Geometry-Driven Eye Gaze Estimation

Eye gaze estimation plays a critical role in computer vision and human-computer interaction, with applications in virtual reality, augmented reality, driver monitoring, and medical diagnostics. Existing systems face challenges including proprietary hardware requirements. This paper proposes a low-cost, high-accuracy eye gaze estimation framework leveraging the geometric principle that circular pupil boundaries project as ellipses. Our approach includes four stages: pupil ellipse detection, quadratic cone formation, circular intersection constraint application, and 3D gaze vector reconstruction through eigen decomposition. Experimental results demonstrate angular errors below 0.5° in real human model datasets, while maintaining complete mathematical interpretability.