项目名称: 心音特征提取和身份识别中新方法研究及应用
项目编号: No.61271334
项目类型: 面上项目
立项/批准年度: 2013
项目学科: 无线电电子学、电信技术
项目作者: 成谢锋
作者单位: 南京邮电大学
项目金额: 75万元
中文摘要: 本项目针对心音生物特征识别的基本原理和关键技术开展创新研究,提出一种以实时性和高识别率为目标的心音生物识别新方法。首先研究独立子波函数码快速提取算法、心音子波族(HSWF)构造方法和基于HSWF、内心音震动特征、PCG形变特征的三种心音合成模型等基础理论,以构成一种心音特征识别的基本理论架构,并据此基本理论进行心音身份识别系统的设计以充分描述心音信号在时域和频域上产生差异的原因,解决由于心音生理机制的特殊性所造成的特征参数存在变异问题;其次研究能体现心音细节特性的表征形式,通过引入心音特征向量分布相图、周期-功率-频率3D图等特征参数表征形式,拓展生物特征参数提取与形象表示的方法;还研究心音信号采集装置,智能预处理技术和建立高效的鲁棒的心音分类器;最终使实际心音身份识别率达到98%以上。本研究成果将丰富生物信息识别的新内容,并为研制具有自主知识产权的新型智能心音处理设备提供理论和实验依据。
中文关键词: 心音;身份识别;特征表征;识别算法;应用研究
英文摘要: In this program, we carry on the research of the basic principles and the key technology on heart sound identi?cation,and propose a new real-time method of heart sound identity recognition for getting better recognition performance. First, we further study the basic theories: the fast extraction method of independent sub-band function codes, the construction method of heart sounds wavelet family(HSWF), the synthetic model of heart sounds based on the characteristics such as HSWF, PCG deformation, and heart vibration, then constitute the basic theory framework for heart sound identification . According to the basic theory ,we design the identification system can fully describe the characteristics of heart sounds in the time domain and frequency domain to solve the problems caused by the heart sounds characteristic variation. Second, we study the other representations can reflect the characteristics of heart sounds. For instance, we have proposed the representation forms namely distribution phase diagram of heart sound feature vector, and cycle-power-frequency graph. The study can expand the area of biological characteristic parameters extraction and the representation method. Then,we focus on the research of heart sounds extraction device in the complex environment, the heart sound signals intelligence pretrea
英文关键词: heart sound;identity recognition;surface representation;recognition algorithm;applied research