Hierarchical distillation for image compressive sensing reconstruction

Bokyeung Lee; Bonhwa Ku; Wanjin Kim; Hanseok Ko

doi:10.1049/ell2.12284

Hierarchical distillation for image compressive sensing reconstruction

Bokyeung Lee, Bonhwa Ku, Wanjin Kim, Hanseok Ko

School of Electrical Engineering

Research output: Contribution to journal › Article › peer-review

1 Citation (Scopus)

Abstract

Compressive sensing (CS) is an effective algorithm for reconstructing images from a small sample of data. CS models combining traditional optimisation-based CS methods and deep learning have been used to improve image reconstruction performance. However, if the sample ratio is very low, the performance of the CS method combined with deep learning will be unsatisfactory. In this letter, a deep learning-based CS model incorporating hierarchical knowledge distillation to improve image reconstruction even at varied sample ratios. Compared to the state-of-art methods with all compressive sensing ratios, the proposed method improved performance by an average of 0.26 dB without additional trainable parameters.

Original language	English
Pages (from-to)	851-853
Number of pages	3
Journal	Electronics Letters
Volume	57
Issue number	22
DOIs	https://doi.org/10.1049/ell2.12284
Publication status	Published - 2021 Oct

Bibliographical note

Publisher Copyright:
© 2021 The Authors. Electronics Letters published by John Wiley & Sons Ltd on behalf of The Institution of Engineering and Technology

Keywords

Computer vision and image processing techniques
Image and video coding
Optical, image and video signal processing

ASJC Scopus subject areas

Electrical and Electronic Engineering

Access to Document

10.1049/ell2.12284

Cite this

@article{697c52af23144fd1a6a222ae28e4c882,

title = "Hierarchical distillation for image compressive sensing reconstruction",

abstract = "Compressive sensing (CS) is an effective algorithm for reconstructing images from a small sample of data. CS models combining traditional optimisation-based CS methods and deep learning have been used to improve image reconstruction performance. However, if the sample ratio is very low, the performance of the CS method combined with deep learning will be unsatisfactory. In this letter, a deep learning-based CS model incorporating hierarchical knowledge distillation to improve image reconstruction even at varied sample ratios. Compared to the state-of-art methods with all compressive sensing ratios, the proposed method improved performance by an average of 0.26 dB without additional trainable parameters.",

keywords = "Computer vision and image processing techniques, Image and video coding, Optical, image and video signal processing",

author = "Bokyeung Lee and Bonhwa Ku and Wanjin Kim and Hanseok Ko",

note = "Publisher Copyright: {\textcopyright} 2021 The Authors. Electronics Letters published by John Wiley & Sons Ltd on behalf of The Institution of Engineering and Technology",

year = "2021",

month = oct,

doi = "10.1049/ell2.12284",

language = "English",

volume = "57",

pages = "851--853",

journal = "Electronics Letters",

issn = "0013-5194",

publisher = "Institution of Engineering and Technology",

number = "22",

}

TY - JOUR

T1 - Hierarchical distillation for image compressive sensing reconstruction

AU - Lee, Bokyeung

AU - Ku, Bonhwa

AU - Kim, Wanjin

AU - Ko, Hanseok

PY - 2021/10

Y1 - 2021/10

N2 - Compressive sensing (CS) is an effective algorithm for reconstructing images from a small sample of data. CS models combining traditional optimisation-based CS methods and deep learning have been used to improve image reconstruction performance. However, if the sample ratio is very low, the performance of the CS method combined with deep learning will be unsatisfactory. In this letter, a deep learning-based CS model incorporating hierarchical knowledge distillation to improve image reconstruction even at varied sample ratios. Compared to the state-of-art methods with all compressive sensing ratios, the proposed method improved performance by an average of 0.26 dB without additional trainable parameters.

AB - Compressive sensing (CS) is an effective algorithm for reconstructing images from a small sample of data. CS models combining traditional optimisation-based CS methods and deep learning have been used to improve image reconstruction performance. However, if the sample ratio is very low, the performance of the CS method combined with deep learning will be unsatisfactory. In this letter, a deep learning-based CS model incorporating hierarchical knowledge distillation to improve image reconstruction even at varied sample ratios. Compared to the state-of-art methods with all compressive sensing ratios, the proposed method improved performance by an average of 0.26 dB without additional trainable parameters.

KW - Computer vision and image processing techniques

KW - Image and video coding

KW - Optical, image and video signal processing

UR - http://www.scopus.com/inward/record.url?scp=85127333165&partnerID=8YFLogxK

U2 - 10.1049/ell2.12284

DO - 10.1049/ell2.12284

M3 - Article

AN - SCOPUS:85127333165

SN - 0013-5194

VL - 57

SP - 851

EP - 853

JO - Electronics Letters

JF - Electronics Letters

IS - 22

ER -

Hierarchical distillation for image compressive sensing reconstruction

Abstract

Bibliographical note

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this