三级训练BP神经网络遥感图像超分辨重建

朱福珍; 王晓飞; 丁群; 何宏昌

doi:10.3788/OPE.20152313.0723

您当前的位置：

首页 >

文章列表页 >

三级训练BP神经网络遥感图像超分辨重建

更新时间：2020-08-13

- 三级训练BP神经网络遥感图像超分辨重建
- Super-resolution reconstruction of remote images based on three level training BP neural network
- 光学精密工程 2015年23卷第10z期页码：722-729
- 作者机构：
  
  1. 黑龙江大学电子工程学院, 黑龙江哈尔滨 150080
  2. 黑龙江大学电子工程学院电子科学与技术博士后流动站,黑龙江哈尔滨,150080
- 作者简介：
- 基金信息：
  
  黑龙江省教育厅科学技术研究项目(No.12531490)();黑龙江大学青年科学基金资助项目(No.QL201209)();黑龙江大学电子科学与技术博士后流动站资助项目
- DOI：10.3788/OPE.20152313.0723
  中图分类号： TP183;TP391.4
- 收稿日期：2015-06-12，
  
  修回日期：2015-06-20，
  
  纸质出版日期：2015-11-14
- 稿件说明：
移动端阅览
朱福珍, 王晓飞, 丁群等. 三级训练BP神经网络遥感图像超分辨重建[J]. 光学精密工程, 2015,23(10z): 722-729

ZHU Fu-zhen, WANG Xiao-fei, DING Qun etc. Super-resolution reconstruction of remote images based on three level training BP neural network[J]. Editorial Office of Optics and Precision Engineering, 2015,23(10z): 722-729
朱福珍, 王晓飞, 丁群等. 三级训练BP神经网络遥感图像超分辨重建[J]. 光学精密工程, 2015,23(10z): 722-729 DOI： 10.3788/OPE.20152313.0723.

ZHU Fu-zhen, WANG Xiao-fei, DING Qun etc. Super-resolution reconstruction of remote images based on three level training BP neural network[J]. Editorial Office of Optics and Precision Engineering, 2015,23(10z): 722-729 DOI： 10.3788/OPE.20152313.0723.

摘要

为了进一步提高遥感图像超分辨效果

降低超分辨重建时间

建立了一种三级训练BP神经网络(BP Neural Network

BPNN)超分辨重建方法

重点研究了网络训练样本的图像获取、输入输出样本的图像筛选、网络结构及训练算法的设计等。建立遥感图像退化模型

采用亚像素位移欠采样的方法获取网络训练样本;然后以方差比较法筛选出各级网络训练的输入/输出样本图像;最后

采用3组超分辨映射模式的遥感图像分别作为同一结构BPNN的输入/输出训练样本图像

连续进行3个周期的训练和学习

从而使图像尺寸映射模式和空间分辨率依次提高3次。仿真和泛化实验表明

三级训练BPNN较其他常见超分辨算法的峰值信噪比最高提高了6 dB左右

超分辨重建图像细节更丰富

重建时间大大降低

更适合遥感图像的实际应用。

Abstract

To further improve the effect of the super-resolution reconstruction(SRR) of remote sensing images and reduce its time-consuming

a three level training BP Neural Network(BPNN) was established. The research was focused on the acquisition of training samples

selections of input-output training samples

and the design of BPNN structure and training algorithm. A remote sensing image degradation model was set up. Then

training sample images were got by undersampled and subpixel-shifted method. The input-output training sample images were selected by variance comparison. Finally

three groups remote sensing images with different super-resolution mapping modes were used as the input-output training samples for the same BPNN. The net was continuously trained and learned three cycles

and image size and spatial resolution were improved three times. Experimental results indicate that the three level training BPNN for the SRR of remote sensing image can obtain better SRR effect and higher spatial resolution in the process of fitting remote sensing image SRR mapping

and the Peak Signal to Noise Ratio(PSNR) is improved about 6 dB than that of other ordinary super-resolution algorithm. For preserving more image details and reducing reconstructing time

it is more suitable for practical applications of remote sensing images.

关键词

Keywords

references

邓承志,田伟,汪胜前,等. 近似稀疏正则化的红外图像超分辨率重建[J]. 光学精密工程,2014,22(6):1648-1654. DENG CH ZH, TIAN W, WANG SH Q, et al.. Super-resolution reconstruction of approximate sparsity regularized infrared images[J]. Opt. Precision Eng., 2014,22(6):1648-1654.(in Chinese)

LI X L, HU Y T, GAO X B. A multi-frame image super-resolution method[J]. Signal Processing, 2010, 90(2):405-414.

TSAI R Y, HUANG T S. Multipleframe image restoration and registration[J]. Advances in Computer Vision and Image Processing, 1984, 1:317-339.

PATANAVIJIT V. A Leclerc Bayesian approach for video reconstruction based on a robust iterative SRR and a general observation model[C]. 2010 International Symposium on Communications and Information Technologies, 2010, 856-861.

龚卫国,潘飞宇,李进明. 用双层重建法实现单幅图像的超分辨率重建[J]. 光学精密工程,2014,22(3):720-729. GONG W G, PAN F Y, LI J M. Single-image super-resolution reconstruction via double layer reconstructing[J]. Opt. Precision Eng., 2014,22(3):720-729.(in Chinese)

PROTTER M, ELAD M, TAKEDA H, et al.. Generalizing the nonlocal-means to super-resolution reconstruction[J]. IEEE Transactions on Image Processing, 2009, 18(1):36-51.

ZHANG L P, ZHANG H Y, SHEN H F, et al.. A super-resolution reconstruction algorithm for surveillance images[J]. Signal Processing, 2010,90(3):848-859.

SONG H H, HUANG B, LIU Q S, et al.. Improving the spatial resolution of landsat TM/ETM+ through fusion with SPOT5 images via learning-based super-resolution[J]. IEEE Transactions on Geoscience and Remote Sensing, 2015, 53(3):1195-1204.[9] GUO K, YANG X, LIN W, et al.. Learning-based super-resolution method with a combining of both global and local constraints[J]. The Institution of Engineering and Technology Image Processing, 2012, 6(4):337-344.

YU J F, GAO X B, TAO D C, et al.. A Unified learning framework for single image super-resolution[J]. IEEE Transactions on Neural Network and Learning Systems, 2014, 25(4):780-792.

XIE Q L, CHEN H, CAO H M. Improved example-based single-image super-resolution[J]. Third International Congress on Image and Signal Processing, 2010, 3:1204-1207.

DANG C, AGHAGOLZADEH M, RADHA H. Image super-resolution via local self-learning manifold approximation[J]. IEEE Signal Processing Letters, 2014, 21(10):1245-1249.

YANG M C, WANG Y C. Self-learning approach to single image super-resolution[J]. IEEE Transactions on Multimedia, 2013, 15(3):498-508.

HE C, LIU L Z, XU L Y, et al.. Learning based compressed sensing for SAR image super-resolution[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2012, 5(4):1272-1281.

SU Y F, FOODY G M, MUAD A M, et al.. Combining Hopfield neural network and contouring methods to enhance super-resolution mapping[J]. IEEE Journal ofSelected Topics in Applied Earth Observations and Remote Sensing,2012,5(5SI):1403-1417.

LI X D, YUN D, FENG L, et al.. Superresolution mapping of remotely sensed image based on Hopfield neural network with anisotropic spatial dependence model[J]. IEEE Geoscience and Remote Sensing Letter, 2014,11(7):1265-1269.

YU J F, GAO X B, TAO D C, et al.. A unified learning framework for single image super-resolution[J]. IEEE Transactions on Neural Networks and Learning Systems, 2014, 25(4):780-792.

朱福珍,李金宗,朱兵,等. 基于径向基函数神经网络的超分辨率图像重建[J]. 光学精密工程,2010,18(6):1444-1451. ZHU F ZH, LI J Z, ZHU B, et al.. Super-resolution image reconstruction based on RBF neural network[J]. Opt. Precision Eng., 2010,18(6):1444-1451.(in Chinese)

朱福珍,朱兵,李培华,等. 图像超分辨BP神经网络的改进[J]. 系统工程与电子技术,2014,36(6):1215-1219. ZHU F ZH, ZHU B, LI P H, et al.. Improvement of BP neural network for image super-resolution[J]. System Engineering and Electronics, 2014,36(6):1215-1219.

朱福珍,李金宗,李冬冬. 基于BP神经网络的超分辨率图像重建[J]. 系统工程与电子技术,2009,31(7):1746-1749. ZHU F ZH, LI J Z, LI D D. Reconstruction of super-resolution image based on BP neural network[J]. System Engineering and Electronics, 2009,31(7):1746-1749.

浏览量

110

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

融合边缘增强与非局部模块的遥感图像超分辨率重建生成对抗网络

基于径向基函数神经网络的超分辨率图像重建

面向单幅遥感图像超分辨率的空间自适应及频率融合网络

基于多尺度自适应注意力的图像超分辨率网络

基于注意力和宽激活密集残差网络的图像超分辨率重建