ECG Compression Based On The Empirical Mode Decomposition

Wearable physiological detection equipment is playing an important role in remote ECG monitoring system,while the portability,transportable,and real-time monitoring characteristics of the equipment also brings great challenge to the data wireless transmission and the storage capacity.ECG data compression technology is essential to solve these problems.Aiming at the overwhelmingly dependence on basic functions of the classical compression method based on the transform domain,a new ECG compression method based on the combination of the empirical mode decomposition(EMD)and the wavelet transform was presented in this paper.The main research contents and the innovation points of this paper were as follows:1.Analyzed the advantages and disadvantages of wavelet transform in ECG compression,and the applying of the adaptive of the empirical mode decomposition and the characteristics of the intrinsic mode functions in signal compression.2.Combining the related knowledge of the wavelet theory and the intrinsic mode function,this study applied both the wavelet transform and the empirical mode decomposition to ECG data compression.ECG data was first decomposed into a set of intrinsic mode functions by the empirical mode decomposition algorithm,and these intrinsic mode functions were divided into two groups according to their data characteristics,and then remixed into two mixed functions.The first mixed function could be completely described by its extrema,while the second mixed function was further decomposed by the wavelet transform.3.By analyzing the energy contained in each mixed frunction,different degrees of distortion were assigned to the two functions.The threshold of the first mixed function was determined by the multiple of the maximum extreme value,and the method of fixed percentage of wavelet coefficients to be zeroed was used to determine the threshold of the second mixed function.4.A uniform scalar dead zone quantizer was used to quantize the feature points to be coded.The quantizer put the threshold processing and the quantization together,associating the quantization error with the value of the threshold,so that we could adjust the retrieved quality by only changing the value of the threshold.Then we divided the non-zero feature points and their locations through different ways according to the characteristics of the feature points after quantization of each mixed function.At last the Huffman coding was used to both the amplitudes and the locations of the feature points.This paper employed the MIT-BIH ECG arrhythmia database provided by Massachusetts Institute of Technology to test the performance of the compression algorithm.Experimental results showed that the proposed method exhibited competitive performance compared to other compressors for ECG compression.