The use of sampling frequency and wavelet analysis to denoise a signal with a high content of white noise

Abstract A procedure for removing most of the Gaussian white noise present in a signal is proposed and demonstrated. It uses a high sampling rate, removes more details from the wavelet decomposition and then down samples to a sample frequency which will not burden further analysis with a large number of samples. It is a useful preprocessing step for signals contaminated with Gaussian white noise, particularly those with a low signal to noise ratio (snr). The procedure is illustrated with signals having a low snr. The results are very encouraging. The values of the mean square error, snr and plots validate the proposed procedure. This approach provides a practical demonstration of a technique that will be useful in processing biomedical signals.