D. Mottet et al., A NOTE ON DATA SMOOTHING FOR MOVEMENT ANALYSIS - THE RELEVANCE OF A NONLINEAR METHOD, Journal of motor behavior, 26(1), 1994, pp. 51-55
The goal of this experiment was to test a potentially useful nonlinear
method for smoothing noisy position data, which often is encountered
in the analysis of movement. This algorithm (7RY) uses a nonlinear smo
othing function and behaves like a low-pass filter, automatically remo
ving aberrant points; it is used prior to differentiation of time seri
es so that usable acceleration information can be obtained. The experi
mental procedure comprises position data collection along with direct
accelerometric data recording. From the position-time data, (a) 7RY an
d (b) Butterworth algorithms have been used to compute twice-different
iated acceleration curves. The directly recorded acceleration measurem
ents were then compared with the acceleration computed from the origin
al position data. Although the results indicated an overall good fit b
etween the recorded and the calculated acceleration curves, only the n
onlinear method led to reliable acceleration curves when aberrant poin
ts were present in the position data.