3D reconstruction of a human face with monocular camera based on head movement
VIP '05 Proceedings of the Pan-Sydney area workshop on Visual information processing
IEEE Transactions on Circuits and Systems for Video Technology
| Hi-index | 0.00 |
The lack of eye contact in video conference degradesthe user experience. This problem has been known andstudied for many years. To correct the eye contact, itrequires face-reorientation and eye gaze correction. Posedetermination and viewpoint determination is a vital stepprior re-orientating the face. This paper suggests anovel, simple and effective approach for thedetermination of pose and viewpoint based on headmovement. Our approach is designed for the video-basedsituations, that it takes the advantage of the availability ofa previous image, and initially with a known pose orviewpoint. It needs one feature point to determine thepitch angle and the yaw angle, and two feature points todetermine the roll angle. The number of feature points tobe tracked is significantly less than many otherresearches suggested in this area. Our approach haslittle overhead to the face re-orientation process. It is fastand efficient, and produces reasonable accuracy.