Research On ECG Signal Processing Technology Based On Sparse Representation And Morphological Component Analysis

For a long time,cardiovascular diseases have been the major factors affecting people’s physical health.Electrocardiogram(ECG),as the main observation method of heart rhythm changes,has always been an important tool in the diagnosis of heart diseases.However,as a physiological weak signal with low frequency,ECG is susceptible to external factors during the collection process,which affects the normal changes of its waveform.How to ensure the accuracy of ECG information plays a very important role in the analysis and judgment of diseases.For the above reasons,in line with the learning research idea of "ECG principle Sparse principle Morphological analysis ECG noise reduction",according to the sparse characteristics of ECG signals as well as the needs to assist professionals in the diagnosis,two kinds of noise reduction algorithms based on the sparse characteristics of ECG are proposed.These two algorithms have been verified in the Massachusetts Institute of Technology Arrhythmia Database(MIT-BIH).The innovation points of this paper are as follows:(1)Aiming at the problem that ECG signals contain baseline wander noise,this paper uses the sparse characteristics of ECG signals itself,a denoising algorithm based on the group sparsity characteristic and low-pass filtering is proposed by decomposing the ECG signals,this algorithm is used to remove baseline wander and random noise in ECG components(2)The traditional total variation(TV)denoising algorithm can produce step-shaped artifacts and underestimation of peak and trough information in the processing of digital signals,this paper proposes a total variational denoising algorithm based on group sparse threshold function(GSTV).The structural changes of ECG signals are extracted effectively when the sparse characteristics of ECG signals are selected.The proposed algorithm has been verified in MIT-BIH library.Compared with traditional methods,the proposed algorithm can effectively improve the underrated information components and the fidelity of signal recovery.(3)Aiming at the white Gaussian noise contained in ECG signals,the ECG noise reduction technology based on sparse recovery is studied.A new fusion algorithm(RRSD-MN)is proposed,which includes the sparse decomposition model of ECG resonance signals(RSSD)based on morphological component analysis,the median average filter algorithm and the normalized least mean square adaptive filter algorithm.As a digital signal processing method based on resonance characteristics rather than frequency,RSSD can decompose and process the signal effectively according to the vibration characteristics of different components in ECG signal,and use specific processing methods to remove the noise.Compared with the traditional noise reduction methods such as wavelet and empirical mode decomposition,the proposed algorithm can increase the SNR of ECG signals by up to 87% on average,and the root mean square error can be reduced to 0.08 at most.The algorithm proposed in this paper achieves the desired effect on the whole,and has validity and application value for ECG signal processing.