Skip to main navigation Skip to search Skip to main content

An innovative image enhancement method for edge preservation in wavelet domain

  • Harbin Institute of Technology

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

Abstract

In this paper, an innovative method is proposed for both noise repression and edge preservation. This method is distinct in that the structure tensor is applied in wavelet domain to detect the edge information. And both the shrinkage and the detection process are quantified and unified in the form of the matrix mask. In this way, the image denoising can be conducted with the detailed information maintained. To be specific, two matrix masks respectively in charge of suppressing the noise and preserving the edge pixels are combined to undertake this dual mission. And such an approach is proven to be beneficial to both the noise repression and the edge preservation.

Original languageEnglish
Title of host publication2015 IEEE International Instrumentation and Measurement Technology Conference - The "Measurable" of Tomorrow
Subtitle of host publicationProviding a Better Perspective on Complex Systems, I2MTC 2015 - Proceedings
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages52-56
Number of pages5
ISBN (Electronic)9781479961139
DOIs
StatePublished - 6 Jul 2015
Event2015 IEEE International Instrumentation and Measurement Technology Conference, I2MTC 2015 - Pisa, Italy
Duration: 11 May 201514 May 2015

Publication series

NameConference Record - IEEE Instrumentation and Measurement Technology Conference
Volume2015-July
ISSN (Print)1091-5281

Conference

Conference2015 IEEE International Instrumentation and Measurement Technology Conference, I2MTC 2015
Country/TerritoryItaly
CityPisa
Period11/05/1514/05/15

Keywords

  • Compound mask
  • Denoising
  • Edge preservation
  • Structure tensor
  • Wavelet threshold shrinkage

Fingerprint

Dive into the research topics of 'An innovative image enhancement method for edge preservation in wavelet domain'. Together they form a unique fingerprint.

Cite this