Development of Pattern Recognition Methods for Golf Swing Motion Analysis

The golf swing is one of the most complex movement sequences in any sports. Even human experts can be overwhelmed by the amount of details that have to be taken into account for its analysis. We present a novel pattern recognition approach that can help in this analysis by automatically and robustly evaluating even tiny swing differences. Our approach is based on the accurate 3D spatiotemporal information about the posture of the golfer and the position of the club that the TaylorMade MAT-T (TM) motion-capture based swing measurement system provides. Golf club fitting experts have been using these systems throughout the last decade and have captured more than 500 000 swings worldwide. Based on the positional data contained in this unique database, we developed a feature description of the golf swing with the goal of classifying even small differences between groups of players. In this manuscript, we show the results of the application of several classifiers to two selected problems of group classification. The presented system can be used for example to distinguish expert from novice players. The information that is calculated by our software tool can substantially support the process of golf club fitting and furthermore assist coaches and golfers to improve performance.