Bayesian detection with amplitude, scale, orientation and position uncertainty

Eric Clarkson, Harrison Barrett

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

4 Scopus citations

Abstract

The likelihood ratio is computed in the case where the background is known, the signal has a known shape, but the amplitude, scale, orientation and location of the signaI are unknown. The noise is assumed to be additive i.i.d. Caussian. The result is an expression involving the mutidimensional wavelet transform of the data with respect to the signal. The linear and quadratic approximations to the full likelihood statistic are examined in detail and are seen to involve operations commonly used in this context. Finally the full nonlinear ratio is formulated in germs of an amplification and modulation mechanism from the data to the prior probabilty distribution on the unknown parameters.

Original languageEnglish (US)
Title of host publicationInformation Processing in Medical Imaging - 15th International Conference, IPMI 1997, Proceedings
EditorsJames Duncan, Gene Gindi
PublisherSpringer-Verlag
Pages549-554
Number of pages6
ISBN (Print)3540630465, 9783540630463
DOIs
StatePublished - 1997
Event15th International Conference on Information Processing in Medical Imaging, IPMI 1997 - Poultney, United States
Duration: Jun 9 1997Jun 13 1997

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume1230
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other15th International Conference on Information Processing in Medical Imaging, IPMI 1997
Country/TerritoryUnited States
CityPoultney
Period6/9/976/13/97

ASJC Scopus subject areas

  • Theoretical Computer Science
  • General Computer Science

Fingerprint

Dive into the research topics of 'Bayesian detection with amplitude, scale, orientation and position uncertainty'. Together they form a unique fingerprint.

Cite this