| 基于调和模型的连续Hopfield神经网络次优图像复原 |
| |
| 引用本文： | 姜明勇,陈向宁,喻夏琼.基于调和模型的连续Hopfield神经网络次优图像复原[J].测绘科学,2012,37(3):121-123. |
| |
| 作者姓名： | 姜明勇 陈向宁 喻夏琼 |
| |
| 作者单位： | 装备指挥技术学院,北京,101416 |
| |
| 摘 要： | 本文提出了一种基于调和模型的连续Hopfield神经网络正则化次优图像复原算法。针对传统正则化图像复原由于"模糊矩阵"和"高通滤波器"规模庞大而带来的复原过程中占用存储资源多的问题,提出了一种基于部分图像信息的次优复原算法,该算法能在性能下降不大的前提下,较好地解决传统复原资源消耗问题。同时算法采用由梯度算子生成的调和模型作为正则项,能在复原的同时保留图像边缘。仿真结果表明了算法的有效性。
|
| 关 键 词： | 图像复原 调和模型 Hopfield神经网络 正则化 次优算法 |
Continuous Hopfield Neural Network sub-optimal image restoration based on Harmonic model |
| |
| JIANG Ming-yong
,
CHEN Xiang-ning
,
YU Xia-qiong.Continuous Hopfield Neural Network sub-optimal image restoration based on Harmonic model[J].Science of Surveying and Mapping,2012,37(3):121-123. |
| |
| Authors: | JIANG Ming-yong CHEN Xiang-ning YU Xia-qiong |
| |
| Institution: | (Academy of Equipment Command and Technology,Beijing 101416,China) |
| |
| Abstract: | A continuous HNN sub-optimal image restoration based on Harmonic model was proposed in the paper.To solving the drawback of huge memory resources cost due to the big ’blur matrix’ and "high-pass filter",a sub-optimal restoration algorithm base on partial image information was proposed which could solve the memory cost problem in traditional restoration perfectly while its performance is tolerable.The algorithm used Harmonic model based on gradient operator as the regularization,which can reserve edges of image in restoration.Simulation results validated the efficiency of proposed algorithm. |
| |
| Keywords: | image restoration harmonic model Hopfield neural network regularization sub-optimal algorithm |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
