An algorithm based on anisotropic markov random field(MRF) model for 3-Dimensional microscopical image restoration was proposed in this paper. Wavelet is a research hotspot nowadays

it has prominent ability to describe the edges of images. To preserve the edges of restored images

the following three aspects were studied. Firstly

since the wavelet coefficients in different subbands have different orientational characteristics

a priori probability constraint based on anisotropic markov random field was introduced as a priori penalized term in the regularization process. Secondly

the regularization coefficients were adaptively updated according to the restored results during each iteration. So that the regularization parameter decreased step by step while the estimated images became close to the real ones. Thirdly

taking the effects of noise into consideration

an auxiliary step to alleviate it was introduced in the iteration process. Experiments showed that the new algorithm could preserve the detail information effectively. In the meanwhile

it could also reduce the interruption between layers of 3-Dimensional slices and the effects of noise both satisfactorily.