实时视频检测系统中背景更新及复杂目标定位算法

本文提出了针对固定背景帧差法的背景更新算法和多目标定位算法。在背景更新中，通过建立目标、噪声矩阵来保存图像的特征信息，利用分类的思想优化更新策略，基于像素对背景进行更新；在目标定位中，提出了基于目标对角线的连通判断模板，该模板不断生长进而检测目标的中心点和体积，可实现实时场景中多目标和形态差异目标的定位：户外实验证明了算法的健壮性和鲁棒性，可以实现背景更新和复杂目标识别．

Background Updating and Complex Targets Locating Algorithms in Real Time Video Detecting

In view of the faults of traditional method, a new background updating and complex targets locating method for background subtrac-updating with the method of classification and pixel; In the targets locating, connected judgment template which was derived from diagonal of the target. This template can get the location and volume of the targets which include different volume. From the result we can see the system is flexible enough to handle variations in multiple moving objects and other arbitrary changes to the observed scene.