复杂背景下减少红外小目标检测虚警率的算法  

Algorithm to reduce false alarm of IR small targets detection in complex background

作　　者：张晟翀[1] 蔡军[1] 孙玉铭 王立新[3] 

机构地区：[1]光电信息控制和安全技术重点实验室,河北三河065201 [2]第二炮兵驻锦州地区专用保障装备军事代表室,辽宁锦州121000 [3]北京理工大学光电学院,北京100081

出　　处：《激光与红外》2015年第2期221-224,共4页Laser & Infrared

摘　　要：为了减小复杂背景下红外小目标检测的虚警率,采用自适应平滑再中值滤波的方法抑制背景噪声,边缘增强阈值分割检测方法突出和提取目标,利用小目标是孤立奇异点且不能发生突变的特点,通过多特性联合判断方法实现虚警目标剔除算法。实验表明,本文的算法计算量小、实时性好、实现灵活,在复杂背景下能有效降低虚警又不漏检目标,并实际完成了FPGA和DSP的硬件实现。In order to reduce the false alarm of IR small targets detection under complicated background,adaptive smooth and median filtering algorithm were used to suppress background noise,and edge enhancement threshold segmentation method was used to highlight and extract the target. Because small target is the isolated singular point and can＇t mutate,false alarm deletion algorithm was realized through multiple features combined with judgment method.Experiment results show that this algorithm has the merits of simple calculation,good real-time and flexible implementation. Under the complex background it can effectively reduce the false alarm and don＇t leave out target. It has been realized with FPGA and DSP hardware structure.

关 键 词：复杂背景 红外小目标检测 虚警 奇异点 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]