Motion pattern tracking classification in bowling matches
Loading...
Date
2017-06
Authors
Daryl Tan Hock Ann
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Motion pattern tracking is the tracing of object movements and transferring the informative data for analyses. Previous motion tracking studies in sports focused on the changing position of the players rather than actual body segment interactions. Thus, a gap is identified for further research works. Therefore, this paper presents a motion pattern tracking approach for bowling game posture classifications. The objectives are to (i)design a framework to recognize and classify motion patterns, (ii)explore sequences of motion strategies, (iii)analyze motion to recognize bowlers‟ relative movements by dominant posture sequences and (iv)examine motion pattern characteristics by classification analysis and rules-reasoning under several parameters namely shoulder, body bend, balance, swing angles and distance of feet. Motion is tracked on sequential image frames of video records and the numeric data retrieved using the Photoshop tool. Preprocessed data are classified using WEKA software for analytical information and grouping body motions into three predefined classes: GOOD, MODERATE, BAD. The main classifier chosen is the Random Tree. The findings show four main conditions concerning body motion to result in GOOD body motion postures for bowling mainly Rule 1: final shoulder angle is < final body bend angle39.17Rule 4: maximum change in distance of feet> 69.82cm and final body bend angle<56.19⁰.