Signal-dependent noise characterization for mammographic images denoising
Marcello Salmeri, Arianna Mencattini, Giulia Rabottino, R. Lojacono · 2008
Abstrac t – The paper deals wi th the no ise character iza t ion under the assumption of a he teroscedast ic signal –dependent no ise model in the context o f medical imaging. In par t icular, in this kind of applica t ion, a sophis t icated noise var iance es t imat ion algori thm is appl ied us ing robust es t imators and nonlinear regress ions. A direct rela t ion be tween no ise var iance and pixe l intensi ty values i s obtained and used wi thin a mul t ireso lut ion denoising algor i thm, performed by Wavelet Thresholding (WT). We wi l l provide result s o f the no ise es t imation, by app lying the proposed method to mammographic images. I.