Abstract:Due to the classical wavelet had the shortcoming of computation complexity and constant decomposition levels in inertial sensors de-noising, a thresholding de-noising method based fuzzy logic and lifting wavelet transform was brought out. Firstly, the motion type was determined by a fuzzy expert system, and the bandwidth of true motion dynamic was chosen accord to the spectrum estimation base on the priori knowledge. Then, a suitable wavelet decomposition levels for each axis of inertial sensors was automatically chosen in accordance with the relationship between wavelet decomposition levels and cut-off frequency. Finally, a discrete wavelet transformation was obtained under lifting scheme, and the signal was de-noised by compresses the coefficients of wavelet transform of signal at a threshold value. Test results prove that the proposed de-noising method can distinguish the motion type correctly, and has a better results and higher computational efficiency than classical wavelet de-noising with the constant decomposition levels.