Shift-invariant wavelet packet for signal de-noising in ultrasonic testing

摘要

Wavelet transform has introduced innovative changes in different fields of science and engineering. One of its important applications is in the de-noising of signals and images. The wavelet packet (WP) is an efficient de-noising method, which has been used for ultrasonic testing (UT) signal de-noising; WP is a generalisation of the discrete wavelet transform (DWT). This paper proposes the use of undecimated wavelet transform (UWT) in implementing WP to overcome the limitation of the shift variance encountered in DWT. Wavelet transforms and their shift-invariance property are briefly explained. Simulations and experiments were carried out on a flow signal contaminated with a material's grain noise. Various wavelet transform processing parameters were investigated, including the number of decomposition levels, analysing wavelets and threshold settings. The results showed a superior de-noising effect of the proposed method over the conventional one.