运动背景下弱小目标的提取算法  

Capture of dim target in moving background

作　　者：宋海峰[1] 陈兴无[1] 王磊[1] 游安清[1] 

机构地区：[1]中国工程物理研究院应用电子学研究所,四川绵阳621900

出　　处：《强激光与粒子束》2006年第10期1625-1628,共4页High Power Laser and Particle Beams

基　　金：国家863计划项目资助课题

摘　　要：对低信噪比运动背景下序列图像中弱小目标的提取算法进行了研究,提出一种新的目标提取算法。根据运动背景的相关性和目标运动的连续性,采用背景模板匹配技术,计算各帧背景运动量,去除复杂背景,用自适应预估和阈值分割技术捕获跟踪目标。实验中,结合所研究的算法,对100帧包含低对比度弱小目标的运动图像序列进行了处理。实验结果表明,此方法能有效地实现目标的自动捕获与跟踪,跟踪效率高,跟踪稳定。A suitable method is proposed for detecting and tracking dim target in IR image sequence with low contrast and moving background. According to the correlation of moving background and position continuity of the moving target, the movement of background is computed with image matching. Complicated background is removed. Target is detected and tracked through adaptive prediction and threshold segmentation. Sequence images of one hundred frames with dim target are processed. The experimental results show that the method can automatically detect the target and track it effectively and steadily.

关 键 词：运动背景 弱小目标 捕获跟踪 模板匹配技术 阈值分割技术 

分 类 号：O438[机械工程—光学工程]