2P1-D18 Measurement of Pedestrian Groups Using Subtraction Stereo
Yuki HASHIMOTO, Makoto Arie, Yuma Hoshikawa, Toru Ubukata, Kenji Terabayashi, Kazunori Umeda · The Proceedings of JSME annual Conference on Robotics and Mechatronics (Robomec) · 2010
This paper proposes a method to distinguish a state of walk of pedestrian groups by using "subtraction stereo" which is a stereo camera system that focuses on moving regions to make stereo matching fast and robust to the correspondence problem in computer vision. The state is classified into four categories: convergence, divergence, parallel shift, and stop. The state of walk is distinguished by an angle and a speed in 3D space of feature points of pedestrians obtained by the KLT Tracker. The effectiveness of the proposed method is verified by several experiments.