Fast object reconstruction in block-based compressive low-light-level imaging

Jun Ke*, Dong Sui, Ping Wei

*Corresponding author for this work

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

2 Citations (Scopus)

Abstract

In this paper, fast object reconstruction is studied for block-based compressive low-light-level imaging (BCLimaging). Instead of object pixels, linear combinations of object pixels, referred to as features, are magnified by an intensifier or MCP, and then collected as system measurements. Gaussian random projection and measurement SNR sorted Hadamard projection are studied. Then linear Wiener operator and nonlinear OMP method using parallel computing with GPU and serial process with CPU are used for reconstruction. With the help of GPU, more than 100X acceleration and less than 3ms precessing time is obtained for object reconstruction using Wiener operator.

Original languageEnglish
Title of host publicationInternational Symposium on Optoelectronic Technology and Application 2014
Subtitle of host publicationImage Processing and Pattern Recognition
EditorsGaurav Sharma, Fugen Zhou
PublisherSPIE
ISBN (Electronic)9781628413878
DOIs
Publication statusPublished - 2014
EventInternational Symposium on Optoelectronic Technology and Application 2014: Image Processing and Pattern Recognition, IPTA 2014 - Beijing, China
Duration: 13 May 201415 May 2014

Publication series

NameProceedings of SPIE - The International Society for Optical Engineering
Volume9301
ISSN (Print)0277-786X
ISSN (Electronic)1996-756X

Conference

ConferenceInternational Symposium on Optoelectronic Technology and Application 2014: Image Processing and Pattern Recognition, IPTA 2014
Country/TerritoryChina
CityBeijing
Period13/05/1415/05/14

Keywords

  • Block-wised compressive imaging
  • Compressive imaging
  • GPU
  • Low-light-level imaging
  • Parallel computing

Fingerprint

Dive into the research topics of 'Fast object reconstruction in block-based compressive low-light-level imaging'. Together they form a unique fingerprint.

Cite this