Variation-based methods with different priors have been proven their ability in preserving edges for image restoration. Blind image decomposition is an inverse problem that is much harder to be solved than non-blind image decomposition from noisy images, commonly producing staircase effects in flat regions and smoothing fine structures. In this paper, we have tried to use spatially adaptive sparse representation (SASR) prior to restore a clean result from a blurred and noised image. In order to fastly and efficiently solve the SASR model, the alternating direction method of multipliers (ADMM) ...