A robust human pointing location estimation using 3D hand and face poses with RGB-D sensor

We present a robust method to estimate human directed target positions in a display. In the proposed method, we utilize 3D hand and face poses instead of hand pose only, to compute human pointing directions and increase accuracy of location estimation. To combine hand and face pointing location estimates, a soft-switching fusion is applied. In the proposed approach, we use a RGB-D sensor which gives more information (depth information) than conventional camera sensor. As a result, we present an experiment that shows the proposed method is more accurate than a single hand or a face pose based target point estimates.