Research Of ECG Compression Algorithm Based On Compressed Sampling

With the development of the society living standards, the health problem and aging population has becoming important, followed the demand for the health care. The Body Area Network(BAN) is arose at the moment. The system demands the low power, especially for a long-term monitoring. A lot of energy in BAN is used for the data transmission between the sensor nodes. The theory of compressive sampling can realize the sampling and compression at the same time, this method can reduce the energy consumption effectively in the process of data collection and transmission. The electrophysiological (ECG) is one of the common signals which can used into the WBAN, how to reduce the energy consumption of the sensor and at the same time ensure the accuracy of the ECG signal reconstruction is one of the key issues in WBAN.This paper mainly analyzes the ECG signal compression methods which based on the CS theory at the single-lead and multi-lead. The paper first proposes the algorithm of ECG signal sparse representation which based on K-SVD and segmented block sparse bayesian reconstruction algorithm which based on deterministic matrix. Then for multi-lead ECG the paper proposes the Variational Bayes compression sampling framework, lastly, use the data from the standard ECG database to verity the algorithm. The main works of the paper are listed as fellow:1. The algorithm of the single-lead ECG sparse representation is proposed based on K-SVD. Firstly, to use low frequency coefficient vector of signal as a sample set, select cascade orthogonal basis as the initial dictionary, and secondly through the K-SVD algorithm trained out adaptive redundant dictionary and then completed the ECG sparse representation. Simulation shows that the algorithm has a good sparse approximation performance even though the sample data size decreases by half. For the large data, the algorithm will be able to reduce memory space and the operation time.2. The ECG compression algorithm is proposed which based on deterministic matrix. Firstly, construct a deterministic polynomial matrix which combines with the segmentation block bayesian framework. Secondly, the deterministic polynomial matrix is optimized by QR decomposition. Simulation shows that this matrix can meet the requirements of real-time and high precision, the optimized matrix further improves the algorithm performance.3. The algorithm of the multi-lead ECG compressed sampling frame is proposed which based on Variational Bayes (VB). Firstly, research the theory of the Multiple Measurement Vectors(MMV) and the Variational Bayes; Secondly, construct the mathematical model of VB which based on CS; Lastly, use the data from the standard ECG database to verity the algorithm, experimental results show that the algorithm becomes much more stable and enhances the robustness in noisy.The proposed algorithms effectively reduce the energy consumption in the WBAN, improve the reconstruction precision; correlation analysis of the algorithms also can provide the technical support for the other physiological signals such as the brain electricity and the muscle electricity.