ECG signal baseline wander removal using wavelet analysis

Abstract

One of the most common problems of ECG recordings is the baseline wandering in the ECG signals during data collection. Baseline wander elimination is considered as a classical problem in ECG signal filtering. This paper presents two wavelet analysis (WA) based ECG signal baseline wander removal methods, The Discrete Wavelet transform based method uses a high level decomposition and eliminates the lowest frequency component. The wavelet packet based searching algorithm uses the energy of the signal in different scales to identify the baseline wander. The algorithm calculates the corresponding energy of wavelet packet coefficients at each scale. After a comparison, the branch of the wavelet binary tree corresponding to higher energy wavelet spaces is chosen. These procedures are tested using specific data records.