Real-time head tracking from the deformation of eye contours using a piecewise affine camera

A computer vision based approach for human–computer interaction through head movements is presented and evaluated in a non-immersive virtual reality context. Once intercepted and tracked in real-time using a piecewise affine camera model and affine-deformable eye contours, user head displacements are estimated and remapped onto the tridimensional graphic environment according to a natural interface metaphor. Both the real-time performance of the tracker and the improved head parameter estimation accuracy – as compared to the one obtainable using globally affine camera models – encourage the use of this approach to support diverse advanced interaction scenarios and applications.