基于Android平台的心音识别系统的研究

摘要

搭载Android智能手机平台，结合了梅尔倒谱系数和共振峰特征参数，在安卓平台上设计开发出一款集心音采集、心音显示以及心音识别等功能的心音身份识别系统。在该系统中，首先对采集的待测心音信号进行数字化、预加重、分帧、加窗处理，然后提取出心音信号的梅尔倒谱系数特征参数，利用动态时间规整算法将其与数据库模板心音逐一进行模式匹配，并将拥有最小欧氏距离的数据库模板心音作为最终识别结果，最后展示出待测心音和数据库所有模板心音的共振峰特性对比图，定性地反映待测心音和数据库模板心音的差异，给用户直观的判断和感受，进一步认证识别的结果。系统界面简洁，操作方便，识别速度较快，实现了真机运行，并达到了预期效果。%In this paper,MFCC and formant parameters are used for the identification of heart sounds based on Android platform. The heart sound identification system is designed with lots of functions on Android platform,such as the collection,displaying and identifica-tion of heart sounds. In this system,firstly the heart sound signal of collection is conducted for digitalization,pre-emphasis,framing,and windows adding. Then the MFCC characteristics parameters of heart sound is extracted,use DTW algorithm to do pattern matching with heart sound of database template and choose the heart sound which has the minimum Euclidean distance as the final identification result. Finally,the comparison chart about formant features of the collection heart sound and heart sound templates in database will be demonstra-ted which reflect the differences between them qualitatively. It will give users intuitive judgment and experience which will further authen-ticate the result of identification. The system interface is simple and easy to operate. It has a fast speed of identification and is able to run on a real machine,reaching the expected results.