Improved Hopfield Network for Image Restoration
Xiaohai He · Jisuanji gongcheng · 2007
Based on the analysis of Paik's Hopfield network for image restoration,this paper presents a modified full parallel self-feedback continuous Hopfield network model to restore degraded image.And by introducing a calculated parameterγrather thanγobtained by experiments or experience,the modified network converge can be made faster.The result of the model in restoring blurred image is compared with that of J.K. Paik's method,which shows the method makes the restoration process faster and output better restoration quality.In experiment 2,compared with the constant parameter,it testifies that the calculated parameterγcan give better convergence rate for the network.