基于地铁自动售票口的人群密度检测及预警

摘要

The popularity of subway automatic ticket machines is convenient for people to buy tickets,but the crowded crowd before the ticket machine has brought great risks to the subway security,people density detection and early warning of the automatic ticket port is necessary.Based on the large density,serious stranded and severe blockage characteristics of the subway,a fast re?al-time detection and early warning method is proposed.First,the video is perspective corrected and divides the detection sub-re?gion,if the sub-region foreground pixel numbers exceed the set threshold,the sub-region uses the texture feature to estimate the people density,and when multiple sub-regions are estimated to be high-density levels,an alert is issued.Experimental results and field applications show that this method is simple and effective.Able to detect the people density of the ticket port in real time and can accurately make early warning to guide the evacuation of the crowd.%地铁自动售票机的普及方便了人们购票,但售票机前拥挤的人群给地铁站安全带来了巨大的隐患,对自动售票口人群密度的检测及预警非常有必要.基于地铁人群密度大,人群遮挡且滞留严重的特点,提出了一种快速实时检测预警方法.先对全场景进行透视矫正划分检测子区域,如果子区域前景像素数超过设定阈值,则对该子区域使用纹理特征估计人群密度,当多个子区域被估计为高密度等级时,则发出预警.实验结果及现场应用表明,这种方法简单有效,能够实时检测售票口的人群密度并能准确做出预警,指导管理人员疏散人群.