Real-time Eye Tracking Technique for Multiview 3D systems

초록

This paper presents the real-time multi-user eye tracking technique for multiview 3D systems. The proposed technique used the Haar feature-based face detection and classification. Then, it calculated the best matching template, and extracts eye positions based on biological proportion. Simulation results showed the proposed method enhanced the average F-1 score up to 0.312, compared with conventional methods.