基于小波分解和数据融合方法的ECG身份识别.PDFVIP

22 4 Vol. 22 No. 4 2009 8 SpaceM edicine M edical En ineerin Au . 2009 基于小波分解和数据融合方法的 ECG身份识别* 1 1 1 2 3 杨向林 , 严 洪, 李延军, 魏 莉 , 孙即祥 ( 1. , 100193; 2. 100022; 3. , 410073) : 研究一种新的基于 ECG 的身份识别方法选取 35位健康人 ECG 波形的波形特征小 波特征和融合特征作为特征向 , 通过相关系数阈值法进行 ECG 身份识别 对 35位被试者的另 外 40段数据进行身份识别验证, 基于波形特征小波特征融合特征的身份识别正确率分别为 82. 5% 87. 5%95% 采用小波特征身份识别的正确率优于波形特征的正确率, 而采用融合特征识别的正确率 优于其他两种特征的正确率 实验表明, 基于 ECG 的身份识别技术是可行的, 可以和指纹等多种 生物特征联合使用开发多导生物识别系统, 并且本文所提方法算法简单, 实时性好, 准确度高 : ECG; 身份识别; 波形特征; 小波特征; 融合特征; 相关系数 : R8573; TP391 : A : 2009) H uman Id en tfi ication B ased on ECG w ith WaveletD ecomp osition and D ata F us ion M ethod YANG X ian lin, YAN Hon , L IYanjun, WE I Li, SUN Jix ian . SpaceM edicine M ed icalEn ineerin , 2009, 22( 4): 296 ~ 301 Abstract: Objective A new method is researched based on ECG for human iden tification. M ethod sThree fea tures ofw ave feature, w avelet feature, fusion featurew ere selected as the feature vectors for 35 normal subject ECG s, and correlation coefficient discrmi inate analysis w as used for human identification. Resu lts The other 40 sessions from 35 subjects w ere used for verify in the accuracy of each feature. They w ere 82. 5%87. 5% and 95% respectively for w ave feature, w avelet feature and fusion feature. ECG w avelet feature outperformed ECG w ave feature. W h ileECG fusion feature performed better than either feature. Conc lusionThe verification expermi ent demonstrates that the technolo y of human identification based on ECG is feasib le. ECG can be comb inedw ith other b iometric features such as fin erprint for developin themultib iom etric identification sys tem. The method proposed in th is paper is smi p le, realtmi e and exact for human identification. K ey w ord s: ECG; hum an identification; w ave feature; w avelet feature; fusion feature; correla