一种基于高斯混合模型的视频监控运动物体提取系统

A Moving Object Extracting System for Surveillance Based on Mixer Gaussian Probability Model

现有的视频监控系统必须要依靠人对监控图像的实时观察才能发挥作用。设计并实现了一种基于高斯混合模型的自适应的视频监控运动物体提取系统,使用高斯混合模型对监控场景进行建模,利用对象形状的空间连通性和最小像素尺寸约束去除噪声影响,从而实时地分离出前景中的运动物体并对其进行追踪和计数,仿真实验的结果证明了系统的有效性。

Now the working of video surveillance must depend on the man＇s real time observation of video. This paper designs and implements an adaptive moving object extracting system for surveillance, which explores mixer Gaussian probability process to model the video pixels, uses object spatial connectivity and mini size pixels constraints to reducing noise. The system can effectively isolate the moving object from background and counts the moving object number. The good results of simulate experiments prove that the system is effective for surveillance.