Camera Placement for Minimizing Occlusion in Object Tracking Systems

In this paper, we present camera placement approaches which consider both object occlusion and object resolution for multi-camera surveillance systems. The approaches aim to minimize occlusion and to obtain clear objects for effective object detection and tracking. To determine camera placement, we utilized various methods based on the K-means and the Principal Component Analysis (PCA) using object positions. The proposed approaches have robustness against occlusion and incremental detection capability of moving objects. To evaluate approaches, we have acquired object positions from cameras and compared three methods.