Pedestrian walking direction from video
Slimane Larabi, Amina Bensebaa · 2017
In this paper we consider direction estimation of pedestrian motion. The idea is to base estimation on image sequence analysis, where the detection of the head and toe points provide measurements for vanishing point estimation (roughly, these points move along parallel lines in 3D, when the direction of walk is fixed). For each frame, the top and the bottom points of the segmentation are extracted. One line is fitted to the collection of top points and one to the collection of the bottom points. Given these two lines, the vanishing point is estimated and from this the direction of the pedestrian. Experiments conducted on database available in public demonstrate the efficiency and robustness of the proposed method. Obtained results are compared to the state of the art and discussed.