People Detection in Downward-Viewing Fisheye Camera Networks Using Fuzzy Integral

摘要

In this paper we study techniques for accurate detection and localization of multiple people in an indoor scene covered by multiple downward-viewing fisheye cameras. In particular, we propose a fuzzy integral based method for aggregating the information from the multiple cameras to form a combined occupancy map of people in the scene. The experimental results on test videos exhibit good performance for practical use. We also propose methods to account for occlusion by scene objects at different stages of the algorithm that lead to improved results.