【摘要】 本文提出了一种天基高光谱图像仿真算法。首先利用传感器参数将输入多/高光谱图像转换为地物辐射亮度图像,利用大气校正方法和辐射传输模型将其转换为地物反射率图像。接着进行端元提取,利用端元集中的光谱信息和原始光谱库匹配构建解混光谱库。然后利用解混光谱库进行混合像元分解,计算场景的丰度矩阵和地物光谱信息,将地物光谱信息在原始光谱库中匹配得到地物类型及全波段反射率。最后利用丰度矩阵、地物全波段反射率计算感兴趣波段的地物反射率图像,通过大气模型、传感器模型生成感兴趣波段天基高光谱仿真图像。实验结果表明,本文算法生成的仿真图像与原始图像具有较高的相似性。更多还原

【Abstract】 A spaced-based hyperspectral image simulation algorithm is proposed in this paper. Firstly,according to sensor parameters,multispectral/hyperspectral image is transformed into ground-object radiance image which is then transformed into ground-object reflectance image using atmospheric correction method and radiative transfer model.Secondly,endmember set is constructed by means of endmember extraction so that unmixing spectral library is constructed through match of spectral information in endmember set and original spectral library. Thirdly,abundance matrix and ground-object spectral information are obtained by means of unmixing using unmixing spectral library so that groundobject types and full band reflectance are gained through match of ground-object spectral information and original spectral library. Finally,band-of-interest ground-object reflectance image is figured out via abundance matrix and full band reflectance. So band-of-interest hyperspectral simulation image is generated through atmospheric model and sensor model. Experimental result shows that our simulation image is similar with the original image.更多还原