基于SVM的ECG传感器信号身份识别方法  

ECG sensor signal identification method based on SVM

作　　者：陈曦[1] 陈冠雄 沈海斌[1] 

机构地区：[1]浙江大学超大规模集成电路设计研究所,浙江杭州310027 [2]杭州易和网络有限公司,浙江杭州310012

出　　处：《传感器与微系统》2014年第10期40-42,46,共4页Transducer and Microsystem Technologies

摘　　要：通过心电图(ECG)传感器采集的信号在身份识别中得到了越来越广泛的应用。但小波滤噪结果往往通过主观判断,没有量化指标,滤波效果不理想;同时,对于ECG特征的提取没有考虑心率变化的影响,鲁棒性不佳。针对这2个问题,提出了一种通过信噪比和相关系数衡量预处理结果的办法,并且在特征的提取上只采用QRS波形,避开了易受心率影响的间期特征。最后使用了多种分类识别方法进行测试,得到了小样本下支持向量机(SVM)最适用于ECG识别的结论。ECG signal collected by ECG sensor is widely used in field of identification. Firstly, wavelet de-noising results are often judged subjectively, no quantitative indicators, and filtering effect is not ideal. Secondly, influence of change of heart rate isn＇ t taken into consideration, robustness is poor. In order to solve these two problems, put forward a kind of methods to measure results of pretreatment by SNR and correlation coefficient, and only adopt QRS waveform in feature extraction, avoiding interval feature easily influenced by heart rate. Finally, use a variety of classification and recognition methods for testing~ for small sample, SVM is most suitable for ECG identification.

关 键 词：心电图 小波变换 支持向量机 相关系数 特征提取 分类识别 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置] TP18[自动化与计算机技术—控制科学与工程]