Deconvolution of Image Sequences with a Learning FFT-based Approach

Iaroslav Koshelev, Andrey Somov, Stamatios Lefkimmiatis, Antonio Rodriguez-Sanchez

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

1 Citation (Scopus)

Abstract

In this work, we address the problem of fast and accurate non-blind motion deblurring of a sequence of frames. We propose an approach based on single image non-blind deconvolution methods by extending them from two spatial dimensions onto a 2D+ time space. The resulted algorithms utilize recent advances of deep learning and heavily rely on learning the spatial and time correlations of frames from data which allows to perform the deblurring in an end-to-end fashion without the manual tuning of parameters. Developed algorithms were trained and tested on blurred video sequences, and as a result an improvement in restoration quality by around 1dB in terms of PSNR was achieved, comparing to a single-frame deconvolution. The proposed algorithms perform video restoration in a close to real-time speed, having potential for real applications in both portable consumer electronics and machine vision systems.

Original languageEnglish
Title of host publication2020 IEEE 29th International Symposium on Industrial Electronics, ISIE 2020 - Proceedings
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages1381-1386
Number of pages6
ISBN (Electronic)9781728156354
DOIs
Publication statusPublished - Jun 2020
Externally publishedYes
Event29th IEEE International Symposium on Industrial Electronics, ISIE 2020 - Delft, Netherlands
Duration: 17 Jun 202019 Jun 2020

Publication series

NameIEEE International Symposium on Industrial Electronics
Volume2020-June

Conference

Conference29th IEEE International Symposium on Industrial Electronics, ISIE 2020
Country/TerritoryNetherlands
CityDelft
Period17/06/2019/06/20

Keywords

  • deblurring
  • image processing
  • video processing

Fingerprint

Dive into the research topics of 'Deconvolution of Image Sequences with a Learning FFT-based Approach'. Together they form a unique fingerprint.

Cite this