| 摘要: |
| 针对现有基于视频监控的人流量统计方案成本高、算法复杂且不利于个人隐私保护的局限性，利用毫米波雷达体积小、成本低、分辨率高的特点，提出了一种基于双时间点检测的人流量监测方法。该方法先获取人体目标散射点位置和多普勒频移信息来构成点云数据，然后根据多普勒频移正负来判断人体的运动方向，并筛选具有高多普勒频移值的点云数据以降低干扰点对聚类结果的影响；在双时间点对特定区域内人员数量进行统计，并根据双时间点之间所获取的点云数据聚类结果对所统计人员数据进行修正。实验结果表明，该方法能够用匿名的方式以较高的正确率统计人员进出。 |
| 关键词: 毫米波雷达 人流量统计 点云数据 多普勒频移 双时间点检测 |
| DOI： |
|
| 基金项目:国家自然科学基金资助项目（62071238）；江苏省自然科学基金（BK20191399） |
|
| A pedestrian counting method based on double time point detection using millimeter wave radar |
| CHEN Jinli,QU Yantao,FENG Yangge,FU Shanteng,WANG Lizheng,FAN Chenyang |
| (School of Electronic and Information Engineering,Nanjing University of Information Science and Technology;Nanjing Changge Technology Development Co.,Ltd.) |
| Abstract: |
| Considering the limitations of the existing traffic statistics schemes based on video surveillance,such as high cost,complex algorithm and being unfavorable to the protection of personal privacy,this paper proposes a pedestrian counting method based on double time point detection by integrating the characteristics of small size,low cost and high resolution of millimeter wave radar.First,the position of the scattering points and the Doppler shift information of the human body are obtained to form the point cloud data.Then the direction of the human movement is judged according to the positive and negative Doppler shift,and the point cloud data with high Doppler shift values is reserved to reduce the influence of interference points on the clustering results.The number of people in a given area is counted at two time points,and the statistical personnel data is modified according to the clustering results of point cloud data achieved between two time points.The experimental results show that this method can count the entry and exit of personnel with a high accuracy rate in an anonymous manner. |
| Key words: millimeter wave radar pedestrian counting point cloud data Doppler shift double time point detection |
