Electrocardiogram Noise Removal And Feature Recognition Algorithm

With the development of computer technology, ECG telemonitoring system has made great progress. The ECG signals conllected by telemonitoring ware actually dynamic ECG, it has more noise which has a new characteristic. For signal denoising is very important to ECG diagnosis, the study of ECG denoising and feature extracting is very necessary. An economical, simple and convenient telemonitoring was designed final.For the non-uniform distribution noise, a denoising method, which was based on singular value decomposition and wavelet transform, was proposed. ECG with non-uniform distribution noise was decomposed into several orthogonal-components by singular value decomposition, and then Each component was denoised by wavelet threshold denoising method, the de-noised ECG was obtained by rebuilding de-noised orthogonal-components at last. The essence of the SVD-WT method is to decompose ECG into different orthogonal spaces by K-L transform; it can reduce the degree of non-uniform noise distribution. Compared the SVD-WT method and other three denoising methods, experimental results show that the SVD-WT method's Euclid Distance is the least and it lead to small signal distortion.For HRV has the nonlinear dynamics features, the fractals analysis based on wavelet transform was be used in feature extracting of HRV. HRV was proved to be one kind of 1/f fractals signal, and the variance of wavelet coefficients at all levels of HRV is a line with fix slope. The slope is the feature value which called r and was extracted in the paper. The results show that the value rises when atria fibrillation begins, while the value falls when atria fibrillation ends. Thus, the method is effective.An ECG telemonitoring system based on low-cost hardware, multi-level framework and the diagnostic processing algorithms was designed; it contributed to the promotion of telemedicine technology.