Contextual Hidden Markov models for wavelet-domain signal processing

Matthew S. Crouse, Richard G. Baraniuk

Research output: Chapter in Book/Report/Conference proceedingConference contribution

47 Scopus citations

Abstract

Wavelet-domain Hidden Markov Models (HMMs) provide a powerful new approach for statistical modeling and processing of wavelet coefficients. In addition to characterizing the statistics of individual wavelet coefficients, HMMs capture some of the key interactions between wavelet coefficients. However, as HMMs model an increasing number of wavelet coefficient interactions, HMM-based signal processing becomes increasingly complicated. In this paper, we propose a new approach to HMMs based on the notion of context. By modeling wavelet coefficient inter-dependencies via contexts, we retain the approximation capabilities of HMMs, yet substantially reduce their complexity. To illustrate the power of this approach, we develop new algorithms for signal estimation and for efficient synthesis of non-Gaussian, long-range-dependent network traffic.

Original languageEnglish (US)
Title of host publicationConference Record of the Asilomar Conference on Signals, Systems and Computers
EditorsM.P. Farques, R.D. Hippenstiel
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages95-100
Number of pages6
Volume1
StatePublished - 1998
EventProceedings of the 1997 31st Asilomar Conference on Signals, Systems & Computers. Part 1 (of 2) - Pacific Grove, CA, USA
Duration: Nov 2 1997Nov 5 1997

Other

OtherProceedings of the 1997 31st Asilomar Conference on Signals, Systems & Computers. Part 1 (of 2)
CityPacific Grove, CA, USA
Period11/2/9711/5/97

ASJC Scopus subject areas

  • Hardware and Architecture
  • Signal Processing
  • Electrical and Electronic Engineering

Fingerprint Dive into the research topics of 'Contextual Hidden Markov models for wavelet-domain signal processing'. Together they form a unique fingerprint.

Cite this