采用背景抑制和自适应阈值分割的高光谱异常目标检测  

Hyperspectral anomaly detection based on background suppression and adaptive threshold segmentation

作　　者：赵春晖[1] 王佳[1] 王玉磊[1] 

机构地区：[1]哈尔滨工程大学信息与通信工程学院,黑龙江哈尔滨150001

出　　处：《哈尔滨工程大学学报》2016年第2期278-283,共6页Journal of Harbin Engineering University

基　　金：国家自然科学基金资助项目(61571145;61405041);黑龙江省自然科学基金重点资助项目(ZD201216);哈尔滨市优秀学科带头人基金资助项目(RC2013XK009003)

摘　　要：高光谱图像小目标异常检测存在着大面积背景异常的干扰,直接采用传统的阈值分割方法会产生较高的虚警。针对核RX异常检测算法存在较大面积背景干扰的现象,结合形态学滤波方法提取大面积背景杂波干扰并对其进行抑制,滤除背景干扰。然后采用自适应阈值方法对处理后的灰度值图像进行异常目标的分离。仿真实验结果表明,该方法较好地实现了对大面积背景干扰的抑制和对异常目标的保持,改善了现有的核RX算法用于高光谱异常检测的性能。Hyperspectral small target anomaly detection algorithms have the problem of a large area of background interference,and using the traditional threshold segmentation yields a high rate of false alarms. In order to better handle a large area of background interference in a Kernel RX detector,this paper developed a new morphology filter method to effectively extract and suppress large cluttered background areas. Further an adaptive threshold method was used to segment anomaly targets on grayscale images. Simulation experiments show that the proposed method provides very good anomaly detection with the advantage of large area cluttered background suppression. This dramatically improves the performance of the Kernel RX detector.

关 键 词：高光谱图像 背景抑制 形态学滤波 异常检测 自适应阈值 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]