摘要
由于图像复杂背景信息的干扰,一般检测算法的应用受到了限制,致使异常检测的虚警率较高,而基于支持向量数据描述(Support Vector Data Description,SVDD)的异常检测算法不需要对背景或者目标数据作任何分布假设,可将原始数据映射到高维特征空间进行异常检测。基于此,本文提出了一种基于非下采样Contourlet变换的异常检测SVDD算法。算法首先对高光谱数据进行NSCT(Nonsubsampled Contourlet Transform)分解,得到高频信息图像和低频信息图像;然后对低频信息作差,得到背景残差数据,抑制了背景信息;接着通过加权融合得到背景抑制后的高光谱图像,最后利用非线性SVDD将背景抑制后的高光谱图像映射到高维特征空间,完成异常目标的检测。通过仿真实验验证可知,所提出的算法与RX算法、KRX算法和未进行背景抑制的SVDD算法相比,具有较低的异常检测虚警率和优良的检测性能。
Due to the complex background information interference in the image, the application of general detection algorithm has been curbed and the false alarm rate of anomaly detection is higher. Anomaly detection algorithm based on Support Vector Data Description （Support Vector Data Description, SVDD） does not need to make any background or target Data distribution assumption, and the original data can be mapped to high-dimensional feature space for anomaly detection. On the basis of this, the paper puts forward a kind of anomaly detection based on the nonsubsampled Contourlet transform SVDD algorithm. First of all, the NSCT decomposition of hyperspectral data is carried out to obtain the high frequency and low frequency images. Then the low frequency information is used to get the background information, and the background information is suppressed. Then the high spectral image is mapped to a high dimensional feature space by the weighted fusion algorithm, and then the abnormal target is detected by SVDD. Through the simulation experiments, we can verify that the proposed algorithm has lower false alarm rate and better detection performance compared with RX algorithm, KRX algorithm and SVDD algorithm.
出处
《红外技术》
CSCD
北大核心
2016年第1期47-52,共6页
Infrared Technology
基金
全国教育信息技术研究课题项目(146241819)
