“…Image restoration is an effective technique for recovering incomplete or damaged images approximate to the ideal images. Up to now, various image restoration strategies have been proposed, which can be classified into several categories: (i) Time domain analysis methods, such as adaptive filter denoising (AFD) [ 6 , 7 ]; (ii) Frequency domain analysis methods, such as waveletβwavelet packet denoising (W-WPD) technique [ 8 , 9 ]; (iii) Data-driven approaches, such as partial differential equation (PDE) [ 10 , 11 , 12 , 13 ], and wavelet hidden Markov random field (WHMRF) [ 14 ]; (iv) Sparse representation (SR) techniques, such as redundant dictionary and non-convex penalty regularization [ 15 , 16 , 17 , 18 ], etc. Although the damaged images can be restored more precisely by the above methods, the drawbacks are also obvious.…”