The Research On Bioelectric Signal Compression And Reconstruction Based On Compressed Sensing

With the development of technologies such as embedded and wireless network communication,bioelectric signal acquisition devices are gradually becoming portabler,wearable lower-power.The difficulty of signal detection is decreasing day by day,and some portable devices based on bioelectric signal remote identification have attracted more and more attention.However,these devices have limited resources.If the traditional Nyquist sampling method is used to collect signals,it will bring some pressure to these devices.Compressed Sensing(CS)is a data compression acquisition method proposed.The sampling and compression of data is completed synchronously.It avoids a waste of resources and meets the requirements of small size and limited power consumption of wearable portable devices.In this thesis,for the limited resource of portable remote bioelectrical signal identification equipment,a bio-electrical signal compression and reconstruction method based on compressed sensing was studied.Taking the electrocardiogram(ECG)signal and the photoplethysmography(PPG)signal in bioelectrical signals as the research signals,a Kasami small set sequence deterministic measurement matrix was constructed and an improved segmented weak orthogonal matching pursuit algorithm was designed to compress and reconstruct the signal.Finally,the eigenvalues were extracted from the reconstructed signals for identification,which was used to analyze the accuracy of the proposed method.The main research contents of this thesis were listed as follows.(1)The reconstruction performance of ECG and PPG signals under orthogonal sparse basis and redundant dictionary constructed by K-SVD was detailed.The experimental results shown that the reconstruction effect of the redundant dictionary constructed by K-SVD was better than that of a single orthogonal sparse basis.The reconstruction quality of two kinds of signals under common measurement matrix was compared,and a deterministic measurement matrix of Kasami small set sequence was constructed.Experiments shown that when the ECG and PPG signals were compressed and reconstructed,the performance of the Kasami small set sequence measurement matrix was much lower than the Gaussian random measurement matrix,but it better than the Toeplitz measurement matrix.In addition,the Kasami small set sequence measurement matrix was used as the deterministic measurement matrix,which was easier to implement than the Gaussian matrix.(2)On the basis of the fact that the orthogonal matching pursuit algorithm needs sparsity as a priori information and insufficient reconstruction efficiency,the segmentation weak orthogonal matching pursuit algorithm was used as the reconstruction algorithm for two bioelectric signals.Besides,concerning the problem that the segmentation weak orthogonal matching pursuit algorithm is unstable and the reconstruction accuracy is relatively low,an improved segmentation weak orthogonal matching pursuit algorithm was designed.The simulation results shown that the improved segmented weak orthogonal matching pursuit algorithm has higher reconstruction precision,stable reconstruction effect and maintains the high efficiency of segmentation weak orthogonal matching pursuit algorithm.It can be utilized for ECG and PPG signal compression and reconstruction.(3)The proposed method was used to compress and reconstruct the signal,and the reconstructed ECG and PPG signals were respectively utilized to extract feature values for identification.The results shown that the recognition rate before and after reconstruction is the same,which shows the effectiveness of the proposed method for the compression reconstruction of two bioelectric signals and the practicality of compressed sensing in portable bioelectric signal identification equipment.