Application of adaptive generalized mathematical morphology theory in de-noising of ECG signals

The traditional methods have disadvantages of large computation and poor results in the preprocessing of electrocardiogram (ECG) signal. Morphological filter is faster and simple, besides, the effectiveness of it is approved adequately. However, it will cause statistical bias phenomenon. This paper proposes the adaptive generalized morphological filter can improve the accuracy of filtering and reduce the bias caused by the inappropriate selection of the structural elements. By this method the characteristics of the ECG signal can be effectively maintained and distortion of signals is small. Various experimental results show that the method can effectively remove noise in ECG signals, and better filtering than wavelet method and traditional morphological method. Experiments results being satisfactory, for a robust de-noising which provides a cleaned ECG signal for detection of QRS complexes.