A Fast Fabric Defect Detection Framework for Multi-Layer Convolutional Neural Network Based on Histogram Back-Projection

Abstract

In this paper we design a fast fabric defect detection framework (Fast-DDF) based on gray histogram back-projection, which adopts end to end multi-convoluted network model to realize defect classification. First, the back-projection image is established through the gray histogram on fabric image, and the closing operation and adaptive threshold segmentation method are performed to screen the impurity information and extract the defect regions. Then, the defect images segmented by the Fast-DDF are marked and normalized into the multi-layer convolutional neural network for training. Finally, in order to solve the problem of difficult adjustment of network model parameters and long training time, some strategies such as batch normalization of samples and network fine tuning are proposed. The experimental results on the TILDA database show that our method can deal with various defect types of textile fabrics. The average detection accuracy with a higher rate of 96.12% in the database of five different defects, and the single image detection speed only needs 0.72s. key words: back-projection, gray histogram, fabric detection, multi-layer convolutional neural network Guodong Sun is a professor in the School of Mechanical Engineering at the Hubei University of Technology. He received his BS degree in energy and power engineering and his PhD degree in mechanical and electronic engineering from Huazhong University of Science and Technology in 2002 and 2008, respectively. His current research interests include machine vision and imaging processing. Zhen Zhoureceived the B.S. degree from in 2017. He is currently pursuing the M.S. degree with the School of Mechanical Engineering, Hubei University of Technology, Wuhan, China. His current research interests are machine learning and imaging processing.

Classifying fabric defects with evolving Inception v3 by improved L2,1-norm regularized extreme learning machine

“…Sun et al. 25 used an end-to-end convolutional network model for defect classification on the textile texture database (TILDA) with five different defects and obtained 96.12% average classification accuracy in the experiment. Chen et al.…”

“…However, this involved the processing of simple images and did not reflect the classification performance of the algorithm on a complex-texture fabric defect dataset. Sun et al 25 used an end-to-end convolutional network model for defect classification on the textile texture database (TILDA) with five different defects and obtained 96.12% average classification accuracy in the experiment. Chen et al 26 proposed a Gabor-filtering faster R-CNN fabric defect classification method to solve the interference problem caused by complex background textures.…”

Deep learning-based fabric defect detection: A review

“…Sun et al. 53 proposed a rapid fabric fault detection method relying on grayscale histogram back-projection and an end-to-end multiconvoluted network model. The results are obtained for average detection accuracy of 96.1%.…”

“…Investigated papers were analyzed and summarized in terms of method, dataset, classification or number of classes, performance as success, and comparison. 15,24,44–49,52–77…”

Fabric Defect Detection Using Customized Deep Convolutional Neural Network for Circular Knitting Fabrics

“…The proposed method can detect different defect types by optimizing a deep semantic segmentation network. According to gray histogram back-projection, Guodong [10] developed a fast defect-detection-framework (Fast-DDF), To address the problem of adjusting network model parameters and long training time, an end-to-end multi convoluted network model is used for defect classification, as well as batch normalization of samples and a network fine tuning process. Jing [11] introduces a deep convolutional neural network-based (CNN) detection approach for autonomous fabric defect detection.…”