针对薄雾天气造成的能见度低的问题，提出了一种利用高光谱图像混合像元分解技术去除雾的方法。建立了薄雾天气下的传感器成像物理模型，对含有雾端元的线性光谱混合数学模型进行解混。然后通过丰度反演方法得到雾端元的丰度后加以去除，将剩余地物端元的丰度调整后即获得去雾后的图像。该方法相比于基于单波段或全色图像的去云雾方法，物理意义更明确。从客观评价指标上也可以看出该方法的薄雾去除效果佳，去雾后图像细节更加丰富。

For the problem of low visibility caused by fog weather, a defogging method based on hyperspectral image unmixing technique is proposed. A physical model of the imaging sensor in fog weather is established, and the mathematical model of linear spectral mixture with fog endmember is unmixed. The fog endmember obtained from the abundance inversion is removed. The defogging image is achieved after the abundance adjustment of the remaining endmembers. Compared with the fog/cloud removal based on the single- band or full-color image, the physical meaning of this approach is clearer. From the objective evaluation, it can be seen that the defogging effect of the proposed method is good. The defogging image has richer details.