Fabric defect inspection based on lattice segmentation and Gabor filtering

Abstract

Fabric defect inspection aims at detecting the defects presented on a patterned fabric surface to achieve high quality. However, visual inspection is challenging due to the diversity of the fabric patterns and defects. This paper presents an automatic defect inspection method which compares the similarities of semantic sub-images conformed to crystallographic groups called lattice. The lattices are automatically segmented based on morphological component analysis (MCA). The defect inspection is then formulated as a novel voting procedure depending on an ideal lattice artificially generated by investigating the distributions of responses given by convolving lattices with Gabor filters. The performance of the proposed method LSG (lattice segmentation assisted by Gabor filters) is evaluated on the databases of star- and box-pattern images. By comparing the resultant and ground-truth images, an overall detection rate of 0.975 is achieved, which is comparable with the state-of-the-art methods.