Combining Fisher Criterion and Deep Learning for Patterned Fabric Defect Inspection

Yundong LI; Jiyue ZHANG; Yubing LIN

doi:10.1587/transinf.2016EDL8101

Combining Fisher Criterion and Deep Learning for Patterned Fabric Defect Inspection

Yundong LI, Jiyue ZHANG, Yubing LIN

Full Text Views

0

Cite this

Summary :

In this letter, we propose a novel discriminative representation for patterned fabric defect inspection when only limited negative samples are available. Fisher criterion is introduced into the loss function of deep learning, which can guide the learning direction of deep networks and make the extracted features more discriminating. A deep neural network constructed from the encoder part of trained autoencoders is utilized to classify each pixel in the images into defective or defectless categories, using as context a patch centered on the pixel. Sequentially the confidence map is processed by median filtering and binary thresholding, and then the defect areas are located. Experimental results demonstrate that our method achieves state-of-the-art performance on the benchmark fabric images.

Publication: IEICE TRANSACTIONS on Information Vol.E99-D No.11 pp.2840-2842

Publication Date: 2016/11/01

Publicized: 2016/08/08

Online ISSN: 1745-1361

DOI: 10.1587/transinf.2016EDL8101

Type of Manuscript: LETTER

Category: Image Recognition, Computer Vision

Authors

Yundong LI
  North China University of Technology
Jiyue ZHANG
  North China University of Technology
Yubing LIN
  North China University of Technology

Keyword

deep learning, fabric defect inspection, stacked autoencoders, Fisher criterion

Cite this

Copy

Yundong LI, Jiyue ZHANG, Yubing LIN, "Combining Fisher Criterion and Deep Learning for Patterned Fabric Defect Inspection" in IEICE TRANSACTIONS on Information, vol. E99-D, no. 11, pp. 2840-2842, November 2016, doi: 10.1587/transinf.2016EDL8101.
Abstract: In this letter, we propose a novel discriminative representation for patterned fabric defect inspection when only limited negative samples are available. Fisher criterion is introduced into the loss function of deep learning, which can guide the learning direction of deep networks and make the extracted features more discriminating. A deep neural network constructed from the encoder part of trained autoencoders is utilized to classify each pixel in the images into defective or defectless categories, using as context a patch centered on the pixel. Sequentially the confidence map is processed by median filtering and binary thresholding, and then the defect areas are located. Experimental results demonstrate that our method achieves state-of-the-art performance on the benchmark fabric images.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2016EDL8101/_p

Copy

@ARTICLE{e99-d_11_2840,
author={Yundong LI, Jiyue ZHANG, Yubing LIN, },
journal={IEICE TRANSACTIONS on Information},
title={Combining Fisher Criterion and Deep Learning for Patterned Fabric Defect Inspection},
year={2016},
volume={E99-D},
number={11},
pages={2840-2842},
abstract={In this letter, we propose a novel discriminative representation for patterned fabric defect inspection when only limited negative samples are available. Fisher criterion is introduced into the loss function of deep learning, which can guide the learning direction of deep networks and make the extracted features more discriminating. A deep neural network constructed from the encoder part of trained autoencoders is utilized to classify each pixel in the images into defective or defectless categories, using as context a patch centered on the pixel. Sequentially the confidence map is processed by median filtering and binary thresholding, and then the defect areas are located. Experimental results demonstrate that our method achieves state-of-the-art performance on the benchmark fabric images.},
keywords={},
doi={10.1587/transinf.2016EDL8101},
ISSN={1745-1361},
month={November},}

Copy

TY - JOUR
TI - Combining Fisher Criterion and Deep Learning for Patterned Fabric Defect Inspection
T2 - IEICE TRANSACTIONS on Information
SP - 2840
EP - 2842
AU - Yundong LI
AU - Jiyue ZHANG
AU - Yubing LIN
PY - 2016
DO - 10.1587/transinf.2016EDL8101
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E99-D
IS - 11
JA - IEICE TRANSACTIONS on Information
Y1 - November 2016
AB - In this letter, we propose a novel discriminative representation for patterned fabric defect inspection when only limited negative samples are available. Fisher criterion is introduced into the loss function of deep learning, which can guide the learning direction of deep networks and make the extracted features more discriminating. A deep neural network constructed from the encoder part of trained autoencoders is utilized to classify each pixel in the images into defective or defectless categories, using as context a patch centered on the pixel. Sequentially the confidence map is processed by median filtering and binary thresholding, and then the defect areas are located. Experimental results demonstrate that our method achieves state-of-the-art performance on the benchmark fabric images.
ER -