Development of system for people counting by image sequence processing
Kenta MASUDA, Tsukasa TOYAMA, Nobuaki NAKAZAWA, Toshikazu Matsui · The Proceedings of JSME annual Conference on Robotics and Mechatronics (Robomec) · 2019
People counting provides valuable information on evaluating an event, marketing, and security. Therefore, it has been proposed that the systems count passing people by using a top-view camera installed at a high position in general. Here, we take into account that it can be easily installed anywhere, we propose a portable system for counting passing people by using a side view camera with a fisheye lens. In order to count passing people, we used a background subtraction method to detect moving region, and calculated its vertex. Furthermore, the direction of passing people was estimated by using optical flow. In the verification experiment, it was confirmed that the developed system could correctly recognize a walking direction and count the number of people.