基于时空域融合滤波的弱小运动目标检测算法  

Detection of a Dim and Small Moving Target Using a Temporal-spatial Fusion Filtering Algorithm

作　　者：孙瑾秋[1] 张艳宁[1] 姜磊[1] 黄伟[2] 

机构地区：[1]西北工业大学计算机学院,西安710072 [2]西安武警工程学院电子技术系,西安710086

出　　处：《机械科学与技术》2009年第1期20-24,共5页Mechanical Science and Technology for Aerospace Engineering

基　　金：国家教育部“优秀人才计划”项目(NCET-05-0866);国家自然科学基金项目(60472072);航空科学基金项目(20060853010)资助

摘　　要：针对CCD光电序列探测图像中的弱小运动目标检测问题,提出了一种运动星空背景下的基于时空域相结合的弱小运动目标检测算法,在空域上采用形态学Tophat滤波降低噪声增强目标,在时域上利用运动背景补偿和多帧差分滤波增强目标的运动特性,通过基于Hough变换的方法进行目标轨迹识别,完成目标检测。实验结果表明,该算法充分考虑了时域和空域中弱小运动目标的特性,能有效地在低信噪比条件下检测出复杂运动背景下的弱小运动目标。For detecting a dim and small moving target in CCD photoelectric image sequences, a temporal-spatial fusion filtering algorithm is proposed. It reduces noise and enhances targets by morphologic Tophat transform in spatial domain, improves the moving characteristic by background register and multi-frame difference filter in temporal domain, and identify the target track with Hough transform of the fusion result. It is shown that the algo- rithm has fully taken into account the characteristic of the dim and small moving target in temporal and spatial domain, and it is effective to detect the dim and small targets in moving background with low signal to noise ratio （SNR）.

关 键 词：弱小运动目标检测 时空域融合滤波 Hough变换检测 

分 类 号：TP391.2[自动化与计算机技术—计算机应用技术]