.
• Detection performance is evaluated by using TILDA Textile Texture Database images.
• Comparison with human-identified defect locations and existing literature results.
• The detection rate is 98.8% whereas the false alarm rate is 0.20–0.37%.

This paper describes an algorithm for texture defect detection in uniform and structured fabrics, which has been tested on the TILDA image database. The proposed approach is structured in a feature extraction phase, which relies on a complex symmetric Gabor filter bank and Principal Component Analysis (PCA), and on a defect identification phase, which is based on the Euclidean norm of features and on the comparison with fabric type specific parameters. Our analysis is performed on a patch basis, instead of considering single pixels. The performance has been evaluated with uniformly textured fabrics and fabrics with visible texture and grid-like structures, using as reference defect locations identified by human observers. The results show that our algorithm outperforms previous approaches in most cases, achieving a detection rate of 98.8% and a false alarm rate as low as 0.20–0.37%, whereas for heavily structured yarns misdetection rate can be as low as 5%.