WAVELET BASED DE-NOISING OF NON-STATIONARY KINEMATIC SIGNALS

  • Sarah Domone
  • Jon Wheat
  • Simon Choppin
  • Nick Hamilton
  • Ben Heller
Keywords: impact, filtering, acceleration, badminton

Abstract

Wavelet Based De-Noising (WDN) is a time-frequency filtering technique that can localise the frequency content of signals. Developments such as translation invariant de-noising and the optimisation of basis function selection have addressed some of the issues associated with WDN reported in the past. The aim of this study was to explore the use of improved WDN techniques for processing non-stationary racket kinematic signals during a badminton smash. WDN was able to better preserve signal features than digital filters and this was especially evident for acceleration. However, 'pseudo-Gibbs' artefacts appeared to be present in some of the signals after double differentiation. Further work should focus on the application of WDN to a wider variety of non-stationary kinematic signals such as the golf swing, baseball batting and tennis strokes.