A gesture description model based on synthesizing fundamental gestures

The purpose of this work is to describe images that enable accurate portrayal of the intentions of a multimedia provider when making videos in which humans are the subjects and when communicating using images. Human gestures as a means of portraying intentions are studied and a gesture-description model that takes the meaning of such actions into consideration is developed. In the developed gesture-description model, a meaningful action is defined as a "fundamental gesture" and gestures are described using a combination of these fundamental gestures. The authors analyzed the polysemous nature and significance of single actions, extracted from images of interactive scenes containing people, and extracted effective single gestures. These were defined as "fundamental gestures" and placed in a database. A trial for two-dimensional human-animation system was then used to synthesize these fundamental gestures and conduct a subjective evaluation test to determine the effectiveness of the synthesized video animations in expressing meaning. The trials showed that by synthesizing gestures with the developed model, the ambiguity of gesture meanings could be reduced and the quality with which the meaning of gestures is expressed could be improved.