New smooth space G for image denoising

The noise image is decomposed into the unknown true image u and the noise v by a new image de-noising method based on image decomposition. The common decomposition models are all dense and can only be transformed into high-order partial differential equations to solve, which are heavy computations. DT model and the Jiang model are sparse image decomposition models. A new image de-noising model based on the above two models is put forward in this paper. This new model is sparse which is defined in the new smooth space G^β_{p, q} (smooth Besov space embedding) variational functional. The variational functional can be solved by second-generation Curvelet contraction threshold. Experimental results show that de-noising effect is better of the proposed model than these common models.