We provide an overview of the modeling and removal of correlated noise (including crosstalk and fixed pattern) in the context of transform-domain filters, and introduce a method for both the exact computation and effective approximations of the noise spectrum in nonlocal collaborative transforms. Extensive experiments support the proposed method over earlier crude approximations used by filters such as BM3D. We conclude with a discussion on how these models interplay with signal-dependent observations and variance stabilization.

This presentation is part of Minisymposium “MS14 - Denoising in Photography and Video (2 parts)”
organized by: Stacey Levine (Duquesne University) , Marcelo Bertalmío (University Pompeu Fabra) .

Authors:

Ymir Mäkinen (Tampere University of Technology)

Lucio Azzari (Tampere University of Technology)

Alessandro Foi (Tampere University of Technology)

Keywords:

image restoration, noise modeling