| 摘要: |
| 针对现有SAR与可见光遥感影像融合算法的计算复杂度较高，细节信息保留较差等问题，提出了一种NSST-IHS结合自适应PCNN改进的融合算法。该方法先利用IHS 变换提取可见光图像的亮度分量I，并将得到的亮度分量I与SAR图像分别进行NSST变换；然后，针对低频子带分量采用方向信息即空间频率和平均梯度自适应调整PCNN的外部刺激与链接强度；高频子带分量上运用改进的拉普拉斯能量和(SML)的融合规则；最后，运用逆 NSST变换和逆IHS变换得到最终融合图像。实验表明，本文算法所得融合图像比传统算法在视觉效果方面提升明显，光谱信息及线性结构特征得到更多保留、各类评价指标上比传统算法要更好。 |
| 关键词: 遥感影像 图像融合 非下采样剪切波变换 脉冲耦合神经网络 |
| DOI：DOI:10.3969/j.issn.1672-2337.2020.06.011 |
| 分类号:TN958；TP391 |
| 基金项目: |
|
| An Improved Fusion of SAR and Visible Images |
|
ZHANG Rui, DONG Zhangyu
|
|
1. School of Computer and Information, Hefei University of Technology, Hefei 230601, China;2. Key Laboratory of Industrial Safety and Emergency Technology, Hefei 230601, China
|
| Abstract: |
| Aiming at the problems of high computational complexity and poor retention of detailed information of the existing SAR and visible light remote sensing image fusion algorithms, an improved fusion algorithm combining NSST-IHS and adaptive PCNN is proposed. This method first uses the IHS transform to extract the luminance component I of the visible light image, and performs the NSST transform on the obtained luminance component I and the SAR image respectively. Then, it uses the direction information, that is, the spatial frequency and average gradient, to adaptively adjust the external stimulus and link strength of the PCNN for low frequency sub-band components. The sum-modified-Laplacian (SML) is applied to the high-frequency sub-band components. Finally, the inverse NSST transform and the inverse IHS transform are used to obtain the final fusion image. The experiments show that the fusion image obtained by the algorithm in this paper improves visual effects significantly compared with the traditional algorithms, the spectral information and linear structure features are more retained, and various evaluation indicators are better than the traditional algorithms. |
| Key words: remote sensing image image fusion non-subsampled shearlet transform(NSST) pulse coupled neural network (PCNN) |
