高光谱遥感图像实时目标检测对于实际应用具有十分重要的意义。针对目标和背景光谱均已知的高光谱遥感图像实时目标检测的问题,在正交子空间投影算法的基础上,利用矩阵分析理论,推导出逐波段处理的实时正交子空间投影算法,加强了原算法的实时处理能力。通过真实图像的实验结果表明,逐波段处理算法具有在实时性基础上提前结束检测过程从而减少目标检测过程所需时间的能力,并且具有数据存储空间和算法运算时间上的优越性。
In this paper,a progressive orthogonal subspace projection(POSP) algorithm is proposed by introducing matrix analysis theory under the condition of prior knowledge of target and background information.This method is based on the orthogonal subspace projection(OSP) algorithm and enhance the original method to process image in real-time.Experimental results of real data image demonstrate that the proposed POSP algorithm has the potential to finish the detection process earlier,and the advantage of time and space occupation.