Abstract:Baseline wander seriously influences the feature extraction and recognition of Electrocardiography (ECG) signals. The effect of the baseline correction method determines the accuracy level of medical diagnosis. A baseline correction algorithm for ECG signals based on empirical wavelet transform and piecewise polynomial fitting theory is proposed in this paper. Firstly, empirical wavelet transform is used to adaptively segment the spectrum of ECG signal, on the segmentation interval a suitable wavelet window is constructed to extract the empirical modal component with tight support. The empirical modal component with the baseline wander component removed is reconstructed. Then, the piecewise polynomial fitting is performed to remove the residual baseline wander from the ECG signal. The test results for the same ECG signal show that compared with the original empirical wavelet transform algorithm, the proposed algorithm improves the signaltonoise ratio (SNR) by more than 19 dB. The proposed algorithm can effectively correct the baseline wander distortion while maintaining good morphological characteristics of the ECG signal.