In this project, we propose a novel virtual training system for baseball batting by means of using motion segmentation, keyframe extraction, example-based learning, and markerless motion capture technologies. We divide the proposed system into two stages. In the offline stage of the system, some professional baseball players throw and hit baseballs in order to capture their ball throwing trajectories and batting motions. These batting motion clips are automatically segmented and classified to accomplish the example-based learning for further baseball batting training in the online stage. During the online training stage, we design a markerless motion capture to track the user’s continuous batting motions; meanwhile, the user can take an augmented-reality-batted training via a 3D AR glasses. Finally, after the motion tracking and analysis by the system, the user will obtain the quantitative evaluation of the virtual baseball training against the professional baseball players and make progress in batting baseballs increasingly.

1.低成本設備
2.市面上任何攝影機皆適用本技術
3.使用者無需穿戴任何特殊衣物
4.可錄製人體標準動作

近幾年因虛擬實境技術崛起，同步帶動遊戲、動畫、電影產業相關技術發展，本技術對於角色動作的擷取、錄製提供一種簡便且低成本的技術，本技術之動作擷取功能僅需使用一般光學式攝影機取得來源影像，未來可延伸於行動裝置如手機、平板等大眾化裝置，對於一般民眾而言，此種技術更容易取得且可應用於生活周遭，此項優點大幅增加市場開發的潛力，也更容易讓一般民眾所接受，未來可延伸開發之應用也相對廣泛。