On the experiment of multi camera tracking using Kalman filter and FOV lines

This paper discusses the experiment of single human tracking using multi camera. The contribution of this paper is on the use of Kalman filter as an object tracker combined with FOV lines method to distinguish the source of camera and to label the passing of moving human as a single object to track. In this paper, the experiment is limited only to track human using three still cameras with a static video background, which is emphisized on the moving human. However, the experiment does not handle occlusion between objects in the border of camera footprint. As the results, tracking algorithm that uses Kalman filter as a tracker combined with FOV lines can be used as well with multiple cameras. Since we are using centroid of human object, the performance measure in this experiment is the distribution of distance error, which represents how far the distance in pixels between the centroid of tracker bounding box and the object bounding box.