Abstract
A study on the motion picture plays an important role in the fields of authentication technology and sports managements for planning optimal trainings. However, such works depend on filming situation, and its difficult to analyze frontal view of scale-changing objects. The standard motion analysis software like "Dartfish" cannot handle the objects approaching to the camera. In this case, such software only shows series of picture image and it is hard to get numerical information from them. In this article we propose a statistical methodology that manage to extract information from the scale-changing objects based on the movie data by a consumer video camera and also apply this to real walking data.
