摘要
时域递归最大值滤波是一种较易实现的红外小目标检测方法,但是在递归过程中存在目标膨胀的缺陷,影响了该方法的应用。通过研究分析,提出了条件最小值滤波替换最小值滤波抑制目标膨胀,较好解决了目标膨胀问题。结合空域最大中值滤波预测背景,将点目标和强噪声保留在预测残差中,再通过递归最大值滤波对预测残差进行时域递归处理,以完成能量累积提高信噪比,设计完成的算法实现了对尺度为 1 个像素,运动速度小于 1 像素每帧的点目标的可靠检测。
Temporal recursive max filter (RMF) is a point target detection algorithm that is easy to realize but has a shortcoming of target dilation in the filtering process. Here a new method called conditional minimum filter is designed to deal with this problem and it is testified effective. With the combination of spatial max-median filter (MMF) and modified temporal recursive max filter, an effective point-target detection algorithm is developed, which uses max-median filter to predict the background and gets the residual image and then puts the residual image sequences into a RMF for the purpose of energy accumulation to improve the ratio of signal to noise. Thus, target with 1 pixel size and speed less than 1 pixel per frame can be credibly detected.
出处
《光电工程》
CAS
CSCD
北大核心
2004年第12期17-19,67,共4页
Opto-Electronic Engineering
基金
国家 863 计划资助项目(2002AA776031
2003AA775045)
