用改进的Paik型Boltzmann机实现图像复原

张煜东; 吴乐南

doi:null

您当前的位置：

首页 >

文章列表页 >

用改进的Paik型Boltzmann机实现图像复原

信息科学 | 更新时间：2020-08-12

- 用改进的Paik型Boltzmann机实现图像复原
- Realization of image restoration by improved Paik's Boltzmann machine
- 光学精密工程 2009年17卷第7期页码：1737-1744
- 作者机构：
  
  东南大学信息科学与工程学院,江苏南京,210096
- 作者简介：
- 基金信息：
  
  国家863高技术研究发展计划资助项目(No.2008AA01Z227)();国家自然科学基金资助项目(No.60872075)();高等学校科技创新工程重大项目培育基金
- DOI：
  中图分类号： TP391.41
- 收稿日期：2008-07-12，
  
  修回日期：2008-08-15，
  
  网络出版日期：2009-07-25，
  
  纸质出版日期：2009-07-25
- 稿件说明：
移动端阅览
张煜东, 吴乐南. 用改进的Paik型Boltzmann机实现图像复原[J]. 光学精密工程, 2009,17(7): 1737-1744

ZHANG Yu-dong, WU Le-nan. Realization of image restoration by improved Paik's Boltzmann machine[J]. Editorial Office of Optics and Precision Engineering, 2009,17(7): 1737-1744
张煜东, 吴乐南. 用改进的Paik型Boltzmann机实现图像复原[J]. 光学精密工程, 2009,17(7): 1737-1744 DOI：

ZHANG Yu-dong, WU Le-nan. Realization of image restoration by improved Paik's Boltzmann machine[J]. Editorial Office of Optics and Precision Engineering, 2009,17(7): 1737-1744 DOI：

摘要

为解决传统的Boltzmann机方法不仅容易陷入局部最小点

而且收敛速度慢问题

对传统的Boltzmann机进行了改进。将Paik's算法与Boltzmann机结合

使串行模式推广到并行模式以加快收敛速度;使用亚单位步长增进技术增加计算精度;最后

为折中收敛速度与收敛精度这一对矛盾

采用了自适应步长策略。对算法的改进进行了理论验证、收敛性分析并对残差变化进行了讨论。实验表明

该方法能够收敛到全局最优

复原结果的峰值信噪比比改进的Boltzmann机法获得的峰值信噪比高0.5~0.8 dB

且收敛速度仅为该方法的1/3

证明了本文提出的改进的Paik型Boltzmann机对图像复原是有效的。

Abstract

An improved Paik's Boltzmann machine was presented to optimize the traditional Boltzmann method that suffers from not only being trapped into a local extreme but also a slow convergence. By proposed the method

the Paik's algorithm is integrated into the Boltzmann machine to restored images. Then

serial models are extended to parallel models to fasten the convergence speed and a subunit step is adopted to increase calculation precision. Finally a step adaptive method is used to trade off the contradiction of convergence precision and convergence velocity. Moreover

each improvement for the origonal algorithm is expatiated by theoritical validation

convergence analysis

and discussions of residual variations. Experiments demonstrate that the proposed method can be converged to the global minimum

and the Peak Signal Noise Ratio (PSNR) of the restored image is 0.5-0.8 dB higher than that of the improved Boltzmann method proposed by Bellouquid

while the consumed time is only 1/3 that of the latter. These results show that the proposed method is effective and valid for image restorations.

关键词

Keywords

references

朱俊, 文玉梅, 李平. 一种像场弯曲的图像复原方法[J]. 光学精密工程, 2003,11(6):621-626. ZHU J, WEN Y M, LI P. Restoration of field-curved images[J]. Opt. Precision Eng., 2003, 11(6): 621-626. (in Chinese)[2] 韩玉兵, 吴乐南. 基于状态连续变化的Hopfield神经网络的图像复原[J]. 信号处理, 2004,20(5):431-435. HAN Y B, WU L N. Image restoration using a modified hopfield neural network of continuous state change[J]. Signal Processing, 2004,20(5):431-435. (in Chinese)[3] 许廷发, 张敏, 顾海军. 改进的BP算法在多目标识别中的应用[J]. 光学精密工程, 2003,11(5):513-515. XU T F, ZHANG M, GU H J. Multi-target recognition with improved BP Algorithm . Opt. Precision Eng., 2003,11(5):513-515. (in Chinese)[4] 李鸣鸣, 龚振邦, 欧阳航空, 等. 实验数据RBF神经网络模型中噪声的处理方法[J]. 光学精密工程, 2005,13(1):227-231. LI M M, GONG ZH B, OUYANG H K, et al.. Strategies to the noise contained in experimental data in RBF neural network model[J]. Opt. Precision Eng., 2005,13(1):227-231. (in Chinese)[5] PAIK J M, GOODMAN J W. Neural networks for computation: number representation and programming complexity Appl[J]. Opt., 1986,25(18):3033-3046.[6] PAIK J K, KASAGGELOS A K. Image Restoration using a modified hopfield network[J]. IEEE Transactions on Image Processing, 1992,1(1):49-63.[7] BELLOUQUID A. From discrete Boltzmann equation to compressible linearized Eluer equations[J]. Electronic Journal of Differential Equations. 2004(104):1-18.[8] 汪源源, 孙志明, 蔡铮. 改进的奇异值分解法估计图像点扩散函数[J]. 光学精密工程, 2006,14(3):520-525. WANG Y Y, SUN ZH M, CAI ZH. Estimation of PSF of image system using modified SVD method[J]. Opt. Precision Eng., 2006, 14(3): 520-525. (in Chinese)[9] ZHANG Y D, WU L N. Improved image filter based on SPCNN[J]. Science in China F edition: Information Science, 2008, 51(12): 2115-2125.[10] ZHANG Y D, WU L N. Multi-resolution rigid image registration using bacterial multiple colony chemotaxis . Visual Information Engineering,2008:528-532.[11] ZHANG Y D, WU L N. A novel pattern recognition method via PCNN and Tsallis entropy[J]. Sensor, 2008,8(11):7518-7529.

浏览量

555

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

基于改进Paik型Boltzmann机的图像复原

基于神经网络的全斯托克斯光电探测器

基于圆卷积神经网络的粘连导电粒子检测

面向领域自适应的部分最优传输高光谱图像分类

基于热传导和卷积神经网络的磨床主轴热误差预测