摘要
针对视频中运动行人遮挡问题,提出了一种基于区域特征的顶视运动行人检测算法。首先结合三帧差分和背景减除算法检测出所有可能的行人运动区域,然后找出运动区域的轮廓以及区域的外接矩形,并计算运动区域的面积,最后根据外接矩形的长宽比以及运动区域面积进行区域筛选,找出有效的运动区域,从而检测出行人。实验结果表明,该方法能够快速准确地检测出视频中的行人。
A top view video pedestrian detection algorithm based on regional features was proposed for the problem of pedestrian occlusion.First,it combined three frame difference and background subtraction algorithm to detect all the probable pedestrian moving regions,and then found out the contours and boundary rectangles of each moving region,and computed the areas of moving regions.Finally,it searched out the effective moving regions from all moving regions in each frame according to the length-width ratio of boundary rectangles and the area values,consequently realized pedestrian detection.Results show that this method can detect pedestrian rapidly and accurately from videos.
出处
《微型机与应用》
2011年第10期38-40,43,共4页
Microcomputer & Its Applications
基金
国家自然科学基金(60802040)
关键词
区域特征
运动行人检测
区域筛选
顶视视频
regional features
moving pedestrian detection
region selecting
top view video
