Abstract des Autors

In this study, the respective sensitivity of 2 smoothing and filtering techniques, namely, polynomial functions, splines, Fourier series and Butterworth filter have been assessed. The accelerations derived from the smoothed or filtered data were compared to those measured directly on a skier to quantify the vertical hip oscillation during 4 basic movements involving flexion and extension of the lower extremities and trunk. Furthermore, these numerical techniques were applied for a double flexion movement taken from Pezzack et al. (1977). In all cases, the accelerations calculated from data processed by the Butterworth filter compare very well to those measured directly. With regard to the 4 types of movement studied, the techniques are equally as good each others; even though, for the polynomial functions, the splines and the Fourier series, it is extremely difficult to correctly determine, a priori, the proper coefficients of the functions. Finally, only the Butterworth filter and to a certain extent the Fourier series, provide acceleration data comparable to that given by Pezzack et al. (1977). We conclude that the numerical techniques and especially the Butterworth filter are satisfactory for smoothing or filtering displacement data and that the calculated accelerations are comparable to those measured, clearly illustrating the advantages of high speed photography or video techniques. Verf.-Referat