Abstract: In order to denoise the noisy image, a new image denoising method based on a multivariate shrinkage model in the non-subsampled shearlet (NSST) domain is proposed. A multivariate shrinkage model which takes into account the NSST coefficients'relationship was derived by using maximum a posterior estimator. Then, the noise-free coefficients were estimated by the multivariate shrinkage function. And the highest scale coefficients were estimated by hard-shrinkage method. Finally, the inverse NSST was applied to these estimated shearlet coefficients to obtain the denoised image. The proposed model utilizes the shift-invariance and the sparse representation of NSST, as well as the interscale and intrascale dependency correlation of NSST coefficients. Experiment results show that the proposed method can effectively remove the noise and avoid Gibbs phenomenon while preserving edges and texture details.