Summary
Smart Info-Media Systems in Asia
2019
Session Number:RS3
Session:
Number:RS3-2
Body Part Categorization and Occlusion Detection Based Volleyball Players’ Spike Height Analysis
Ziken Li, Xina Cheng, Takeshi Ikenaga,
pp.136-140
Publication Date:2019/9/4
Online ISSN:2188-5079
DOI:10.34385/proc.57.RS3-2
PDF download (3.5MB)
Summary:
Volleyball players’ spike height is very important in sports analysis of volleyball as it provides a quantitative criterion to judge spike motion. Spike height is the highest point of trajectory of spike hand, which is obtained by 3D tracking. There are two problems in acquiring the position of hand: the deformation of hand and the occlusion of different camera views. This paper proposes body part categorization and occlusion detection based observation model to solve these problems. Body part categorization is a detection based observation model, which tracks the target not by image similarity but by category, so it is robust to deformation. Occlusion detection solves the problem that unknown number of occluded views by detecting occluded views and eliminating their influence. The experiments are based on the videos of final game and semi-final game of 2014 Japan Inter High School Mens Volleyball in Tokyo. The experiment results of proposed method are that: 92.86% of test sequences can be successfully detected the spike height, and in which the average error of spike height is 7.45cm.