Kernel based empirical mode decomposition and its application in gait signal de-noise

Gait signal analysis of quantitative has been a challenging task over the past decades for its non-linear and non-stationary nature. Empirical Mode Decomposition (EMD) is a data-driven signal analysis method developed by Norden. E. Huang, it is especially suitable for non-linear non-stationary signal processing. It has been successfully applied in many problems, but the envelop algorithm using traditional cubic spline interpolation by Norden. E. Huang have border swing problem and extra oscillations, it is because that the cubic spline interpolation couldn´t adapt to the nature of the signal envelop, inspired by the ideas from machine learning, a new algorithm which is improved kernel ridge regression to estimate envelop of signal using the extrema is proposed in this paper. The new algorithm can used to recover the corrupted test signal. Numerical simulations show higher performance of the proposed algorithm than the traditional one.