采用NSCT分解和主成分分析的高光谱异常检测

High spectral anomaly detection using NSCT decomposition and principal component analysis

针对复杂背景信息对高光谱异常检测干扰而导致检测效果降低的问题,提出了一种新的基于背景抑制的异常检测算法。对高光谱图像进行NSCT分解,得到各波段的高频信息和低频信息;对高频系数进行重构,得到只含有高频信息的高光谱数据;对原图像与低频信息作差,得到背景残差图像,利用主成分分析法对其进行背景抑制,得到背景信息充分抑制的高光谱数据,并与重构后的高频信息加权融合,得到有效抑制背景并凸显目标的高光谱数据,最后采用KRX算法对处理后的图像异常检测。结果表明,本方法检测性能优越,取得了较好的检测效果和较低的虚警率。

To solve the problem that complex background information leads to reduction of anomaly detection effectiveness,a novel anomaly detection algorithm is proposed to improve detection effect by suppressing background of imagery.Each band of hyperspectral imagery was decomposed by NSCT transformation to obtain the high frequency and low frequency information.To reconstruct the high frequency coefficients,it gets hyperspectral data just containing the high frequency information.The background residual error data which was the minus of the hyperspectral imagery and low frequency images was processed to get prominent target.Furthermore,hyperspectral data background suppressed is received by using principal component analysis（PCA）.Hence,ultima hyperspectral data is gained by fusing high frequency information with hyperspectral data background suppressed.At last,the images processed is detected by KRX algorithm.It turns out that the proposed algorithm outperformed the other algorithms and obtained a better effect of detection and a lower false alarm rate.