摘要
在视频监控中,需要对特定区域进行实时监控。为了克服传统的视频监控系统存在的数据量大、响应时间长和人类自身固有的弱点等缺陷导致的监控效率低下等问题,文中实现了一种智能区域入侵的检测算法。该算法是基于一种三层高斯背景建模的方法,在提取前景图像后,对前景图像进行阴影检测、全局灰度检测、目标完整性检测,同时在提取更加精确的前景目标后,分析运动目标轮廓,对目标的轮廓信息进行统计,计算出特定区域中入侵的人数。最后通过一段实际的监控视频验证了文中提出算法的有效性。
In video surveillance,a specific area needs real-time monitoring. In order to overcome the problems of low monitoring efficien-cy caused by the defects including large amount of data,long response time,and intrinsic weaks of humans in the traditional video surveil-lance system,propose an intelligent regional intrusion detection algorithm. The algorithm is based on a three-Gaussian background model-ing method,after extracting the foreground images,carry on the shadow detection,global grayscale detection and object integrity detection to them. Then statistically analyze the more precise foreground extraction,contour information can be used to calculate the number of in-vasion in the specific area. Finally,the effectiveness of the proposed algorithm is verified through a piece of actual surveillance video.
出处
《计算机技术与发展》
2014年第10期159-162,166,共5页
Computer Technology and Development
基金
国家自然科学基金资助项目(61272084
61202004)
江苏省高校自然科学研究重大项目(11KJA520002)
江苏省科技支撑计划(社会发展)项目(BE2011826)
高等学校博士学科点专项科研基金资助课题(20113223110003
20093223120001)
关键词
自适应高斯混合模型
背景建模
运动检测
阴影检测
adaptive Gaussian mixture model
background modeling
motion detection
shadow detection
