A Study of an Automated Scoring System for the Twist Skill in Horizontal Bar of Artistic Gymnastics

Abstract

This paper presents an automated scoring system for the twist skill in Horizontal bar based on motion image analysis. In this system, training scenes of Horizontal bar are taken by a high-speed video camera, and then a gymnast’s region is extracted from a video image frame by frame based on a background subtraction method. Next, the body axis of the gymnast in the moment when a twist skill completed is estimated from the gymnast’s region. Finally, the deduction point for the twist skill is automatically decided according to the rules described in the Code of Points. In experiments using video images for 26 practices, it was shown that about 80.8% of the correspondence rate between estimated deduction scores and true ones calculated by the correct ‘SCF’ and ‘Body axis’ was obtained. This can be promising as a result at the preliminary stage of this research.