“…In the earlier stage, the classic methods, such as spatial filtering methods [4, 5] and wavelet‐based methods [6–8], play an important role for MNR. More recently, owing to the successful of regularisation theory, the variational regularisation methods, the maximum a posteriori (MAP)‐based methods, and the sparse representation‐based methods have been widely used for many image processing tasks, such as inpainting, compressive sensing, restoration, and reconstruction [9–14]. Thus, the variational regularisation method also becomes a promising research way for MNR.…”