摘要 |
<p>A method of coding an image of a human face and shoulders from a sequence of images comprising: applying a three-dimensional (3D) model to the human face of at least one of the images in the sequence; estimating global movement of the human face using the 3D model; applying facial expression coding to estimate local motion of the human face; applying a two-dimensional (2D) model to the shoulders of at least one of the images in the sequence; and estimating global motion of the shoulders in the sequence of images using the 2D model.</p> |