改进的多光谱遥感影像超分辨率重构算法

Improved Super-resolution Reconstruction Algorithm for Multi-spectral Remote Sensing Image

针对现有遥感影像重构算法数据资源有限、配准精度低等问题,结合遥感影像的光谱特征,提出一种改进的多光谱遥感影像超分辨率重构算法。提取场景结构特征作为重构的正则化约束条件,保持重构结果中的高频信息。利用波段间的交叉相关,获得场景的结构特征信息。通过迭代反投影算法对单波段影像进行重构,将其合成为全色高分辨率遥感影像。仿真实验结果表明,该算法的重构效果较优。

Aiming at the problem of limited data resourse and lower precision of registration of the present super-resolution reconstruct method,combined with the abundant spectral information of remote sensing image,an improved super-resolution reconstruction method is proposed.It uses cross correlation between two bands,which is useful for the preserving of high frequency component.The correlation of bands can obtain the feature information,iterative back projection is used to reconstruct every single band image,and these single high resolution images are synthesized to a panchromatic remote sensing image.Experimental results show that the reconstruction effectiveness of the algorithm is good.