Abstract
One of the major problems encountered in recording ECG is the appearance of unwanted distortions induced by power line interference in the electrocardiogram. In addition, infections due to impulse noise leads to variations in the amplitudes which represent the abnormalities associated with the heart. This paper proposes a novel approach for addressing both the aforementioned issues in ECG signals employing sub-band decomposition using wavelets analysis. Morphological filtering is applied to the detail sub-bands for removal of impulse noise using one-dimensional structuring element. Further, the powerline interference is removed using IIR Butterworth filter providing significant reduction in power spectral density levels between 50 to 60 Hz. The finally reconstructed ECG signal yields reasonably good impulse noise as well as powerline suppression using the proposed approach.