基于NSST-IHS变换稀疏表示的SAR与可见光图像融合

Fusion of SAR and Visible Images Based on NSST-IHS and Sparse Representation

针对合成孔径雷达(SAR)与可见光图像成像原理不同,其融合图像常常存在感兴趣目标不突出及光谱失真的问题,提出了一种基于NSST-IHS变换稀疏表示的融合算法。对源图像进行IHS和NSST变换,在所得低频分量上采用基于结构相似性和亮度差异性的稀疏表示融合规则,高频分量上则采用基于改进的拉普拉斯能量和的融合规则,融合结果再通过NSST和IHS逆变换得到。实验以哨兵1号SAR图像与landsat-8可见光图像进行验证,并与传统的IHS、Wavelet、NSCT、IHS-Wavelet-SR和NSST-IHS算法进行比较。结果表明,该算法不论视觉还是评价指标都有了明显提高,空间结构信息和光谱信息得到有效的保持,有利于后续目标检测与识别工作。

In order to solve the problem that the interested aims are not prominent and spectral distortion caused by different imaging mechanism of synthetic aperture radar(SAR)and visible images,this paper proposes a fusion algorithm based on NSST-IHS and sparse representation.Firstly,source images are transformed by intensity-hue-saturation(IHS)and non-subsampled shearlet transform(NSST).Secondly,a fusion rule based on the structure similarity and luminance difference of the sparse representation is used in low-frequency components,while a fusion rule based on sum-modified-Laplacian is used in high-frequency components.Finally,the fusion results are obtained by inverse transformation of NSST and IHS.Experiments are carried out with Sentinel-1A SAR images and landsat-8 visible images,and compared with the traditional algorithms of IHS,Wavelet,NSCT,IHS-Wavelet-SR and NSST-IHS.The results show that the new algorithm has obvious improvement whether in visual or evaluation as well as to maintain the spatial structure information and spectral information,which is beneficial to target detection and recognition.