关键词: 心电图,小波变换,特征提取,属性约简,粒计算

Abstract: The selection and extraction of electrocardiogram feature parameter are the base of the analysis of electrocardiogram(ECG).To improve recognition rate of detection algorithm and classification accuracy is the key to automatic analysis technology.Thus,a hybrid algorithm based on wavelet transform(WT) and attribute reduction of granular computing(GC) to detect premature beat signal of electrocardiogram(ECG) was present.At first,12 electrocardiogram feature parameters are chosen based on diagnostic criteria from cardiovascular experts.Then the feature detection algorithm based on wavelet transform is used for feature extraction,and an attribute reduction algorithm based on granular computing is also used for attribute reduction.Finally,the data are put into pattern classification and the result is verified by MIT-BIH database.As the experiment shows,the classification accuracy after reduction is much higher than it before reduction.Therefore,that reasonable selection of feature parameter is an important factor to improve the recognition efficiency was justified in this article.

Key words: ECG,WT,Feature extraction,Attribute reduction,Granular computing