Pedestrian Tracking Based On Multiple Laser Scanners

Pedestrian tracking, as an important fundamental technology, shows a great application value in some fields such as security monitoring,traffic statistics and so on. Currently, pedestrian tracking based on computer vision has already well developed, and it is applied in many fields effectively. But tracking based on computer vision has its own inherent defects, for example badly affect of illumination; the loss of depth information during the imaging process; high complexity of processing algorithm and so on, which makes the result of computer vision tracking unsatisfied under some situations such as blocking,dark and so on.As an effective distance sensor, laser scanner can obtain 3-D distance information directly, and it is completely not affected by the situations outside such as illumination, so during these years laser scanner is wildly used in robot navigation,pedestrian tracking and other areas.This article proposed a method of pedestrian tracking based on multiple laser scanners system. By distributing multiple laser scanners in a wild area reasonably, setting the lasers' scan plane near the height of pedestrian's ankle to obtain the location data of pedestrian's both feet, at same time collect the background information of environment; after Merging the datas from different laser canners under the same Cartesian coordinate by the calibration among multiple laser scanners; then using the movement law of pedestrian to build up a model of pedestrian; at last to achieve the tracking of pedestrian through Kalman filter.According to the method we proposed above, this article constructs a real experimental system of pedestrian tracking. This system sets two single-row laser scanners in hall and gallery to collect real datas of pedestrians seperately for the experiment. The result of the experiment shows that multiple laser scanners system can track the multiple pedestrians simultaneously, at the same time under the blocking situation the tracking can still works well. The experiment meets the requirements which we expected before.Compare to the pedestrian tracking by computer vision, the method of pedestrian tracking based on multiple laser scanners this article proposed has some advantages as follows:not affected by illumination of environment,easy processing algorithm,stable and reliable target tracking. It appears more appropriate under the situation of crowed and wild places, and it shows a well application prospect.