Robust Head Pose Estimation with a Depth Camera
The subject invention cover systems and methods for estimating head pose data from a sequence of depth images of a human subject, and processing the data to generate a continuous estimate of the head pose in a 3-dimensional ( 3 D) space, and to generate a 3 D head model for display and exploitation by a human subject. The invention includes a method for capturing head pose data in which the subject is provided instructions to rotate their head in a first direction until a threshold angle of rotation is reached and then are provided instructions to rotate their head in a second direction.
1 . A computer-implemented method for estimating head pose, comprising:
detecting that the face of the subject is positioned correctly within the field of view of a depth camera;
providing a first instruction to a human subject regarding a first direction in which to rotate his head;
receiving a first sequence of images from a depth camera wherein the images correspond to a first side of the subject's head;
generating a mesh that that estimates head pose based on the first sequence of images;
determining that that the subject's head has rotated to a threshold angle in the first direction;
providing a second instruction to a human subject regarding the direction in which to rotate his head;
receiving a second sequence of images from the depth camera wherein the images correspond to a first side of the subject's head;
extending the head mesh to further estimate head pose based on the second sequence of images;
determining that the subject's head has rotated to a threshold angle in the second direction; and
displaying the completed head mesh.