Video surveillance system has been applied widely in many fields such as intelligent security,it is one of research focuses in recent years.Segmentation of pedestrians in a video surveillance system is significant,and the effectiveness of the segmentation has a direct impact on the object recognition and behavior understanding.An automatic pedestrian segmentation algorithm is proposed in this paper.Firstly,the target is located based on the HOG pedestrian detection algorithm,and the foreground and background information is roughly obtained ; then,the pre-segmentation information is used as the prior knowledge of GrabCut algorithm,without interactive operation,and the pedestrian is segmented automatically and precisely; Finally,the experimental results illustrate that the proposed algorithm is effective and meets the application reluirements.%视频监控系统在智能安防等众多领域有着广泛的应用空间,是近年来研究的热点之一.行人的分割在视频监控系统中极其重要,其分割的有效性直接影响目标识别、行为理解等.本文提出一种自动的行人分割算法,使用基于HOG特征的行人检测算法对目标进行定位,获得粗略的前景和背景信息,把获得的预分割信息作为GrabCut算法的先验知识,无需交互式操作,自动地对行人进行精确分割.实验证明该算法不仅有效,而且能满足应用需求.
展开▼