为解决飞机平台强干扰环境下红外多光谱扫描仪的图像混合噪声问题，提出了一种基于低秩矩阵近似与常量统计的光谱图像去噪方法。从仪器的噪声产生机制出发将图像噪声区分为谱间差异噪声与谱间共有噪声，并根据噪声在光谱数据中的特点分别采用低秩矩阵近似与常量统计的方法进行去除。经仿真分析去噪后的光谱图像具有较高的峰值信噪比与去噪鲁棒性，并通过机载遥感试验数据证明在大量噪声的干扰下对于响应偏低的边缘谱段仍能有效地恢复出图像信息。该方法在其它类似光谱图像去噪方面具有一定的推广应用价值。

A denoising method of an airborne infrared multispectral scanner was proposed to solve the problem of image-mixed noise in strong interference conditions. Starting from the noise generation mechanism, the mixed noise was divided into two types: spectral common noise and spectral differentiated noise. According to the feature of these noises, low-rank matrix approximation and constant statistics were applied in this method. Our simulation proved the method had a high peak-signal-to-noise ratio and denoising robustness. Experiments using real data were conducted in low-spectral responsivity to verify the denoising performance of the proposed method. This method has certain advantages and application values in other similar spectral denoising fields.