为解决高光谱探测条件下的假目标光谱设计问题,从特征空间中目标显著性的角度分析了假目标的设计原则。区别于其他目标探测,高光谱目标探测是高维数据的处理过程,目标显著性反映在数据经降维处理后形成的特征子空间中。据此提出的假目标设计原则如下。先在特征子空间内设计假目标,使其一方面凸显自身,另一方面则遮掩真目标。而后依据特征子空间与光谱通道的线性关系,通过子空间中假目标与其他目标的相对位置确定假目标的光谱特性。对相关图像的检测表明,这样设计的假目标能够有效降低真实目标的探测效果。相关结论对高光谱隐身技术有一定指导作用。 In order to solve the spectral design problem of fake targets in hyperspectral detection, the design principles of fake targets are analyzed from the perspective of the saliency of the target in the feature space. Differing from other target detection, hyperspectral target detection is the processing of high-dimensional data, and the significance of the target is reflected in the feature subspace formed after the data is reduced dimensionally. The proposed pseudo-target design principles are as follows. First, we design fake targets in the characteristic subspace, so as to make them highlight themselves on the one hand and hide the real targets on the other hand. Then based on the linear relationship between the characteristic subspace and the spectral channel, the spectral characteristics of the false target are determined by the relative positions of the false target and other targets in the subspace. The detection of related images shows that the false target designed in this way can effectively reduce the detection effect of the real target. Relevant conclusions have some guidance on hyperspectral stealth technology.