Edge preserving CFA demosaicking based on nonlinear weighted color differences

Ngai Li, Jim S.Jimmy Li, Sharmil Randhawa, Donald G. Bailey

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

2 Citations (Scopus)

Abstract

In order to preserve an edge in demosaicking, four directional high-order interpolants of the missing color value are determined in four different directions. A nonlinear weight is then assigned to each interpolant, so that the weights for the interpolants are more extreme on different sides of an edge in order to reduce the influence of the interpolants on the opposite side of an edge while increasing the influence on the same side of the edge. In this way, errors due to interpolation across an edge can be reduced. When there is no edge, more evenly weights are assigned to the interpolants to produce a more accurate output by smoothing noise. In this paper, we propose a method to produce nonlinear weights for interpolants for various edge transitions in order to minimize color artifacts. It has been shown that our proposed algorithm outperforms the other state-of-the-art algorithms both quantitatively and visually with the minimum amount of color artifacts.

Original languageEnglish
Title of host publicationProceedings of the 2016 IEEE Region 10 Conference, TENCON 2016
Subtitle of host publicationTENCON
PublisherInstitute of Electrical and Electronics Engineers
Pages1143-1146
Number of pages4
ISBN (Electronic)9781509025961
DOIs
Publication statusPublished - 8 Feb 2017
Event2016 IEEE Region 10 Conference (TENCON) -
Duration: 22 Nov 2016 → …

Publication series

NameIEEE Region 10 Annual International Conference, Proceedings/TENCON
ISSN (Print)2159-3442
ISSN (Electronic)2159-3450

Conference

Conference2016 IEEE Region 10 Conference (TENCON)
Period22/11/16 → …

Keywords

  • demosaicking
  • interpolants
  • interpolation across an edge
  • color artifacts

Fingerprint

Dive into the research topics of 'Edge preserving CFA demosaicking based on nonlinear weighted color differences'. Together they form a unique fingerprint.

Cite this