Research Of Feature Extraction And Classification Method For ECG Based On Morphology

Electrocardiography (ECG) plays a very important role in medicine and pattern recognition fields. It has a very important significance in the field of automatic medical diagnosis.ECG feature extraction method based on ECG morphology feature by using signal segmentation combining3rd order Bezier curve fitting and a classify method by self-organization mapping network are presented in this thesis. The analysis of morphological characteristics, summarizing of current domestic and overseas ECG research state for ECG pattern recognition are presented in thesis. Then, the morphology feature extraction methods based on Bezier curve for ECG signal are analyzed and classify verification by self-organization mapping network for ECG signals have been done. Totally110000heart beats from MIT-BIH arrhythmia database are used to verify the effect of the pattern recognition and classification. According to experiment results, the effect of feature extraction by Bezier curve fitting and classification combining self-organization mapping network can be improved. The experiment results show that the methods presented in this thesis can raise effectiveness of feature recognition and accuracy of classifier.The author has developed an experimental platform of ECG signal diagnosis system. All the algorithms in this thesis have been verified by experiments and have realistic significance.