Extreme value analysis of frame coefficients and implications for image denoising
Preprint
Abstract
Denoising by frame thresholding is one of the most basic and efficient methods for recovering a discrete signal or image from data that are corrupted by additive Gaussian white noise. The basic idea is to select a frame of analyzing elements that separates the data in few large coefficients due to the signal and many small coefficients mainly due to the noise. Removing all data coefficients being in magnitude below a certain threshold yields an approximation to the original signal. In order that a significant amount of the noise is removed and at the same time relevant information about the original image is kept, a precise understanding of the statistical properties of thresholding is important. For that purpose we compute, for the first time, the asymptotic distribution for the maximum of the absolute values of noise coefficients for a wide class of redundant frames. Based on our theoretical results we give a rationale for universal extreme value thresholding techniques yielding asymptotically sharp confidence regions and smoothness estimates corresponding to prescribed significance levels. The results cover many frames used in imaging applications, such as redundant wavelet systems, curvelet frames, or unions of bases. We show that `generically' a standard Gumbel law results as it is known from the case of orthonormal wavelet bases. However, for specific highly redundant frames other limiting laws may occur. We indeed verify that the translation invariant wavelet transform shows a different asymptotic behavior.Keywords
All Related Versions
This publication has 28 references indexed in Scilit:
- Locally adaptive image denoising by a statistical multiresolution criterionComputational Statistics & Data Analysis, 2012
- Concentration inequalities and confidence bands for needlet density estimators on compact homogeneous manifoldsProbability Theory and Related Fields, 2011
- Stein block thresholding for image denoisingApplied and Computational Harmonic Analysis, 2010
- Wavelets with composite dilations and their MRA propertiesApplied and Computational Harmonic Analysis, 2006
- New tight frames of curvelets and optimal representations of objects with piecewise C2 singularitiesCommunications on Pure and Applied Mathematics, 2003
- A normal comparison inequality and its applicationsProbability Theory and Related Fields, 2002
- Smoothness Estimates for Soft-Threshold Denoising via Translation-Invariant Wavelet TransformsApplied and Computational Harmonic Analysis, 2002
- Biorthogonal bases of compactly supported waveletsCommunications on Pure and Applied Mathematics, 1992
- Orthonormal bases of compactly supported waveletsCommunications on Pure and Applied Mathematics, 1988
- The One-Sided Barrier Problem for Gaussian NoiseBell System Technical Journal, 1962