A Wavelet-GSM Approach to Demosaicking

We propose a wavelet-based Gaussian scale mixture (GSM) demosaicking method. The wavelet coefficients of the proposed method corresponding to the luminance and chrominance components are reconstructed using Bayesian minimum mean square error estimation. The proposed wavelet-GSM prior exploits the correlation of neighboring wavelets coefficients to improve upon a previously proposed posterior sparsity directed demosaicking method. As a result, our proposed demosaicking method suppresses the zippering artifacts more effectively than the state of the arts.