Abstract
In knowledgeβbased coding which has attracted considerable attention as a new coding scheme for moving pictures, the model of the moving object must be controlled efficiently in the range from the rough motion of the whole body to the detailed displacement of each part when the scenes are synthesized at the receiving side.
This paper describes both a method of modeling of the structure and the motions of the object using the frame representation scheme, which is often used in the field of the knowledge engineering, and a scheme to synthesize the scenes based on the models. The objective of this paper is to represent the motion of the object rather than the changes of its surface. For simplicity, the object is limited to that which consists of connected rigid parts. In our method, the structure of the object and its motions are represented hierarchically.
Using the motion model, the motion of the whole body can be controlled by the semantic information and the displacement of each part can be controlled by the function. Furthermore, the speed and the amplitude of the motion can be changed efficiently. As an example, we have represented a simple stick model of the human body and its walking motion in the frames, and then synthesized the stick motions.