Priori-block sparse Bayesian learning algorithm for compressed sensing based ECG recovery

Compressed sensing (CS) has advantages in low-cost, low-power and long term wireless electorcardiogram(ECG) applications. However, there are some problems in existing reconstruction algorithms, which include the unsatisfied signal quality, huge computation task and no-adaptive to noise. To accurately reconstruct the non-sparse ECG signal, a priori block sparse Bayesian learning (P-BSBL) algorithm is proposed in this paper. Based on the block sparse Bayesian learning, the P-BSBL introduces priori of ECG signals to enhance the performance of the algorithm, which adopts the " nearby" zero solution space as the initial values and the signal statistical characteristic as the stop condition. The numerical experiments on MIT-BIH ECG database were conducted to verify the algorithm. The results show that the proposed method can efficiently reconstruct the non-sparse ECG signal with high signal quality. The P-BSBL has better performance compared with the convex optimization and greed methods; and it is more suitable for the ECG signal reconstruction with high data compression ratio and variable signal-to-noise ratio.