Foulds Richard A
Department of Biomedical Engineering, New Jersey Institute of Technology, Newark 07102-1982, USA.
IEEE Trans Inf Technol Biomed. 2006 Jan;10(1):199-206. doi: 10.1109/titb.2005.855496.
Computer animation of sign language used by deaf individuals has been produced from 51 time-varying trajectories of the fingertips, the centers of rotation of the joints of the hands and arms, and facial landmarks. These trajectories are sampled at 1/5th the National Television Systems Committee (NTSC) video frame rate and interpolated using piecewise sequences of cubic Bezier splines. The resulting trajectories are used to control sparse, stick figure animations of sign language resulting in considerable spatiotemporal compression with intelligibility in the range of 90%. This method introduces an additional 5:1 compression in the temporal domain that has not been previously exploited, and is potentially useful in sign language telecommunication, multimedia presentations, and gesture recognition.
已根据51个指尖、手部和手臂关节旋转中心以及面部特征点的时变轨迹制作了聋人使用的手语计算机动画。这些轨迹以国家电视系统委员会（NTSC）视频帧率的五分之一进行采样，并使用三次贝塞尔样条的分段序列进行插值。所得轨迹用于控制手语的稀疏简笔动画，从而在时空上实现可观的压缩，清晰度达90%左右。该方法在时域引入了此前未被利用的额外5:1压缩，在手语电信、多媒体展示和手势识别方面可能具有实用性。
