Wavelet denoising of multicomponent images, using a Gaussian Scale Mixture model

In this paper, denoising on multicomponent images is performed. The presented procedure is a spatial waveletbased denoising techniques, based on Bayesian leastsquares optimization procedures, using a prior model for the wavelet coefficients that account for the intercorrelations between the multicomponent bands.The applied prior model for the multicomponent signal is a Gaussian Scale Mixture (GSM) model. The method is compared to single-band wavelet denoising and to multiband denoising using a Gaussian prior. Experiments on a Landsat multispectral remote sensing image are conducted.

• Type: Conference
• Year: 2006
• Where: ICPR
• Authors: Paul Scheunders, Steve De Backer
• Keywords: Computer Vision, ICPR 2006, Multicomponent Signal, Prior Model, Spatial Waveletbased Denoising