To exact more directional information and important detail information from the images effectively, an image fusion algorithm for synthetic aperture radar(SAR) and grayscale visible light images based on the hidden Markov model(HMM) in the non-subsample Shearlet transform(NSST) domain is proposed. In NSST domain, the low frequency factors are fused by standard deviation. Meanwhile, the hidden Markov tree(HMT) model is built to train the high frequency factors. Then the energy gradient is used to select the trained high frequency factors. Thus, the low frequency and high frequency images are fused by inverse transformation of NSST to get the final image. Finally, the simulation results show that, compared with other multi-scale HMT models and traditional NSST fusion strategy, the proposed method can promote the fusion quality and enhance the information of the images, while reducing noise as well, and also show its effectiveness and feasibility.