Applying Wavelet Transformation in Blind Sources Separation

One of the most important problems in heart signal processing is the extraction of fetal electrocardiogram (FECG). One of the reasons that we are interested in FECG extraction is that this signal consists of important characteristics about healthy conditions of fetus. Based on available conditions, Blind Source Separation is a suitable method for this problem. Existence of noise in observed signals from electrodes on the mother's body, can affect the separation performance. Therefore signal de-noising is an important stage in this problem. In this study, using wavelet transform and optimum selection of its parameters in FECG extraction has been investigated. The first reason for using wavelet transform is to remove noise from the observed signals and the second reason is to apply it into BBS algorithms. Depending to the noise level in signals, wavelet transform can be used before or after signal separation, also it can be used both before and after signal separation. Simulation results show the performance of each method in different conditions for obtaining the desired signal at the presence of noise.