中文摘要：

为了进一步提高高光谱遥感图像小目标无监督检测方法的运算速度，并降低其虚警率，提出了一种基于改进蜂群优化投影寻踪与K最近邻的检测方法。首先，采用核主成分分析法对原始高光谱遥感图像进行降维；然后，提出以邻域像元联合定义峰度与偏度的方法，并将两者结合作为投影指标，再以改进后的蜂群算法作为寻优方法，使用投影寻踪从高光谱图像中逐层获取投影图像，再根据其直方图提取小目标；最后，利用线性判别分析进一步提取像元特征，并结合加权K最近邻方法对小目标的检测结果进行提纯。大量实验结果表明，与RX方法、独立分量分析法、混沌粒子群优化投影寻踪法相比，本文方法不但可以更精确地检测出高光谱遥感图像中的小目标，而且具有更快的运算速度。

英文摘要：

In order to further improve the operation speed and reduce the false Mann rate of the unsupervised detection method for small targets in hyperspectral remote sensing images, a detection method based on the projection pursuit （PP） optimized by improved artificial bee colony （ABC） optimization algorithm and K-nearest neighbor （KNN） is proposed in this paper. Firstly, the kernel principal component analysis （KPCA） method is adopted to perform the dimension reduction of the original hyperspectral remote sensing images. Then, the method jointly defining the kurtosis and skewness according to the neighborhood pixels is proposed, the combination of the kurtosis and skewness is taken as the projection index. The improved artificial bee colony algorithm is taken as the optimization algorithm. The projection pursuit is used to obtain the projection images layer by layer from the low dimensional hyperspectral remote sensing images, and the small targets are extracted according to the histogram of these projection images. Finally, the linear discriminant analysis （LDA） is used to extract the features of the pixels, and the weighted K-nearest neighbor method is used to purify the preliminary detection results of the small targets. A large number of experiment results show that compared with the RX method, independent eomponent analysis （ICA） method and the projection pursuit method based on chaotic particle swarm optimization （CPSO） , the proposed method not only can detect the small targets in hyperspeetral remote sensing image accurately, but also has faster operation speed.