实时背景下的遗撒物体检测研究  

Detection Study of Abandoning and Scattering Objects in Real-time Background

作　　者：化莉[1] 沈建新[2] 

机构地区：[1]淮阴工学院计算机工程学院,江苏淮安223003 [2]南京航空航天大学机电学院,江苏南京210016

出　　处：《成都大学学报（自然科学版）》2011年第4期342-344,共3页Journal of Chengdu University（Natural Science Edition）

摘　　要：在视频智能监控系统中,遗撒物体的检测是一个重要的环节.首先通过建立混合高斯模型实时更新视频序列的背景图像,然后利用背景差分实现对运动目标的检测,最后将检测出的运动物体的重心、面积及轮廓作为参数,用来判断一段时间连续视频序列中参数的不变性,从而检测出遗撒物体.实验结果表明,该方法能准确的检测出遗撒物体,具有很强的可靠性和鲁棒性.Detection of abandoning and scattering objects is an important link in video intelligent monitoring systems.Background image of updated real-time video sequence was established based on mixed Gaussian model.Then background subtraction algorithm was used to achieve the detection of moving targets.The center of gravity,area and contour as parameters which were obtained from the detection were used to determine the invariability of parameters in consecutive video sequence in a period of time and further to detect the abandoning and scattering objects.The experimental results show that this method can accurately detect abandoning and scattering objects,which has strong reliability and robustness.

关 键 词：智能监控 混合高斯模型 背景差分 遗撒物 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]