高分五号(GF-5)是我国首颗实现对大气和陆地综合观测的全谱段高光谱卫星，其搭载的可见短波红外高光谱相机(AHSI)提供的遥感数据拥有极高的光谱分辨率。然而，AHSI数据的空间分辨率为30 m，较低的空间分辨率限制了应用场景。为实现GF-5数据的降尺度，本文通过融合10 m哨兵二号(Sentinel-2)多光谱数据，生成10 m GF-5高光谱数据。在方法上，针对现有先进的信息损失引导的图像融合方法(information loss-guided image fusion, ILGIF)在高光谱图像降尺度中计算时间成本高的问题，本文提出了其快速版本(Fast ILGIF, FILGIF)。另一方面，在降尺度过程中，本文考虑并估计了30 m GF-5高光谱数据和10 m Sentinel-2数据之间的尺度转换点扩散函数(point spread function, PSF)，提高融合数据质量。实验结果验证了融合Sentinel-2数据用于GF-5高光谱数据降尺度的可行性。同时，结果表明：在获得与ILGIF相当精度的前提下，FILGIF大幅提高了运行效率；尺度转换PSF对降尺度过程有着重要影响，其准确的估计有助于获得更高精度的降尺度结果。

Gaofen-5 (GF-5) is the first hyperspectral satellite in China, which can acquire comprehensive observations of atmosphere and land surface. The advanced hyperspectral imager (AHSI) aboard GF-5 is a sensor that can acquire data covering visible，near-infrared (VNIR) and short-wave infrared (SWIR) wavelengths with very fine spectral resolution (i.e., 5 nm for VNIR and 10 nm for SWIR). The spatial resolution of GF-5 AHSI data (i.e., 30 m), however, is relatively coarse for extraction of land cover information in a number of cases, such as small-sized buildings and roads in urban areas. To produce GF-5 images with both fine spatial and spectral resolution, in this paper, GF-5 hyperspectral images were downscaled to 10 m by spatial-spectral image fusion with 10 m Sentinel-2 multispectral images. With respect to the method, considering the high computational cost in the advanced information loss-guided image fusion (ILGIF) method, a fast ILGIF (FILGIF) method was proposed, which is computationally more effective than the ILGIF method. On the other hand, the scale transformation optimal point spread function (PSF) between 10 m Sentinel-2 and 30 m GF-5 data was estimated adaptively for each GF-5 band to enhance the performance of downscaling. Experiments showed that the Sentinel-2 images are suitable for downscaling GF-5 hyperspectral images. The proposed FILGIF method can achieve comparable accuracy to ILGIF while the computational cost is decreased significantly. Moreover, when the PSF effect is considered, the downscaling results are more accurate.