BCGM based MAP denoising in wavelet domain
Xutao Li, Jiajia Ren, Yunkai Feng · 2010
This paper is concerned with image denoising approaches. The detail exponents of image transformed by Wavelet have been proved its significant heavy-tails nature. The alpha stable distribution has a generality to represent heavy-tailed and impulsive nature. However, the probability density function of such a statistics model has no closed expression. Based on a tractable approximation, BCGM model, maximum a posteriori (MAP) shrinkage is proposed for removing additive. Since the prior distributions are analytically known after BCGM is introduced, MAP approach becomes now more tractable than fore methods.