ECG Signal Enhancement with Serial Cascade OWA Filter

Abstract

Noises which appear during recording of biomedical signal are seldom characterized by the Gaussian distribution, whereas a noise can have an impulsive nature. An application of traditional filtering method at such assumption can lead to introducing a distortion to a filtered signal. In this paper, an overview of the robust filtering problem is provided. The new serial cascade robust filter with application of the ordered weighted aggregation (OWA) operator is proposed. The OWA operator is described applying a weighted average of ordered signal samples in a moving window. By introducing the nonlinear sorting operation and assigning to the sorted samples appropriate values of weights the improvement of filtering in impulsive environment is achieved. The structure of the proposed serial cascade OWA filter consists of two successive OWA filters. The proposed filter is tested with the help of real signal corrupted with an artificial as well as a real noise. The indicator of the filter's performance quality is the root mean square error (RMSE). For testing purpose the electrocardiographic signal is used. The obtained results show that in the field of impulsive noise suppression, the proposed serial cascaded OWA filter lead to better results compared to the reference filters.