In order to solve the problem of unable to compress the noise and protect the details both well at the same time in the traditional image restoration algorithm based on Hopfield neural network

a new algorithm based on the modified Hopfield neural network of continuous state change and the wavelet domain hidden Markov tree (HMT) model is presented. The wavelet domain HMT model is utilized as the prior information about the statistical relationship between the image wavelet coefficients

and is introduced into the neural network model as the regularization term

the final restoration image is obtained using the energy convergence property of Hopfield neural network. Further more

a highly-parallel weight matrix determination algorithm is proposed

the weight values are computed batch by batch through the operator operation to the pattern images

in order to avoid the multiplication of large scale matrices. Experimental results demonstrate that

for either real image or artificial image

the visual quality of the restoration result is improved evidently

and the ISNR improves more than 0.3dB compared to the traditional algorithms. The objective of compressing the noise and protecting the details at the same time is reached.