Astrophysical map reconstruction from convolutional mixtures

We propose an astrophysical map reconstruction method for multi-channel blurred and noisy observations. We define the problem under Bayesian framework. We use the t-distribution to model the image gradients as a prior and resort the Monte Carlo simulation to estimate the maps and error both in the pixel and frequency domain. We test our method in five different sky patch located at varying positions from galactic plane to high altitude. We give the estimated maps along with the power spectrums and the numerical performance measures.