Abstract: Faced with the non-stationary property of real-world images and the complexity of computation problem in image recovery,an image restoration method using wavelet-domain classified hidden Markov tree（CHMT） model is proposed.According to Bayesian theory of image restoration algorithm,CHMT model is used as a priori information of image in the wavelet-domain,and regularization restriction is made to recover the image.The CHMT model has spatial adaptability,making the modeling more accurate.The restoration equation is solved with the simplified conjugate gradient method.Experimental results showed that this algorithm has a reasonable computational complexity,and the image recovery P_SNR is improved.