基于改进递归最大值滤波的红外点目标检测  

Point-Targets Detection Algorithim Based on Modified Recursive Max Filter

作　　者：江涛[1] 王永仲[1] 

机构地区：[1]军械工程学院光学与电子工程系,河北石家庄050003

出　　处：《红外技术》2004年第3期41-44,共4页Infrared Technology

基　　金：国家863计划资助项目(2003AA775045)

摘　　要：递归最大值滤波检测是一种按照最大值递归累加实现目标检测的方法。针对算法存在的目标膨胀问题原文献给出了一个对累积帧使用最小值滤波的处理方法。通过分析和实验我们发现原文给出的方法存在丢失正确累积点的问题,针对该问题我们提出了一个新的抑制目标膨胀的方法,并在此基础上实现了对点目标的有效检测。Recursive-Max-Filter algorithm accumulates image sequences recursively to detect targets in clutters. As there is a problem of targets dilation, a Min-Filter is introduced into the accumulating process by the author. But we find that Min-Filter has a foible that will discard the right target-accumulated-points in some cases through analysis and experiments. Therefore, we give another method here to deal with targets dilation and an efficient point-targets detection algorithm based on modified Recursive-Max-Filter.

关 键 词：点目标检测 递归最大值滤波 递归累加 红外侦察系统 信噪比 多帧检测技术 

分 类 号：TN215[电子电信—物理电子学]