基于顺序形态滤波的红外运动小目标检测  

Moving Small Target Detection Based on Order Morphology Filtering in Infrared Image Sequences

作　　者：叶斌[1] 彭嘉雄[1] 卢汉清[2] 

机构地区：[1]华中理工大学图像识别与人工智能研究所,武汉430074 [2]中国科学院自动化研究所模式识别国家重点实验室,北京100080

出　　处：《数据采集与处理》2001年第3期315-319,共5页Journal of Data Acquisition and Processing

基　　金：国家重点实验室开放课题基金资助项目

摘　　要：针对红外图像中运动弱小目标的检测问题 ,本文提出了基于顺序形态滤波的小目标检测方法 ,并给出了具体算法。目标检测分两步进行 ,首先利用顺序形态滤波抑制背景 ,检测出候选目标的位置 ,并对候选目标进行区域分割 ,最后利用序列图像中目标运动的连续性和轨迹的一致性筛选出真正的目标。作者通过实验比较了该方法与传统高通滤波检测方法在抗噪声性能、背景抑制性能以及抑制虚警目标性能的差异。实验结果表明 ,顺序形态滤波法在这三个方面都优于高通滤波法 ,它能够快速。To resolve the detection of moving small target in infr ared image sequences, a new small target detection method based on order morphol ogy filtering is presented, and the detailed algorithm is given. Two steps of ta rget detection are adopted. First suppressing background and determining the loc ation of the candidate targets by using order morphology filtering, and segmenti ng the regions where there are possible targets are performed. Finally the true targets are screened according to the principle of moving continuity and traject ory consistency of moving target in the image sequences. Experiments show th at the method is more effective in three performances of anti-noise, suppressin g background and suppressing the number of false targets than that of high pass filterin g method. It can effectively and reliably detect the moving small target with lo w SNR.

关 键 词：小目标检测 高通滤波 顺序形态滤波 图像流 红外图像 序列图像 模式识别 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]