Image compression using empirical mode decomposition

Kourosh Khoshelham, Zhilin Li

Research output: Chapter in book / Conference proceedingConference article published in proceeding or bookAcademic researchpeer-review


Image compression is an important issue in real time multimedia communication. Existing transform based image coding methods suffer from blocking, ringing and other artifacts in high compression ratios. This has motivated the authors to investigate the feasibility of employing a new transformation, Empirical Mode Decomposition (EMD), for image compression. The EMD transform is described and its 2D extension as applied on digital image is illustrated. The feasibility of employing EMD for image compression is discussed in this paper and two approaches to image compression based on EMD is presented. The time domain approach is based on adaptive sampling of IMF functions and reconstructing the image from these samples. The frequency domain approach applies Hilbert transform on IMF functions and forms the Hilbert spectrum using frequency quantization. Results of this research show that reconstruction of the original image from compressed image is possible in both time domain and frequency domain approaches with an acceptable accuracy however designing an adequate entropy coding scheme and modifying the 2D implementation of EMD transform play key rules for an EMD based image compression.
Original languageEnglish
Title of host publicationPicture Coding Symposium
Number of pages5
Publication statusPublished - 9 Dec 2003
EventPicture Coding Symposium - Saint Malo, France
Duration: 23 Apr 200325 Apr 2003


ConferencePicture Coding Symposium
CitySaint Malo

ASJC Scopus subject areas

  • Signal Processing
  • Computer Graphics and Computer-Aided Design


Dive into the research topics of 'Image compression using empirical mode decomposition'. Together they form a unique fingerprint.

Cite this