Defect Detection in Textiles Using Optimal Gabor Wavelet Filter

Abstract

For automatic visual inspection of industrial production on textile materials, a method based on optimal Gabor wavelet filter was presented in this paper. This method used a simple and fast process to get the optimal filer from a set of Gabor wavelet filters and utilized this filter to extract texture features of images. Considering the industrial requirement, we used fast algorithm in feature extraction to reduce the calculation time. Finally, we used a threshold of this features to produce a binary image of defects. The experiments on this approach yielded excellent results for various types of defects. The results show that the developed algorithm is robust, scalable and computationally efficient for defect detection and appropriate for real-time industry production