Skip to main navigation Skip to search Skip to main content

A nonlinear quantitative measure for accessing the performances of hyperspectral image compression

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

Abstract

Image compression is one of the most important processing techniques in hyperspectral image applications. A nonlinear quantitative measure for objectively accessing the performances of image compression techniques is proposed based on the scheme that image compression methods with better performances will maintain more information when compressing images with the same compression ratio, and the information correlation between source image and compressed image is stronger. Using the Nonlinear Correlation Coefficient (NCC) to accurately describe the general relationship between the source image and the compressed image, the performances of different image compression methods can be directly compared. The proposed nonlinear correlation based measure can be considered as an objective enforcement and complementary to Peak Signal-to-Noise Ratio (PSNR) and Mean Square Error (MSE) measure, which are widely used and based on the differences between source image and compressed image.

Original languageEnglish
Title of host publicationIMTC'05 - Proceedings of the IEEE Instrumentation and Measurement Technology Conference
Pages2012-2015
Number of pages4
StatePublished - 2005
EventIMTC'05 - Proceedings of the IEEE Instrumentation and Measurement Technology Conference - Ottawa, ON, Canada
Duration: 16 May 200519 May 2005

Publication series

NameConference Record - IEEE Instrumentation and Measurement Technology Conference
Volume3
ISSN (Print)1091-5281

Conference

ConferenceIMTC'05 - Proceedings of the IEEE Instrumentation and Measurement Technology Conference
Country/TerritoryCanada
CityOttawa, ON
Period16/05/0519/05/05

Keywords

  • Image compression
  • Nonlinear correlation
  • Performances evaluation

Fingerprint

Dive into the research topics of 'A nonlinear quantitative measure for accessing the performances of hyperspectral image compression'. Together they form a unique fingerprint.

Cite this