Image Denoising Employing Two-Sided Gamma Random Vectors with Cycle-Spinning in Wavelet Domain
Pichid Kittisuwan, Sanparith Marukatat, Thitiporn Chanwimaluang, Widhyakorn Asdornwised · ECTI Transactions on Electrical Engineering Electronics and Communications · 2009
In this work, we present new Bayesian estimator for circularly-contoured Two-Sided Gamma random vector in additive white Gaussian noise (AWGN). This PDF is used in view of the fact that it is more peaked and the tails are heavier to be incorporated in the probabilistic modeling of the wavelet coefficients. One of the cruxes of the Bayesian image denoising methods is to estimate statistical parameters for a shrinkage function. We employ maximum a posterior (MAP) estimation to calculate local variances with Rayleigh density prior for local observed variances and Gaussian distribution for noisy wavelet coefficients. Several denoising methods (ProbShrink with redundant wavelet transform) using undecimated wavelet transforms provide good results. The undecimated wavelet transforms can also be viewed as applying an orthogonal wavelet transform to a set of shifted versions of the signal. This procedure wasfirst suggested by Coifman and Donoho where they termed it cycle-spinning method. We apply cycle spinning with orthogonal wavelet transforms in our work. The experimental results show that the proposed method yields good denoising results.