论文部分内容阅读
自动视频监视技术用于探测特定现场的入侵者。这一任务主要归结为下述问题———通过对单摄象机取得图象顺序进行评价来探测移动目标。此种单摄象机方法的主要问题在于不能可靠地测量目标的三维尺寸与三维位置 ,因为目标尺寸及其速度是在二维像平面内估算的。本文提出一种用第二个摄象机获得现场三维信息的方法。此法使用一种有效的三维现场模型将两摄象机数据进行综合评价 ,所用两摄象机视场部分重叠 ,与现有视频监视系统常用的布置方法相同。实验证明 ,当存在移动的阴影、树叶、鸟和昆虫等或摄象机被遮档时 ,运用两摄像机综合评价其误报率可比纯单摄象机评价进一步降低。
Automatic video surveillance technology is used to detect intruders on a specific site. This task is largely attributed to the following issues - the detection of moving targets by evaluating the sequence of images taken by a single camera. The main problem with this single camera approach is that the 3D and 3D position of the target can not be reliably measured since the target size and its velocity are estimated in the 2D image plane. This paper presents a second camera to obtain the scene of three-dimensional information method. This method uses an effective three-dimensional field model to make a comprehensive evaluation of the two camera data. The two cameras used partially overlap the field of view, which is the same as that of the existing video surveillance system. Experiments show that when there are moving shadows, leaves, birds and insects, etc. or the camera is blocked, the use of two cameras to evaluate the false alarm rate can be further reduced compared with the single camera evaluation.