Wavelet-based removal of sinusoidal interference from a signal

This paper presents a wavelet-based filtering approach to the recovery of a signal contaminated with sinusoidal interference of a single frequency component or its harmonics. A key step in the approach is to threshold the detail coefficients of the signal by using a level-dependent threshold. Unlike the use of notch and comb-shaped filters, the new approach does not result in notable attenuation to the signal at interference frequencies and is especially suited for the removal of sinusoidal interference that is located within the signal spectrum. Evaluation of the approach has been conducted using a simulated piecewise smooth signal and a real, rapidly changing pressure signal. Results obtained from the studies demonstrate that both the sinusoidal interference and the white noise in the signal can be removed effectively without excessive attenuation to the signal at the interference frequencies.