On the use of deep learning for single-pixel imaging

Saad Rizvi, Jie Cao, Qun Hao

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

We apply deep learning (DL) to counter three key problems which may occur in single-pixel imaging (SPI) namely noise, appearance of ringing or pixelated artifacts due to undersampling, and effects of projector lens aberration or defocusing. We employ a multi-scale mapping based deep convolutional neural network (DCNN) architecture to rectify undesirable effects in a 96×96 target reconstruction produced by environmental or system conditions, and optical anomalies. We train the proposed DCNN on augmented experimental data as well as simulation data to achieve robust experimental performance. Experimental results on real targets (2D and 3D) demonstrate the superior performance of the proposed method compared to conventional SPI.

Original languageEnglish
Title of host publicationHolography, Diffractive Optics, and Applications X
EditorsYunlong Sheng, Changhe Zhou, Liangcai Cao
PublisherSPIE
ISBN (Electronic)9781510639171
DOIs
Publication statusPublished - 2020
EventHolography, Diffractive Optics, and Applications X 2020 - Virtual, Online, China
Duration: 12 Oct 202016 Oct 2020

Publication series

NameProceedings of SPIE - The International Society for Optical Engineering
Volume11551
ISSN (Print)0277-786X
ISSN (Electronic)1996-756X

Conference

ConferenceHolography, Diffractive Optics, and Applications X 2020
Country/TerritoryChina
CityVirtual, Online
Period12/10/2016/10/20

Keywords

  • Deep learning
  • Denoising
  • Deringing
  • High-resolution imaging
  • Real-time imaging
  • Single-pixel imaging

Fingerprint

Dive into the research topics of 'On the use of deep learning for single-pixel imaging'. Together they form a unique fingerprint.

Cite this