基于遮挡系数和区域划分的人群数目估计方法

Crowd Quantity Estimation Based on Occlusion Coefficient and Region Division

传统的基于像素统计的人数估计方法在行人之间存在遮挡时,无法对人数进行准确的估计。为了减少遮挡引起的误差,提出一种基于遮挡系数和区域划分的人群数目估计方法,利用行人在图像中的高度,前景像素的数量和前景图像连通区域信息,计算遮挡系数,减小了由于人群遮挡引起的人数估计误差,同时采用区域划分,根据行人在图像中的高度分成子区域,减少了摄像机透视效应对人数估计的影响。实验结果表明,相对于传统基于像素统计的人数估计方法,该方法提高了人数估计的准确性。

The traditional crowd counting based on pixel statistics cannot accurately estimate the crowd quantity when there is occlusion between pedestrians.In order to reduce the error caused by the occlusion,a new method is proposed to estimate the crowd quantity based on occlusion coefficient and region division.The occlusion coefficient is calculated by the height of pedestrians in the image,the number of foreground pixels and the connected region information in the foreground image.The method reduces the error of estimating the number of people caused by crowd occlusion.The method adopts region division and divides the image into sub-regions according the height of the pedestrian in the image,which reduces the influence of the camera perspective effect.The experimental results show that the proposed method improves the accuracy of crowd quantity estimation comparing with the traditional method based on pixel statistics.