ECG signal denoising based on similar segments cooperative filtering

摘要

Noise reduction is an important preprocessing for the ECG signal subsequent analysis and diagnosis. Based on the cooperative filtering of similar segments, an effective ECG denoising method is proposed. First, a signal is divided into equal-length segments, and then the similar segments of each segment are found according to the similarity and arranged into a two-dimensional array which is filtered by the Savitzky-Golay method and polynomial fitting method in two directions, respectively. In the last, all the filtered segments are put back to their original positions. The overlapping part of each segment is weighted averaged as the filtering result of this segment. Thus, the final denoised signal is reconstructed. The advantages of the proposed method lie in the information between similar segments is also used, that is, the cooperative filtering of similar segments. As a verification of the proposed method, ECG signals in the MIT-BIH database are denoised and used to other denoising methods reported in references. The results show that the proposed method is effective.