基于短时轨迹检测的红外弱小目标轨迹检测算法  

Trajectory Detection for Infrared Dim Point Target Based on Short Trajectory Detection

作　　者：赵菲[1] 卢焕章[1] 

机构地区：[1]国防科技大学ATR国防科技重点实验室,湖南长沙410073

出　　处：《电子学报》2014年第1期175-181,共7页Acta Electronica Sinica

基　　金：国家863高技术研究发展计划

摘　　要：为解决虚警率较高情况下红外弱小目标轨迹检测效率下降严重的问题,本文提出一种基于短时轨迹检测的红外弱小目标轨迹检测算法.算法首先在相邻帧间进行短时轨迹检测,然后以短时轨迹为基础建立新的轨迹结构并结合轨迹关联算法完成目标轨迹的检测.文章推导了算法的虚警率和检测概率的关系,证明了本文提出的算法相比单帧检测结合轨迹关联的算法具有更好的检测性能.通过对仿真和实际红外图像序列进行实验表明,本文算法能够在准确检测弱小目标的同时有效的减少无效轨迹的数量,从而提高了算法的执行效率.To improve the efficiency under high false alarms probability conditions,an infrared dim point target trajectory detection algorithm based on short trajectories detection is proposed in this paper. The short trajectories are detected in consecutive frames at first, and then the trajectory association which builds a new trajectory depends on the result of short trajectory is applied to detect the target trajectory. The relationship between false alarms probability and detection probability is deduced,and the detection performance is proved to be better than algorithm based on trajectory association combining with segmentation in single frame. The experimental results indicate that the amount of invalid trajectories can be decreased effectively while the target trajectory is being detected accurately, and the efficiency of target detection is improved.

关 键 词：弱小目标 轨迹检测 短时轨迹 轨迹关联 

分 类 号：TN911.73[电子电信—通信与信息系统]