INTRODUCTION: Ski instructors and coaches assess a skier's technique in order to recommend how he or she may improve performance. This assessment is highly subjective and requires years of experience. A quantitative method to evaluate the whole body movement patterns of skiers would allow objective comparisons of individual skiing techniques and may lead to a better understanding of motion control in skiing. In this study, a method was developed to decompose the whole body movement patterns of skiing into a set of principal movements that can be used to quantitatively describe skiing technique. The method was applied to compare the individual skiing technique of six highly-skilled racers in slalom. METHOD: The kinematics of six male members of the Norwegian Europa Cup team (age 17-20, FIS points 22.35 ± 8.21) were captured for two turn cycles during a slalom race simulation using 4 cameras and control points distributed along the course (Reid, 2010). The 3D coordinates for 21 reconstructed landmarks were transformed into a 63-dimensional (21 makers x 3D) "posture vector," p(t). Skier movements corresponded to changes in p(t) over time and were determined by a principal component analysis (PCA), similar to Troje (2002) for gait analysis. The PCA yielded (a) a set of orthogonal, 63-dimensional vectors in the direction of the main changes in posture called "principal components" (PCi); (b) Eigenvalues (EVi) quantifying how much of the variance of posture vectors occurred in direction of the associated PC; and (c) a set of 1-dimensional "principal movements" PMi(t) corresponding to the projection of the change of posture (p(t)-Pmean) onto each principle component vector PCi. RESULTS: For all subjects, the first principal component (PCi) represented the change in posture corresponding to body inclination into the turn (Fig. 1). Comparison of PM1(t) between subjects allowed an assessment of when in the turn and to what degree the movement was executed. For instance, subject 6 leaned into the turn faster, and maintained an inclined position longer, compared to subject 1 (Fig. 2). The higher order PCs represented other skier movements including trunk rotations, vertical, and fore-aft movements. Together, the first five PMs represented 95.8 ± 0.6% and the first ten PMs represented 99.2 ± 0.1% of the entire variability of p(t). DISCUSSION: Principal component analysis of posture, vectors offers a data-driven method to decompose a complex movement into a set of 1-dimensional principal movements that can be studied independently. This method allows direct comparisons between skiers as well as correlation with variables quantifying performance or loads on the skier.
© Copyright 2010 Book of Abstracts. 5th International Congress on Science and Skiing, Dec. 14 - 19, 2010, St. Christoph am Arlberg. Veröffentlicht von University of Salzburg, Interfakultärer Fachbereich Sport- und Bewegungswissenschaft/USI. Alle Rechte vorbehalten.