论文部分内容阅读
在高光谱图像异常检测中,背景存在异常像元会造成背景统计信息失真,这将导致检测结果具有较高的虚警率。针对此问题,本文提出了一种基于密度背景纯化的异常检测算法。首先计算背景中每个像元的密度;然后根据高光谱图像中背景密度远大于异常密度的特性,利用最大类间方差法将异常从背景中分离;最后,将纯化后的背景用于统计信息的估计,通过RX检测算法(Reed-Xiaoli detector,RXD)对高光谱图像进行检测。为验证算法的有效性,利用两组真实的高光谱数据进行仿真实验。实验结果表明与RXD比,所提算法在两组数据下的曲线下面积值分别提高了0.024 6和0.008 6。与当前的异常检测算法相比:所提算法有较好的接收机工作特性曲线。
In hyperspectral image anomaly detection, the presence of an abnormal pixel in the background may cause the background statistical information to be distorted, which will result in a higher false alarm rate of the detection result. In response to this problem, this paper presents an anomaly detection algorithm based on density background purification. Firstly, the density of each pixel in the background is calculated. Then, according to the characteristics that the background density is much larger than the anomalous density in the hyperspectral image, the maximum inter-class variance method is used to separate the anomalies from the background. Finally, the purified background is used for statistical information Hyperspectral images were detected by the RX detection algorithm (RXD). To verify the effectiveness of the algorithm, two sets of real hyperspectral data were used to simulate the experiment. Experimental results show that compared with RXD, the area under the curve of the proposed algorithm increases by 0.024 6 and 0.008 6 respectively. Compared with the current anomaly detection algorithm: the proposed algorithm has a good receiver operating characteristic curve.