Defect Detection and Classification for Plain Woven Fabric Based on Deep Learning

Abstract

Defect detection for plain woven fabric is investigated in this paper. The original RGB images are converted into gray scale images and further enhanced by gray level adjustment. Several different filtering methods are used to denoise for comparison and ideal low pass filter is selected for the filtering of adjusted gray scale images. Deep learning based on convolutional neural network is suggested for the detection and classification of the filtered images using the VGG model. Experiment results verified the effectiveness of the proposed method.