视频监控中多视角目标智能定位追踪方法研究

摘要

在视频监控中对目标进行定位追踪时,容易出现遮挡情况.当前定位追踪方法需预先检测出遮挡状况,再对目标位置进行校正,计算量大,且追踪结果不可靠.为此,提出一种新的视频监控中多视角目标智能定位追踪方法,针对每个摄像头采集到的图像,对其进行二值化处理,通过稀疏表示方法对目标进行定位.构建了视频监控中多视角目标跟踪的概率框架,针对每个智能摄像机,建立视频监控中多视角目标追踪模型.通过不同视角中的视频数据求出本地似然函数.为了避免维度灾难问题,对联合后验概率进行分解获取边缘概率乘积的形式,通过边缘概率求出被追踪目标在摄像机视角下的位置.依据消息传递机制实现信息融合,达到目标追踪的目的.实验结果表明,所提方法有很高的定位和追踪精度.%Fixing the target tracking in video surveillance,prone to condition,current location tracking method to detect the shade conditions in advance,and correction of target location,large amount of calculation,and tracking the result is not reliable.For this,a new kind of video monitoring target smart positioning tracking method was put forward,multiple points of view for each image obtained by camera,binarization processing,through sparse representation method to locate the target.Build the probability of the perspectives of target tracking in video monitoring framework,for each smart camera,perspectives of target tracking in video monitoring model is established,through the different angle of view and the local likelihood function of video data.In order to avoid the problem of dimension disaster,joint posterior probability decomposition to obtain edge probability of product form,through the edge probability and track targets in the position of the camera perspective.Based on message passing mechanism to realize information integration and achieve the goal of target tracking.The experimental results show that the proposed method has a high positioning and tracking accuracy.