Abstract: Improved methods of moving object detection are classified into three categories，and an improved algorithm of Gaussian mixture model is presented for the application on a video monitoring system with a static camera.By analyzing the Gaussian variance function，this paper makes the variance a constant value instead of updating it.At the same time，the learning rate of mean value is fixed.Experiments show that the improved algorithm outperforms the other traditional methods in realtime performance and stability.

Key words: intelligent video monitoring system, moving object detection, Gaussian mixture model, variance

摘要： 将运动目标检测的改进方式分为三类。针对固定摄像机的视觉监控系统，提出了一种改进的高斯混合模型算法。通过对方差在高斯混合模型中的作用进行分析，省略方差更新，将方差设为固定值，均值学习率采用固定值。实验结果表明，同传统检测方法相比，改进的算法具有更好的实时性与可靠性。

关键词: 智能视觉监控系统, 运动目标检测, 高斯混合模型, 方差