中文摘要：

高光谱遥感异常目标检测中,目标和背景光谱信息难以精确地界定,导致目标检测性能下降。针对经典RX检测算法存在虚警概率高、计算量大、过程繁琐等缺点,将Hausdorff度量引入高光谱异常检测,利用改进的Hausdorff距离（MHD）从光谱匹配程度的角度,进行了高光谱异常目标检测,最大程度地将异常目标和背景分离。采用模拟数据和真实高光谱数据进行大量实验,检测性能大幅提升,算法的计算效率提高了60%,证明了本文算法比RX算法、因果RX算法和KRX算法检测效果好,效率高等优势,算法的低复杂度特性为硬件实现提供了良好的算法支持。

英文摘要：

In anomaly target detection in hyperspectral imagery,it can be difficult to accurately distinguish between the spectral information of the targets and background,which leads to a decline in target detection performance. The results of the classic RX detection algorithm have a high false alarm probability,and the process is characterized by a large amount of calculation and complexity. To address these issues,we introduce the Hausdorff metric to hyperspectral anomaly target detection,prove the usefulness of its application,and make a number of improvements to suppress noise interference. In terms of the spectral matching,we separate the targets and background to a greater degree based on the improved Hausdorff distance. Experiments were performed using both synthetic and real hyperspectral data. Moreover,the results show improved detection performance and an increase in computational efficiency of nearly 60%. These experimental results prove that this algorithm has lower computational complexity and better performance than the traditional RX algorithm,casual RX algorithm,and KRX algorithm and can better support the implementation of hardware.