Summary
Smart Info-Media Systems in Asia
2019
Session Number:RS3
Session:
Number:RS3-3
Dense Trajectory Length and Direction Feature Based Player Receive Reaction Time Detection in Volleyball Game
Yanchao Liu, Xina Cheng, Takeshi Ikenaga,
pp.141-145
Publication Date:2019/9/4
Online ISSN:2188-5079
Summary:
The player reaction time plays an important role in volleyball game analysis which has been widely applied in volleyball strategy analysis and coach assistance. In order to achieve the purpose of detecting the reaction time, the difficulties including 1) distinction of feet motion change, 2) small motion of moving arms, 3) various motions of moving arms. Aiming to overcome these difficulties, this paper proposes the feet trajectory length change feature and the arms trajectory direction change feature based on dense trajectory method. Instead of using the whole body trajectory feature, the first proposal analyzes feet trajectory length curve to detect the Standby motion and the reaction motion to distinguish the feet motion effectively. The second proposal uses the vector angle difference between the head vector and the tail vector of arms trajectory to describe the arms motion, which is robust to detect small motion and various motions of arms. The dataset is 2014 Inter High School Mens Volleyball Games held in the Tokyo Metropolitan Gymnasium in Aug. 2014. The proposed reaction time detection system can achieve 50.6%, 79.5% and 94.0% within 17ms, 33ms and 50ms tolerant error respectively.