A MODEL-BASED FRAMEWORK FOR AUTOMATIC TRACKING AND COUNTING OF PEDESTRIANS IN VIDEO SEQUENCES
Gianluca Antonini, Santiago Venegas Martinez, Jean‐Philippe Thiran · 2005
In this paper we propose a framework for automatic detection, tracking and counting of pedestrians in video sequences. The detection and tracking parts are based on an integrated behavioral model for pedestrian dynamics with standard image processing techniques. The target’s counting method is based on a hierarchical clustering of pedestrian trajectories where the data representation is based on the maximum of cross correlation. 1.