应用主成分分析法的ECG身份识别方法-YXF.docVIP

PAGE PAGE 35 应用主成分分析法的ECG身份识别方法-YXF 中文摘要 摘要：在生物特征识别领域，ECG以用作身份识别必备的普遍性、唯一性、稳定性和可测性以及其他独特的性质，而比其他传统的身份识别方法更具优势，成为身份识别领域一个新的研究方向。目前，此方法已经得到了更多的应用，不仅在医疗领域有重大贡献，也是一种新型的安全监测技术。 若想在ECG身份识别的识别结果上有突破，应该把重点工作放在心电信号特征的提取和识别上。但是，由于心电信号微弱，易受各种噪声干扰，影响所提取的特征的品质，因此，对心电信号进行预处理尤为重要，同时，为了提高识别率，对于特征的提取也是关键工作。 通过比较一些常用的滤波方法，本文将应用小波变换进行滤波，完成对信号的去噪去基线漂移处理。接着，为了去除特征空间的冗余，减少计算量，采用主成分分析法将原始数据转化为由维数较少的“有效”特征成分来表示。 最后，本文采用线性判别分类的方法，将经过主成分分析法提取的ECG信号的特征作为特征向量进行识别，取得了较为满意的识别率，用实验证明了该身份识别技术方法的可行性。图13幅，表1幅，参考文献24篇。 关键词：ECG；小波变换；主成分分析；特征提取；分类识别 ABSTRACT ABSTRACT:In the field of human identification, because of the universality, uniqueness, stability and measurability, ECG signal has become a new research direction. Now, ECG signal has been wildly used, not only has great signification to the medical field , but also a new type of safety monitoring technology. The core of ECG identification is feature extraction, classification, and identification. However, the ECG signal is weak, vulnerable to all kinds of noise, which will decrease the accuracy of feature extraction.Therefore, ECG signal preprocessing, denoising is particularly important, meanwhile, if we want to get breakthroughs in the rate of recognition, feature extraction is crucial for the work. By comparing some filtering methods commonly used, this paper will apply wavelet transform, select the best fit wavelet function to filter the signal. Next, for the sake of redundancy’s removing between features, we take a way of Principal Component Analysis(PCA) to reduce the dimension of the original signal, select the most effective characteristic components. Finally, we use Linear Discriminant Analysis(LDA) to achieve the identification based on ECG signal, and get a satisfying result, prove the feasibility of this method of identification technology. 13 figures, 2 tables, references. KEYWORDS：ECG; Wavelet transform; Principal Component Analysis; Feature extraction; Classification 1引言 1.1研究背景及意义 随着信息化的普及，以及近年来社会经济的快速发展，信息安全特别是个人信息的泄露已经成为目前必须