结构稀疏表示与细节注入的遥感图像融合方法

Remote sensing image fusion based on structured sparse representation and detail injection

针对目前遥感图像融合不能为有效兼顾多光谱图像的光谱信息和全色图像的空间信息导致融合图像质量不佳的问题,提出基于结构稀疏表示与细节注入的遥感图像融合方法。该方法首先采用双重稀疏表示分别获取全色遥感图像对应的高、低频结构化字典;其次,通过该字典对全色遥感图像进行稀疏表示,得到其高频细节信息;最后通过细节注入模型(ARSIS)将全色图像中的细节信息融入低分空间辨率的多光谱遥感图像。实验结果表明,与传统基于稀疏表示以及细节注入模型的融合方法相比,本文方法能更好的兼顾融合图像的空间与光谱分辨率,在视觉效果与指标评价上均具有更好的融合效果。

Aiming at the problem that the remote sensing image fusion can not effectively combine the spectral information of the multi-spectral image and the spatial information of the panchromatic image, a remote sensing image fusion method based on structured sparse representation and detail injection is proposed. Firstly, this method uses the double sparse representation to obtain the high and low frequency structured dictionaries corresponding to the panchromatic remote sensing image respectively. Secondly, the sparse representation of the panchromatic remote sensing image is obtained by this dictionary, Then, the high frequency detail information is obtained by the sparse representation of the panchromatic image under the learned dictionary. Finally, the detail information of the panchromatic image inject into the low spatial resolution multi-spectral remote sensing image under ARSIS model. The experimental results show that compared with the traditional sparse representation and the detail injection methods, this method can obtain better fusion results on both the spatial and spectral information, and it has better fusion effect in visual and evaluation indexes.