Abstract
Pedestrian flow is much less regulated and controlled compared to vehicle traffic. Estimating flow parameters would support many safety, security or commercial applications. Current paper discusses a method that enables acquiring information on pedestrian movements without disturbing and changing their motion. Profile laser scanner and depth camera have been applied to capture the geometry of the moving people as time series. Procedures have been developed to derive complex flow parameters, such as count, volume, walking direction and velocity from laser scanned point clouds. Since no images are captured from the faces of pedestrians, no privacy issues raised. The paper includes accuracy analysis of the estimated parameters based on video footage as reference. Due to the dense point clouds, detailed geometry analysis has been conducted to obtain the height and shoulder width of pedestrians and to detect whether luggage has been carried or not. The derived parameters support safety (e.g. detecting critical pedestrian density in mass events), security (e.g. detecting prohibited baggage in endangered areas) and commercial applications (e.g. counting pedestrians at all entrances/exits of a shopping mall).
Highlights
There is strong demand from transportation planning to track and describe the motion of pedestrians, it is still a big challenge
The literature of traffic monitoring speaks about 10 different methods for pedestrians, from personnel counting to mechanical equipment. They mention remote sensing and video image recording, too (Cessford and Muhar, 2003). (Havasi et al, 2007) and (Leibe et al, 2005) presented pedestrian detection from static images, while (Sabzmeydani and Mori, 2007) and (Barsi et al, 2016) worked with videos. (Fuerstenberg and Lages, 2003) and (Gidel et al, 2010) have implemented a system tested in passenger car using laser methods. (Lovas and Barsi, 2015) applied profile laser scanner, (Benedek, 2014) has used rotating multi-beam laser scanner to detect pedestrians. (Gate and Nashashibi, 2008) have improved the pedestrian classification accuracy by recursive estimation. (Kisfaludi, 2004) used security camera images to detect pedestrian passing. (Bauer and Kitazawa, 2010), (Shao et al, 2007) have described the pedestrian motion and applied it in the detection
The raw data obtained by the profile scanning are polar coordinates: it measures distance values at given scan angles
Summary
There is strong demand from transportation planning to track and describe the motion of pedestrians, it is still a big challenge. The literature of traffic monitoring speaks about 10 different methods for pedestrians, from personnel counting to mechanical equipment. They mention remote sensing and video image recording, too (Cessford and Muhar, 2003). The average measurement frequency was 9.1 Hz (repetition rate is roughly 109.3 ms) This data set was excellently suitable to develop the required algorithms. The original data set has been stored in a folder structure having subfolders for each frame. This storage approach made the processing somewhat complicated (in comparison to the Sick single data file style)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
