| Title |
Algorithm for Pedestrian Monitoring System |
| Authors |
사베리에프 블라디미르(Saveljev Vladimir) ; 손정영(Jung-Young Son) ; 허광희(Gwanghee Heo) ; 김용석(YongSuk Kim) ; 문건희(Geon-Hee Moon) |
| DOI |
https://doi.org/10.5573/ieie.2021.58.1.57 |
| Keywords |
Pedestrian monitoring system; False alarm occurring rate; Invariant image; Shadow remove; Long/short term image brightness average; Adaptive threshold level |
| Abstract |
An algorithm for reducing false alarm rate by minimizing the shadow effects which are the major reason of causing the alarm is introduced in a pedestrian monitoring system. The false alarm is induced by shadows and objects other than pedestrian in the boundary region between a safety region and roads for cars, even though the pedestrians are staying at the safety region during the time period before the green traffic light is on. Since this algorithm is based on invariant image that is insensitive to surrounding illuminating conditions or brightness level, it reduces greatly the false alarm rate that are caused by the shadows from surrounding objects in the day time and overlapped to the boundary region. At the same time, the algorithm includes long and short term brightness average of the boundary region to define the adaptive threshold level for the further reducing the false alarms caused by fallen leaves, garbage, road wears and so on. The false alarm rate achieved with this algorithm is less than 5%. |