基于非负矩阵分解的TerraSAR与SPOT5影像融合

Fusion Research on TerraSAR and SPOT5 Image Based on NMF

合成孔径雷达具有全天候、全天时的优势,能够提供高分辨的目标图像,而多光谱遥感影像能够提供丰富的光谱信息,将两种不同影像进行融合,综合两者的优势信息可以得到质量更高、信息更丰富的图像。提出了基于非负矩阵分解算法的融合算法对SAR影像和多光谱遥感影像进行融合。通过对比研究,提出的融合算法在目视判读以及客观评价指标上和传统融合算法相比,都有较明显的改善,尤其是在细节表现力和光谱保持度方面优势显著。

The synthetic aperture radar has the advantages of all-weather and time of day,and can provide high-resolution object images.The multispectral remote sensing images can provide rich spectral information.Fusing the different kinds of images,synthesizing the advantage information of each other into the fusion image and making the fusion image has higher quality and richer information.The fusion method based on the Non-negative matrix factorization is proposed and these methods are used to fuse SAR data and multispectral images.Through comparative study,the proposed fusion algorithm has obvious advantages in ability of details expressive and spectral conservation from object and subject evaluation indexs.