摘要
本文在研究空间自适应正则化和Hopfield神经网络图像复原算法的基础上,综合两者的优点,提出了一种退化图像的盲复原方案,先对退化图像利用SAR算法进行先验辨识,再利用基于连续函数的全并行自反馈改进Hopfield神经网络算法进行精确的复原。仿真证明,此方案的盲复原结果图像的质量有很大改善。
The image restoration technology based on the Hopfield neural network makes use of neural (network's) outstanding merits in solving optimization problems. This paper proposes a new blind restoration method that firstly makes the prior blur identification that identifies PSF by using the space-adaptive regularization (SAR) method before using the parallel self-feedback neural network based on a continuous function. Experimental results show that this method produces good effect in blind restoration.
出处
《计算机工程与科学》
CSCD
2004年第6期31-33,54,共4页
Computer Engineering & Science
