基于特征的单帧红外小目标检测  

Infrared Small Target Detection Based on Character

作　　者：余贵水[1] 李秀峰 魏钟记[1] 

机构地区：[1]海军工程大学,武汉430033 [2]海军933办公室,北京100841

出　　处：《舰船电子工程》2010年第5期78-80,共3页Ship Electronic Engineering

摘　　要：针对红外小目标自身的特点,给出了一种改进的检测方法:首先通过低通滤波进行预处理,然后保留灰度值大于预处理图像均值的点。由于小目标在红外图像上所占像素很少且是区域能量较高的点,遍历整幅图像,滤除连续的像素点,得到小目标或斑目标,为了进一步检测分割出候选目标点,考虑到目标点的局部能量较大的特征,通过合理的设定阈值,缩小目标点的范围,为后续目标轨迹的检测提供基础。According to the characteristics of infrared small target,an improved new algorithm is proposed in this paper.First,let infrared image through low-pass filter,and then retain points whose gray values are larger than the mean one.As small targets are points that have a small quantity of pixels and share higher energy in infrared image,traversing the entire image,taking off successive pixels,retains small targets or spot goals.Furthermore,in order to test candidate target points,considering the local energy of target points is higher,we can set a reasonable threshold to narrow the scope of target points.Through these processes,basis for the detection of target track can be provided.

关 键 词：红外小目标 局部能量 阈值 

分 类 号：TN916.76[电子电信—通信与信息系统]