稀疏表示支撑集的遥感影像融合

针对常用稀疏表示系数融合规则不能完全保留两幅影像的有用信息,该文通过分析稀疏表示系数支撑集空间分布关系,提出一种新的稀疏表示系数融合规则。首先对多光谱影像进行广义IHS变换,将得到的亮度分量与全色影像分别进行稀疏表示;然后分析亮度分量与全色影像稀疏表示解的支撑集,对支撑集中交集部分和差集部分所对应的稀疏表示系数分别利用求和方式与L1范数最大方式进行融合;最后采用加权细节插入方式,将融合后的亮度分量细节信息插入到多光谱影像中,得到高分辨多光谱影像。实验结果表明,该方法能较好地提高空间分辨率并减少光谱损失;在主观视觉和客观评价上,比常用的融合规则方法有所提高。

Remote sensing image fusion based on sparse representation support set

For the common sparse representation coefficient fusion rule cannot completely retain the useful information of two images,this paper proposes a new sparse representation coefficient fusion rule by analyzing the spatial distribution of sparse representation coefficient support set space.Firstly,the generalized IHS transform is applied to multispectral images,and the brightness components and panchromatic images are respectively sparse representation.Secondly,the support set of the brightness component and panchromatic image sparse representation solution is analyzed,and the sparse representation coefficients corresponding to the support set intersection and the difference the sections of the literary collections are respectively used to fuse with the L1 norm.Finally,the weighted detail insertion method is used to insert the fused luminance component details into multispectral images,and the high resolution multispectral image is obtained.The experimental results show that the method can improve the spatial resolution and reduce spectral loss.In the subjective visual and objective evaluation,more than the commonly used fusion rules methods.