针对高光谱图像中目标形状特征已知,背景和目标光谱特征未知时的多类小目标检测问题,给出一种检测算法.通过高光谱图像数据样本二次型的高阶矩控制点扩散函数,获取自适应结构化背景;然后,利用目标形状先验信息构造形状特征子空间,在高维光谱特征空间实现形状特征子空间匹配检测.理论分析和实验结果表明该检测器可同时有效检测具有不同形状特征的多类目标. Aiming at the multi-class small target detection problems in hyperspectral image with known target shape features and unknown background and target spectral features, a detection algorithm is proposed. The point spread function is controlled by the higher moments of the quadratic form of hyperspectral image data samples , And then get the adaptive structured background. Then, the shape feature subspace is constructed based on the prior information of the target shape, and the shape feature subspace matching detection is implemented in the high dimensional spectral feature space. The theoretical analysis and experimental results show that the detector can effectively detect Different types of features of the target.