Design and simulation of ultrasonic denoising algorithm using wavelet transform and ICA

Abstract

In this paper a new algorithm for signal denoising is proposed. The algorithm uses two signal processing techniques, wavelet transform and independent component analysis. The approach is based on removing the limitations of one technique by taking advantages of the other technique. In previous works, denoising process causes removing some signal components. In this work an algorithm is presented that keep all them. The performance of this method is studied on some test standard signals. In addition, as a practical application, this method is applied to ultrasonic signal denoising that widely use in nondestructive testing (NDT). The simulation results show up to 2dB improvement in SNR comparing wavelet transform.