Scalable, non-intrusive person tracking and counting solution using time-of-flight cameras

• Research unit: INNOSUISSE
• Project number: 11580.1;3 PFES-ES

Short description (German)

Scalable, non-intrusive person tracking and counting solution using time-of-flight cameras

Short description (English)

Scalable, non-intrusive person tracking and counting solution using time-of-flight cameras

Abstract (German)

The goal of this project is to develop a people tracking system employing multiple time-of-flight cameras. This system is intended to replace the mechanical and/or optical turnstile based systems, providing higher functionality at a lower overall cost. Time-of-flight cameras are a new technology which has several advantages over traditional cameras, most importantly for this application; they simplify the notoriously hard problem of background to foreground separation in real time video. One drawback of time-of-flight cameras is the small field of view. In this project, this disadvantage is circumvented by employing an array of these cameras in such a way, that a much larger field of view results. Since each camera is coupled with a computing unit, local decisions are made at each camera, and only those decisions need to be feed to the supervisor for the final tracking of the people over several cameras.

Abstract (English)

The goal of this project is to develop a people tracking system employing multiple time-of-flight cameras. This system is intended to replace the mechanical and/or optical turnstile based systems, providing higher functionality at a lower overall cost. Time-of-flight cameras are a new technology which has several advantages over traditional cameras, most importantly for this application; they simplify the notoriously hard problem of background to foreground separation in real time video. One drawback of time-of-flight cameras is the small field of view. In this project, this disadvantage is circumvented by employing an array of these cameras in such a way, that a much larger field of view results. Since each camera is coupled with a computing unit, local decisions are made at each camera, and only those decisions need to be feed to the supervisor for the final tracking of the people over several cameras.