發(fā)布時(shí)間：2018-05-07 22:09

  本文選題：心音 + 共振峰頻率　； 參考：《南京郵電大學(xué)學(xué)報(bào)(自然科學(xué)版)》2017年05期

【摘要】：針對(duì)心音身份識(shí)別過(guò)程中心音特征提取的難點(diǎn),提出了一種以第一心音(S1)和第二心音(S2)共振峰頻率作為特征的心音分類(lèi)識(shí)別方法。對(duì)原始心音通過(guò)小波變換進(jìn)行去噪處理;基于歸一化平均香農(nóng)能量的分段算法對(duì)心音信號(hào)分段獲得S1和S2的時(shí)域波形;采用線(xiàn)性預(yù)測(cè)編碼(LPC)的方法分別提取S1和S2的共振峰頻率;結(jié)合S1和S2共振峰頻率構(gòu)成心音的特征向量,并采用支持向量機(jī)(SVM)的分類(lèi)方法對(duì)心音的特征向量進(jìn)行分類(lèi)識(shí)別。實(shí)驗(yàn)結(jié)果顯示,S1和S2共振峰頻率能夠很好地表征心音信號(hào)的穩(wěn)定性和唯一性,以S1和S2共振峰頻率作為心音特征進(jìn)行分類(lèi)識(shí)別具有非常高的識(shí)別精度,這為基于心音特征的身份識(shí)別技術(shù)以及心臟疾病診斷方法提供了可靠的理論基礎(chǔ)。
[Abstract]:Aiming at the difficulty of heart sound feature extraction in the process of heart sound identification, this paper presents a method of heart sound classification and recognition with the resonance peak frequency of the first heart tone S1) and the second heart tone S2) as the features. The original heart sound is de-noised by wavelet transform, and the time domain waveforms of S1 and S2 are obtained based on the segmented algorithm of normalized average Shannon energy. The resonance peak frequencies of S1 and S2 are extracted by linear predictive coding (LPC) method, and the eigenvector of heart sound is constructed by combining the frequency of S _ 1 and S _ 2 resonance peaks, and the feature vectors of heart sound are classified and recognized by support vector machine (SVM) classification method. The experimental results show that the frequency of S _ 1 and S _ 2 resonance peaks can well characterize the stability and uniqueness of the heart sound signal, and the recognition accuracy is very high by using the S _ 1 and S _ 2 resonance peak frequencies as the heart sound characteristics. This provides a reliable theoretical basis for the identification of heart sounds and the diagnosis of heart disease.
【作者單位】： 南京郵電大學(xué)電子與光學(xué)工程學(xué)院;
【基金】：國(guó)家自然科學(xué)基金(61271334,61073115) 江蘇省研究生培養(yǎng)創(chuàng)新工程(SJCX17_0229)資助項(xiàng)目
【分類(lèi)號(hào)】：R540.4;TN911.7
，

本文編號(hào)：1858661