Abstract :
This paper presents a modeling method of human motion for the purpose of constructing a human body/structure/motion database. Various representation methods of the human body and of its motions are used in Computer Graphics, Virtual Reality, etc. Furthermore, measuring and analyzing methods of human motion using for instance 3D digitizers, high resolution video cameras are presented in many fields such as rehabilitation, ergonomic design and production planning. Besides, many new applications that can analyze the human motion from a medical point of view are available. In the future of these fields, we consider that a database system which can manage various human data and support flexible query operations is needed. In order to compare a new captured human body/structure/motion data with another human data in a computer system, the data need to be unified and to correspond to semantics. We focused on the modeling of human motion by including an example motion, describing its actual motion and semantically associating the example with the motion. In fact, in a previous work, we defined the concept of Hub which represents an example motion. We also defined the concept of Mediator which not only represents the individual motion characteristics but also associates the Hub with the real human motion data detected from multimedia such as videos. To extend our previous work, we now consider the notion of order in the sequences of the human motion. In that we introduce the concepts of Atomic Unit and Motion Graph. These concepts allow us to analyze human motions and to compare their sequences. Indeed they are a good way to unify data from various applications.