This paper diseusses the restoration of multiple blurred and noised images. To construct a valid smoothness constraint, the anisotropic tensor diffusion is used,which can automatically select diffusivity according to the direction of edges. To get a valid weight function,the two aspects of restoration,noise and blur,are considered. In the processing of noise,we consider the normalized equation of the imaging, which can eliminate the influence of the difference on noise level. Especially in the processing of blur,we believe that the weight of each image is relevant to the ill-condition degree of each blur matrix. so a simulation step is set up to estimate the perturbation of each Blur matrix under certain perturbation. Compared with traditional model,the new model can get right weight according to the blur and noise,which can weigh the influence of each input image,and improve the processing results. Numerical results show that the images restored by the new method are better than those restored by traditional models,and the selected weights coincide with the restoration of single image restoration results.