Real-time face identification using two cooperative active cameras

This paper proposes a method for real-time face detection and identification using two cooperative pan-tilt-zoom (PTZ) cameras. For each camera, the human face is detected and segmented using motion and skin color cues. The face segment is then analyzed by considering the relative position of the facial color blob to determine the pose. After facial pose is estimated, the identification is performed using a face matching method based on color-distribution. Identification results with confidence values from both cameras are weighted combined to conclude the final result. Our experimental results demonstrate successful face detection and tracking in uncontrolled background, and the system is capable for real-time face identification. The experiments also confirm the collaboration between cameras improves the identification performance.