| S1和S2共振峰频率在心音分类识别中的应用 |
| Application of formant frequency of S1 and S2 in classification of heart sounds |
| |
| DOI： |
| 中文关键词: 心音；共振峰频率；特征提取；分类识别；支持向量机 |
| 英文关键词:heart sound;formant frequency;feature extraction;classification and recognition;support vector machine(SVM) |
| 基金项目:国家自然科学基金(61271334,61073115)和江苏省研究生培养创新工程（SJCX17_0229）资助项目 |
| 作者 | 单位 | | 成谢锋 | 南京邮电大学 电子与光学工程学院，江苏南京210023 | | 陈亚敏 | 南京邮电大学 电子与光学工程学院，江苏南京210023 |
|
| 摘要点击次数: 11163 |
| 全文下载次数: 3237 |
| 中文摘要: |
| 针对心音身份识别过程中心音特征提取的难点，提出了一种以第一心音(S1)和第二心音(S2)共振峰频率作为特征的心音分类识别方法。对原始心音通过小波变换进行去噪处理；基于归一化平均香农能量的分段算法对心音信号分段获得S1和S2的时域波形；采用线性预测编码(LPC)的方法分别提取S1和S2的共振峰频率；结合S1和S2共振峰频率构成心音的特征向量，并采用支持向量机(SVM)的分类方法对心音的特征向量进行分类识别。实验结果显示，S1和S2共振峰频率能够很好地表征心音信号的稳定性和唯一性，以S1和S2共振峰频率作为心音特征进行分类识别具有非常高的识别精度，这为基于心音特征的身份识别技术以及心脏疾病诊断方法提供了可靠的理论基础。 |
| 英文摘要: |
| To solve the difficulty of extracting the features of heart sounds in the process of heart sound recognition,a recognition method based on the first heart sound (S1) and the second heart sound (S2) formant frequency is proposed.The original heart sound is denoised by wavelet transform.Then,the waveform is split up into S1 and S2 segments based on the algorithm of normalized average Shannon energy.Based on linear prediction encoding method(LPC),the formant frequency of S1 and S2 is extracted.Finally,a feature vector is constituted by the formant frequency of S1 and S2,and the support vector machine(SVM) is used to classify the feature vectors.The experimental result shows that the formant frequency of S1 and S2 can characterize the stability and the uniqueness.With the formant frequency of S1 and S2 as the feature,there is a very high recognition accuracy in the classification,providing a reliable theoretical basis for the identification of heart sound and diagnosis of heart disease. |
| 查看全文 查看/发表评论 下载PDF阅读器 |