利用脉冲耦合神经网络的高光谱多波段图像融合方法
DOI:
作者:
作者单位:

作者简介:

通讯作者:

中图分类号:

基金项目:

国家自然科学基金项目(面上项目,重点项目,重大项目)


HYPERSPECTRAL MULTIBAND IMAGE FUSION ALGORITHM BY USING PULSE COUPLED NEURAL NETWORKS
Author:
Affiliation:

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 文章评论
    摘要:

    针对高光谱图像波段众多、数据量大的特点,提出了一种基于脉冲耦合神经网络(Pulse Coupled Neural Networks, PCNN)模型的高光谱多波段图像融合方法.根据高光谱图像多输入的特点对原始PCNN模型进行了扩充,采用多通道PCNN模型来对输入图像进行非线性融合处理.通过分析传统变阈值衰减模型的特点及其不足,提出了修正的变阈值指数增加模型,以改善融合效果和降低PCNN运行的时间复杂度.利用记录点火时刻的赋时矩阵得到带有一定增强效果的融合结果图像.实验结果表明,该方法的融合效果要优于传统的主成分分析融合方法和小波变换融合方法.

    Abstract:

    Considering hyperspectral images with multiband and large data amount, a novel fusion algorithm of hyperpsectral multiband images based on pulse coupled neural networks (PCNN) was proposed. Firstly, the original PCNN model was expanded according to the multiinput characteristics of the hyperspectral images, and a multichannel PCNN model was applied to fuse the multiple input images in a nonlinear manner. Then, the modified variable threshold exponent increasing attenuation model was proposed to improve fusion effect and reduce time complexity by analyzing the characteristics and shortage of the traditional variable threshold attenuation model. Finally, the fusion image with a certain degree of enhancement effect was obtained by the time matrix which recorded the ignition time. The experiment results show that the proposed algorithm outperforms the traditional fusion algorithms based on principle component analysis (PCA) and wavelet transform.

    参考文献
    相似文献
    引证文献
引用本文

常威威,郭雷,付朝阳,刘坤.利用脉冲耦合神经网络的高光谱多波段图像融合方法[J].红外与毫米波学报,2010,29(3):205~210]. CHANG Wei-Wei, GUO Lei, FU Zhao-Yang, LIU Kun. HYPERSPECTRAL MULTIBAND IMAGE FUSION ALGORITHM BY USING PULSE COUPLED NEURAL NETWORKS[J]. J. Infrared Millim. Waves,2010,29(3):205~210.]

复制
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:2009-03-20
  • 最后修改日期:2009-05-24
  • 录用日期:2009-08-10
  • 在线发布日期: 2010-07-19
  • 出版日期:
文章二维码