Research On ECG Signal Denoising Based On Optimization Of Informax ICA

Blind source separation is restoration of the source signals from some observation signals when the source signals and hybrid systems are all unknown. In scientific research, engineering application and even in real life, many signals can be observed as a mixture of unknown source signals. Independent Component Analysis has arised under the background of blind sources separation, and gradually been developed into an effective method of signal processing and data analysis. To solve the problem of blind source separation by ICA method, it just need to meet some prior conditions such as statistical independence of source signals, then the source signals which can’t be observed can be recovered from the mixed signals. Therefore it has been widely used in biomedical signal processing, identification and separation of speech signals, image processing and recognition, geographical spatial information processing, and many other fields.ECG signal which is one of the biological electrical signals, is firstly used for clinical medical research and disease detection. It is an important way to understand the heart activity and is an important basis for heart and cardiovascular disease prevention and diagnosis. ECG signals got from the surface of the body, are inevitably affected by the noises such as power frequency interference, baseline drift and EMG wake. To get rid of signal noises like these is very important for subsequent processing such as characteristic wave detection and recognition, pathologic diagnosis, data compression and so on. Therefore it has attracted many attentions of the scholars at home and abroad. ECG signal denoising problem can be viewed as a blind source separation problem. Its mechanism, noise environment and the surface signal acquisition basically meet the assumptions and mathematical model of ICA, so it can be handled by the ICA model. It usually assumes that independent source signals have the same or similar distribution by typical ICA algorithm. It has only use second-order and fourth-order statistics by traditional ICA algorithm in the signal separation process, the dealing way has implicitly assumed that source data meet high symmetrical distribution. But this assumption is not always consistent with actual analysis data, once the implicit assumption can’t be established, it will not be able to separate source signals effectively by traditional ICA algorithm. In view of the defect of traditional ICA method can’t effectively solve the asymmetric distribution of the source signal, this paper has based on the source signal density estimation methods to optimize the traditional ICA separation results.This paper has expounded the basic model and algorithm of ICA, then optimized algorithm based on Informax and put forward an optimization ICA denoising method based on Informax algorithm. After that, it has gave a model to simulate, the results show that these three types of noises have been accurately filtered and ECG signal has been obtained ideal effect.Finally, this paper has made comparation between the optimization ICA denoising method and the commonly used Informax ICA algorithm and FastICA algorithm. Denoising results have carried on contrast data of each separation index of these three algorithms and have shown that the performance of optimization ICA denoising method based on Informax ICA are better...