摘要
针对某武器试验中背景复杂,现有的背景差分法在背景模型的维持和更新不能用于长期和复杂的场景,以及对近地目标提取检测困难的问题,提出一种改进的背景差分法。该算法采用结合邻域信息的背景差分法和最大方差阈值法,能够在一定程度上减小背景滞后更新引起的运动目标检测误差,且使目标边缘提取更加明确,从而提高了系统的运行速度,实现复杂背景下的运动目标检测。在Visual C++6.0中用OpenCV实现了相关算法的设计,并给出了完成系统任务所需的部分关键代码,实现了运动目标和试验场景的分离与提取。仿真实验验证了该算法的有效性以及实时性。
Propose a kind of improved background difference method against the complex background in weapons testing,for the reason that the background difference in the current can＇t be used to maintain and update the background model of the long-term and complex scenes.The proposed method combines the background method using neighborhood information and the maximum variance threshold,which can reduce the deviation in moving target detection to some extent,and extract the edge of the target more clearly,thus improves operating speed of the system and achieves moving target detection in complex background.Corresponding algorithm is designed using OpenCV in Visual C＋＋ 6.0,and part of the key codes are given.Separation and extraction of the moving target from the testing scene are achieved by using the codes,and simulation results show the effectiveness and real-time of the algorithm.
出处
《计算机技术与发展》
2012年第2期49-52,共4页
Computer Technology and Development
基金
陕西省科学技术研究发展计划项目(2011K06-22)
关键词
智能视频监控
运动目标检测
OPENCV
背景差分法
intelligent video surveillance
moving target detection
OpenCV
the background difference method
