文章检索

  • 检索
  • 检索词:
  • 高级检索
您是今天第 1968位访问者
您是第 7237333 位访问者
引用本文:
【打印本页】   【下载PDF全文】   查看/发表评论  【EndNote】   【RefMan】   【BibTex】
←前一篇|后一篇→ 过刊浏览    高级检索
本文已被:浏览 698次   下载 494 本文二维码信息
码上扫一扫!
分享到: 微信 更多
Shearlet变换耦合细节强化因子的遥感图像融合算法
崔怡文1, 侯德林2
1.武汉铁路职业技术学院 经济管理学院,湖北 武汉430000;2.武汉纺织大学 管理学院,湖北 武汉 430073
摘要:
为克服当前较多遥感图像融合方法存在间断以及吉布斯现象,本文利用像素点间灰度以及梯度信息,设计了一种采用非下采样Shearlet变换(NSST)耦合细节强化因子的图像融合方法。将多光谱(MS)图像经过强度-色调-饱和度(IHS)变换,分离出强度成分。随后,借助 变换处理强度成分与全色(PAN)图像,获取对应的高频和低频系数。以强度成分对应的低频系数为依据,通过图像的空间频率特性计算加权系数,将PAN图像的低频系数植入到强度(I)成分对应的低频系数中,融合低频系数。采用像素点间灰度以及梯度信息,构造细节强化因子,融合高频系数。最后,采用IHS和NSST反变换重构这些融合系数,获取融合结果。实验结果显示:较当前融合技术,所提算法拥有更为理想的融合效果,具有更高的互信息值和更低的光谱偏差度值。
关键词:  遥感图像融合  空间频率  NSST变换  梯度信息  细节强化因子  IHS变换
DOI:10.11805/TKYDA20447
分类号:
基金项目:国家自然科学基金委青年基金资助项目(71203169);教育部人文社科基金一般资助项目(17YJA630108);湖北省人文社科重点研究基地-企业决策支持研究中心重点资助项目(DSS20180602)
A remote sensing image fusion algorithm based on Non Subsampled Shearlet Transform coupling detail enhancement factor
CUI Yiwen1, HOU Delin2
1.School of Economic Management,Wuhan Railway Vocational College of Technology,Wuhan Hubei 430000,China;2.School of Management,Wuhan Textile University,Wuhan Hubei 430073,China
Abstract:
In order to overcome the discontinuities and Gibbs phenomenon in many remote sensing image fusion methods, this paper designs an image with Non Sampling Shearlet Transform(NSST) coupling detail enhancement factor by using the gray level and gradient information between pixels fusion method. The intensity(I) component of Multi Spectral(MS) image is separated by Intensity-Hue-Saturation(IHS) transformation. The high and low frequency coefficients of I component and Panchromatic(PAN) image are extracted by NSST. Based on the low-frequency coefficient corresponding to component I, the filling coefficient is calculated by the spatial frequency characteristics of the image. The low-frequency coefficient corresponding to the image is filled into the low-frequency coefficient corresponding to component I, and the low-frequency coefficient is fused. The gray level and gradient information between pixels are utilized to construct detail enhancement factors to measure the differences between pixels and their neighbors, and then the high-frequency coefficients are fused. Based on the fusion coefficient, IHS and NSST inverse transforms are adopted to reconstruct the coefficients, and the fusion results are obtained. The experimental results show that the image fusion algorithm has higher mutual information value, lower spectral deviation value and better spectral and spatial characteristics than the current image fusion algorithm.
Key words:  remote sensing image fusion  spatial frequency  Non Subsampled Shearlet Transform  gradient information  detail enhancement factor  IHS transform

分享按钮