Face-direction estimating system using stereo vision

We developed the system which can estimate the face-direction of a person with two cameras using stereo vision. In this system, the eyes and the mouth are extracted in each image for measurement of their 3D position by matching each feature in the one image with the other. Then, face direction is obtained from the normal of the plane which is determined by the 3D coordinates of the features. This system can track face-direction in semi-real time.