基于自反馈连续Hopfield网络复原运动模糊图象的研究  

Analyses of Restoring Motion-Blurred Images Based on Self-Feedback Continues Hopfield neural network

作　　者：王磊[1] 莫玉龙[2] 

机构地区：[1]上海交通大学电子信息学院,上海200030 [2]上海大学(嘉定校区)电子工程系,上海201800

出　　处：《电子器件》1997年第1期296-300,共5页Chinese Journal of Electron Devices

摘　　要：本文提出了一种能够有效复原运动模糊图象的改进Hopfield网络模型.理论分析表明采用该全并行目反馈连续网能使网络能量更精确地收敛到全局最小.该网络对匀速直线运动模糊图进行复原后得到的图象信噪比提高显著,目视效果很好.This paper provides a modified Hopfield neural network model that can resore motion-blurred images effectively. Theoretital analyses show that this full prallel self-feedback continues network ensures its energy convergence to the global minimum. The SNR of the images restored from linor motion-blurred images by this model improves obviously and the visual quality of the restored image is quite good.

关 键 词：神经网络 图象复原 图象处理 

分 类 号：TN919.8[电子电信—通信与信息系统] TP18[电子电信—信息与通信工程]