Bayesian estimation of the multifractality parameter for images via a closed-form Whittle likelihood - Université de Lyon Access content directly
Conference Papers Year : 2015

Bayesian estimation of the multifractality parameter for images via a closed-form Whittle likelihood

Abstract

Texture analysis is central in many image processing problems. It can be conducted by studying the local regularity fluctuations of image amplitudes, and multifractal analysis provides a theoretical and practical framework for such a characterization. Yet, due to the non Gaussian nature and intricate dependence structure of multifractal models, accurate parameter estimation is challenging: standard estimators yield modest performance, and alternative (semi-)parametric estimators exhibit prohibitive computational cost for large images. This present contribution addresses these difficulties and proposes a Bayesian procedure for the estimation of the multifractality parameter c2 for images. It relies on a recently proposed semi-parametric model for the multivariate statistics of log-wavelet leaders and on a Whittle approximation that enables its numerical evaluation. The key result is a closed-form expression for the Whittle likelihood. Numerical simulations indicate the excellent performance of the method, significantly improving estimation performance over standard estimators and computational efficiency over previously proposed Bayesian estimators.
Fichier principal
Vignette du fichier
combrexelles_17037.pdf (209.35 Ko) Télécharger le fichier
Origin Files produced by the author(s)
Loading...

Dates and versions

hal-01511893 , version 1 (21-04-2017)

Identifiers

  • HAL Id : hal-01511893 , version 1
  • OATAO : 17037

Cite

Sébastien Combrexelle, Herwig Wendt, Jean-Yves Tourneret, Patrice Abry, Stephen Mclaughlin. Bayesian estimation of the multifractality parameter for images via a closed-form Whittle likelihood. 23rd European Signal Processing Conference (EUSIPCO 2015), Aug 2015, Nice, France. pp. 1003-1007. ⟨hal-01511893⟩
116 View
67 Download

Share

Gmail Mastodon Facebook X LinkedIn More