RestoreNet: a deep learning framework for image restoration in optical synthetic aperture imaging system

Ju Tang, Kaiqiang Wang, Zhenbo Ren, Wei Zhang, Xiaoyan Wu, Jianglei Di, Guodong Liu, Jianlin Zhao

Research output: Contribution to journalArticlepeer-review

42 Scopus citations

Abstract

Imaging blur is an inevitable problem in optical synthetic aperture imaging system because of the low response of frequency. Non-blind deconvolution algorithms are usually used for image restoration to obtain clear and high-resolution images. However, accurate prior information on the optical transfer function of system is required in the non-blind methods. As a data-driven approach, recent developments in deep learning have shown great potential in image processing. In this paper, a U-shaped deep learning framework named RestoreNet is proposed for image restoration, especially for removing the blur of optical synthetic aperture imaging system in a blind way. Numerical simulation and experiment results show that RestoreNet is an effective alternative with great restoration ability, stability and generalization in the image restoration of optical synthetic aperture imaging system.

Original languageEnglish
Article number106463
JournalOptics and Lasers in Engineering
Volume139
DOIs
StatePublished - Apr 2021

Keywords

  • Deep learning
  • Image deblur
  • Optical synthetic aperture imgaing system
  • Optical transfer function

Fingerprint

Dive into the research topics of 'RestoreNet: a deep learning framework for image restoration in optical synthetic aperture imaging system'. Together they form a unique fingerprint.

Cite this