Multishrinkage: analytical form for a Bayesian wavelet estimator based on the multivariate Laplacian model.

We develop a novel multivariate Bayesian wavelet estimator of a simple analytical form that is computationally effective for the image denoising problem. The estimator is derived from the multivariate Laplacian model by using the maximum a posteriori rule. We find the multivariate estimator produces restoration results of high quality, both visually and in terms of peak signal-to-noise ratio.