Application of optimized wavelet threshold de-nosing method in pedestrian navigation system

In view that the random drift noise of MEMS gyro output signals is relatively large, an optimized wavelet threshold method is proposed based on the analysis of traditional wavelet threshold de-nosing method for indoor pedestrian navigation applications. A continuous wavelet threshold function is constructed by this method, whose wavelet coefficients are between those from the soft and hard threshold function. In a certain degree, the new method overcomes the inherent defects of the traditional threshold function. Finally, the experiments are made by using different methods to de-noise the MEMS gyro data from the pedestrian navigation system, and the results show that the signal processed by the optimized threshold method achieves higher SNR of 40.8748 dB, and the MSE is reduced by 40%. The random noise in the MEMS gyro is removed effectively, and the optimized wavelet threshold method meets the needs of the follow-up study on the pedestrian navigation.