A High Recognition Rate of Feature Extraction Algorithm Without Segmentation

To solve the problem of heart sound segmentation with noises, the feature extraction accuracy could further improved. A high recognition rate of feature extraction algorithm without segmentation was proposed. Firstly, wavelet decomposition was carried out, then representative detail and approximate wavelet coefficient were selected. Next, envelope and autocorrelation feature were calculated. Furthermore, nonlinear dimension processing was used on detail and approximate envelope autocorrelation feature. Finally optimization feature set was obtained. Three public heart sound database were used for classification experiment experimental results shown that this algorithm had obvious feasibility, the recognition rate could reach to 100%.