Presentation 2023-03-02
QR code image dnoising netwroks based on decodability assessment
Kazumitsu Takahashi, Makoto Nakashizuka,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) In this paper, an image denoising method for QR code images is proposed. The image recovery from the degraded QR code image is performed to improve the successful rate of the data decoding from the QR code images. In many image recovery algorithms, which include image denoising, the objective of the algorithm is to decrease the mean square error between the clean image and the recovered image. However, the decrement of the mean square error is not related to improve the decoding rate directly. There exist any possible metric of the recovery that can improve the decoding rate. In this paper, the CNN (convolutional neural network) is applied to decodability assessment of the QR code and the trained network is employed to train the denosing network that improve the decoding rate of QR codes. In the proposed method, the CNN is trained to predict success or fail of decoding of the existing QR code decoder. The output of this CNN is defined as the decodability. Then, the denosing CNN is trained to minimize the loss function that consists of this decodability and the fidelity of the output image. In experiments, the decodability assessment of the CNN is demonstrated. We show that the decoding rate obtained from the denoising neural network that is trained with the decodability is superior to the denoising neural network trained with only mean square error.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) convolutional neural network / denoising / QR code / image recovery / decode
Paper # SIS2022-46
Date of Issue 2023-02-23 (SIS)

Conference Information
Committee SIS
Conference Date 2023/3/2(2days)
Place (in Japanese) (See Japanese page)
Place (in English) Chiba Institute of Technology
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair Tomoaki Kimura(Kanagawa Inst. of Tech.)
Vice Chair Naoto Sasaoka(Tottori Univ.) / Hakaru Tamukoh(Kyushu Inst. of Tech.)
Secretary Naoto Sasaoka(NTT) / Hakaru Tamukoh(Kansai Univ.)
Assistant Yoshiaki Makabe(Kanagawa Inst. of Tech.) / Yosuke Sugiura(Saitama Univ.)

Paper Information
Registration To Technical Committee on Smart Info-Media Systems
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) QR code image dnoising netwroks based on decodability assessment
Sub Title (in English)
Keyword(1) convolutional neural network
Keyword(2) denoising
Keyword(3) QR code
Keyword(4) image recovery
Keyword(5) decode
1st Author's Name Kazumitsu Takahashi
1st Author's Affiliation Chiba Institute of Technology(CIT)
2nd Author's Name Makoto Nakashizuka
2nd Author's Affiliation Chiba Institute of Technology(CIT)
Date 2023-03-02
Paper # SIS2022-46
Volume (vol) vol.122
Number (no) SIS-410
Page pp.pp.33-36(SIS),
#Pages 4
Date of Issue 2023-02-23 (SIS)