NLDNet++: A physics based single image dehazing network

Iris Tal, Yael Bekerman, Avi Mor, Lior Knafo, Jonathan Alon, Shai Avidan

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

Abstract

Deep learning methods for image dehazing achieve impressive results. Yet, the task of collecting ground truth hazy/dehazed image pairs to train the network is cumbersome. We propose to use Non-Local Image Dehazing (NLD), an existing physics based technique, to provide the dehazed image required to training a network. Upon close inspection, we find that NLD suffers from several shortcomings and propose novel extensions to improve it. The new method, termed NLD++, consists of 1) denoising the input image as pre-processing step to avoid noise amplification, 2) introducing a constrained optimization that respects physical constraints. NLD++ produces superior results to NLD at the expense of increased computational cost. To offset that, we propose NLDNet++, a fully convolutional network that is trained on pairs of hazy images and images dehazed by NLD++. This eliminates the need of existing deep learning methods that require hazy/dehazed image pairs that are difficult to obtain. We evaluate the performance of NLDNet++ on standard data sets and find it to compare favorably with existing methods.

Original languageEnglish
Title of host publicationIEEE International Conference on Computational Photography, ICCP 2020
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781728152301
DOIs
StatePublished - Apr 2020
Event2020 IEEE International Conference on Computational Photography, ICCP 2020 - Saint Louis, United States
Duration: 24 Apr 202026 Apr 2020

Publication series

NameIEEE International Conference on Computational Photography, ICCP 2020

Conference

Conference2020 IEEE International Conference on Computational Photography, ICCP 2020
Country/TerritoryUnited States
CitySaint Louis
Period24/04/2026/04/20

Keywords

  • Computational Photography
  • Deep Neural Networks
  • Dehazing

Fingerprint

Dive into the research topics of 'NLDNet++: A physics based single image dehazing network'. Together they form a unique fingerprint.

Cite this