扫描型红外预警系统中弱小目标的检测算法  

Efficient Method of Dim Small Target Detection for Scanning IR Surveillance System

作　　者：胡永生[1] 陈钱[1] 钱惟贤[1] 管志强[1] 

机构地区：[1]南京理工大学电子工程与光电技术学院,南京210094

出　　处：《光子学报》2008年第11期2350-2354,共5页Acta Photonica Sinica

基　　金：国防预研项目(40405050303)资助

摘　　要：提出一种基于面阵探测器的全方位红外预警系统的运动弱小目标检测算法.该算法主要基于灰度形态学滤波和目标概率分布函数匹配两种技术.灰度形态学的高帽变换可以有效地检测出红外图像中的特定目标,但没有考虑目标强度分布特征,检测结果的虚警率较高.本文采用了均值偏移算法常用的概率分布函数匹配技术,对检测出的小目标进行二次确认.试验结果表明,该方法能够有效地抑制背景杂波干扰,降低虚警率.A mixed method of dim small target detection for scanning Infrared surveillance system based on area focal plane array （FPA） was presented. The method combined gray level morphology filter and target probability distribution function matching technique. The top-hat filter of gray level morphology could detect small targets defined by structure element efficiently. But detecting result of top-hat filter had a high false alarm ratio for discarding the intensity distribution information of target. The probability distribution function matching method used in mean-shift was imported for re-validating the detected targets. Results show that the presented method can restrain background clutter and reduce the false alarm ratio.

关 键 词：信号检测 红外预警系统 弱小目标 形态滤波 概率分布匹配 

分 类 号：TN219[电子电信—物理电子学]