Purpose: The purpose of this study was to develop algorithms and software that can track the trajectory of table tennis balls using image-processing algorithms to obtain information quickly for under and establish tactics used in table tennis. Methods: The algorithms used in the field of computer vision were applied on two matches played by novice and two matches during international competitions by elite athletes. Reliability analysis was performed by comparing the table tennis ball bounce frequency in each zone obtained through the automatic method and the manual method. Results: The mean reliability of the two novice games was only 85.1 ± 3.69%, total error was 14.9 ± 3.69%, overestimation error was 52.2 ± 9.78%, and underestimation error was 47.8 ± 9.78%. While the mean reliability of the two international tournaments was 71.8 ± 0.87%, the total error was 28.2 ± 0.87%, overestimation error was 82.0 ± 8.03%, and underestimation error was 19.2 ± 7.75%. Conclusions: Although the target reliability of algorithms and software developed in this study was achieved only in novice competitions with 80%, the over-estimation errors were generally high in international competitions, showing the potential for further improvement.
© Copyright 2020 Korean Journal of Sport Science. Korea Institute of Sport Science. Alle Rechte vorbehalten.