Image Restoration in G_p,q^β Space Shrinking

A new smooth space image restoration model based on DT model, Jiang model, and LHLLAV model is considered. The meanings of the parameters of the new model is discussed. The definition, nature, and norm of Besov space and Gp,qβ space are described. The model in Gp,qβ space is re-described according to the relationship between Gp,qβ space and Besov space. By constructing a surrogate functional that removes the influence of K*Ku, the model solution in the second generation curvelet transform domain is derived and an elegant curvelet shrinkage schemes is obtained. The experiment study shows that the new method has better restoration effect, faster convergence, and lower computational complexity than LHLLAV model.