摘要
约束能量最小化(Constrained Energy Minimization,CEM)目标检测算法广泛应用于高光谱目标检测中。本文在分析CEM算法的推导过程后,发现图像像元的选择,可以改善自相关系数,因此提出一种改进的CEM目标检测算法。该方法首先对高光谱数据集进行光谱重排、一阶微分,增加目标与背景的差异性;计算目标光谱与数据集中光谱点的相似度,求取CEM算法的自相关矩阵时去除与目标相似度高的像元,减少自相关矩阵对目标的抑制。为进一步抑制背景,增加算法的普适性,加入对数算子。最后对合成高光谱数据和真实高光谱数据进行试验,结果表明,与传统算法相比,提出的算法可以对伪装目标进行有效识别,而且对小目标和大面积目标检测都具有适用性。
Target detection algorithm based onconstrained energy minimization (CEM) is widely used in hyperspectral targetdetection. An improved CEM target detection algorithm is proposed. In thismethod, spectral reordering and first order derivation of hyperspectral datasets are firstly used to increase the difference between target and background.The similarity between target spectral and spectral points of data set iscalculated, and the pixels with high similarity are removed when theautocorrelation matrix of CEM algorithm is obtained with the suppression oftarget by autocorrelation matrix reduced. To further suppress the background, alogarithmic operator is added. Finally, experiments on synthetic hyperspectraldata and real hyperspectral data show that the proposed algorithm can recognizecamouflaged targets effectively, and is applicable to small targets and largearea targets detection.
出处
《应用物理》
2019年第2期63-70,共8页
Applied Physics
