Method, System and Computer Program for Remotely Controlling a Display Device via Head Gestures

摘要

A method for remotely controlling a computing device comprises repeatedly capturing an image from a video frame, detecting a human face in the captured image, matching the detected human face to a previously detected human face, extracting facial landmarks from the matched detected human face, estimating a 3D head pose of the matched detected human face based on the extracted facial landmarks, the 3D head pose being represented in an egocentric coordinate system by a 3D pose vector which is directed from the human face, the 3D pose vector being free to rotate around x-, y- and z-axes of the egocentric coordinate system using respective rotation matrices and free to translate along these x-, y- and z-axes using a translation vector, and controlling a user interface on a display screen of the computing device according to the estimated 3D head pose.