基于混合像元分解的薄云下光学遥感图像恢复方法

Spectral unmixing based image restoration for optical remote sensing images covered by thin cloud

云遮挡是限制光学遥感卫星对地观测能力的主要因素之一。针对薄云下光学遥感图像的图像恢复问题,首先将云对光谱观测值的影响在线性混合像元模型中显性地加以表达,提出了针对云特性的改进型线性混合像元模型;其次给出了两种基于混合像元分解的图像恢复方法、直接消除法与丰度调整法;最后分别在两种混合像元分解算法与两种图像恢复方法,即VCA(顶点成分分析)算法/MDC-NMF(最小距离限制的非负矩阵分解)算法与直接消除法/丰度调整法的不同组合下,分别利用模拟数据和真实数据,对相关方法的图像恢复能力和图像恢复效果进行了定性和定量分析。实验结果表明,MDC-NMF算法与丰度调整法的组合处理能够获得最佳的图像恢复效果。

The obscuring effect of clouds is one of the major factors which restrict the observation capabilities of optical remote sensing satellites. In this paper, aimed at image restoration for optical remote sensing images covered by thin clouds, three contributions have been made. Firstly, with characteristics of clouds, an enhanced linear mixing model has been proposed, in which the influence of clouds on the measured spectra has been presented explicitly. Secondly, two spectral unmixing based image restoration methods have been given, namely the direct elimination method （DEM） and the abundance adjusting method （AAM）. At last, with different combinations of the two spectral unmixing algorisms VCA/MDC-NMF and two image restoration methods DEM/AAM, the capabilities and results of relevant methods are analyzed qualitatively and quantitatively, using both simulated and real datasets. Experimental results show that, the combination of MDC-NMF and AAM can achieve the best image restoration result.