摘要
针对静态摄像机条件下的视频监控问题,提出了一种基于背景建模的运动目标检测算法。首先利用统计的方法建立了基于颜色和颜色梯度的背景模型,并实时地对背景模型进行更新,最后将这两种背景模型综合考虑对目标进行有效检测。该算法较好地解决了背景模型的提取、更新、背景扰动、外界光照变化等问题。实验结果证明,该算法对固定场景下运动目标的检测是快速有效的。
A moving object detection algorithm based on background modeling is proposed for video surveillance system with static camera. Statistical models of color and gradients are built and the models are real-time updated. The two models are used to detect moving object. The algorithm effectively deals with the problems of the abstraction and update of background, background disturbance, illumination changes etc. Experiment results show that the presented method is fast and effective.
出处
《系统仿真学报》
EI
CAS
CSCD
北大核心
2005年第1期159-161,165,共4页
Journal of System Simulation
基金
国家863计划资助项目(2001AA422270)
关键词
背景建模
混合高斯模型
颜色梯度
目标检测
background model
mixture Gaussian model
color gradient
object detection