臺灣博碩士論文加值系統

傳統個人身份辨識的方法，大多是利用有照證件或是密碼，由於傳統個人身分辨識只辨物不辨人，系統安全性較低。為了提升安全性，故近幾年利用生物特徵作為個人身分辨識。生物特徵可以分為兩類：一類為生理學的特徵，例如指紋、虹膜、心電圖等…。另一類為行為模式的特徵，例如筆跡、聲紋…。每個人的生物特徵都是獨一無二的。
本篇論文提出在身體內部的生理特徵─心電圖。心電圖是心臟收縮時所產生的心臟電氣活動的一種記錄，利用電極置於身體特定部位，依特定組合所產生十二種電壓記錄，即為十二導極心電圖。心電圖具有快速、非侵入式的優點。每個人的心臟大小、肌肉強度、形狀、位置都不太相同，所以利用心電圖作為個人的身份辨識。
利用小波分解將心電圖分解至各個不同的頻帶，找出隱藏在心電圖中不同個體間的特徵，將找到的特徵利用K-NN分類器去分類。另外，小波轉換具有抗雜訊的能力，所以我們加入了不同訊雜比(SNR)大小的雜訊，考驗系統的抗雜訊能力。
使用各個頻帶之變異數以及R-R間隔做為特徵，每段心電圖長度為14個QRS複合波時，其辨識率達到98%左右。考慮雜訊的影響在SNR=10dB以上，辨識率依然達九成五左右。
採用小波轉換之特徵值可以代替基準點之特徵，再加上R-R間隔更可以得到更好的效果。

Traditional individual identification usually utilizes ID card or password. These methods may be suitable for identifying items but not for individuals, and the system security is low. To enhance security level, method based on biometric have been proposed for individual identification in recent years. Biometric can be divided into two categories, namely physical biometrics such as finger prints, iris and ECG and behavioral biometrics such as hand writing, voice print. Everyone has unique biometrics.
In this research, we propose to study the biometric within human body- electrocardiogram (ECG). The ECG is the record of heart electrical activity produced when the heart functions. Put the electrodes at the particular positions of the body, 12 kinds of voltage signals can be recorded that are the 12-lead ECG. The advantages of ECG are its fast and non-invasive. Each person has different heart size, heart muscle strength, shape and position, so we use ECG for individual identification.
We utilize wavelet decomposition to decompose the ECG into different bands, find out the features hidden in the ECG and use K-NN for classification, Furthermore , wavelet transform have de-noiseing ability, so we add different signal-to-noise ratio (SNR) noise, for testing the de-noise ability of the system.
Using the variance of each band and the R-R interval are used as features, high accuracy can be reached, when using ECG samples with 14 QRS complexes in each sample. Even with high level of noise interference (SNR=10dB), high accuracy of 95% can still be retained
The results demonstrate that the feature based on wavelet transform can replace the feature of the fiducial point. R-R interval can improve the performance of the wavelet transform-based features.

摘要 i
Abstract ii
圖目錄 v
表目錄 viii
第一章 簡介 1
1.1 研究背景與動機 1
1.2 研究目的 2
1.3 全文架構 2
第二章 心電圖原理 3
2.1 心臟簡述 3
2.2 心跳與電之關係 3
2.3 心臟的傳導系統 3
2.4 十二導極心電圖 4
2.4.1 . 標準肢導極(Standard limb leads) 5
2.1.2 . 增大導極(Augmented leads) 6
2.1.3 . 胸導極(Chest leads) 7
第三章 小波轉換 9
3.1 小波轉換(Wavelet Transform)緣起 9
3.2 小波分析基本理論 9
3.3 離散小波轉換(Discrete wavelet transform, DWT) 11
3.4 小波族(Wavelet Families) 13
3.4.1、 Haar 小波，(haar) 13
3.4.2、 Daubechies 小波，(dbN) 14
3.4.3、 Coiflets小波，(coifN) 15
3.4.4、 Symlets小波，(symN) 15
3.4.5、 Morlet小波，(morl) 16
3.4.6、 Mexican hat小波，(mexh) 16
3.4.7、 Meyer小波，(meyer) 17
3.5 多層分解(Multiple-Level Decomposition) 19
3.6 小波重建(Wavelet Reconstruction) 20
第四章 論文提出之方法 22
4.1 資料庫 23
4.1.1、 MIT-BIH資料庫 23
4.1.2、 自己所量測之心電圖 24
4.2 前處理 26
4.2.1、 測試資料之擷取 27
4.2.2、 去雜訊 27
4.3 找出具有代表性之QRS複合波 28
4.4 特徵擷取 29
4.4.1、 平均的QRS 複合波特徵 29
4.4.2、 連續心電圖訊號之特徵 32
4.5 分類器(K-NN) 34
第五章 實驗設計與結果 36
5.1 實驗環境 36
5.2 實驗設定與結果 36
5.2.1、 資料與特徵的選擇(MIT-BIH資料庫) 36
5.2.2、 資料與特徵的選擇結果討論(MIT-BIH資料庫) 38
5.2.3、 雜訊對系統的影響(MIT-BIH資料庫) 40
5.2.4、 資料與特徵的選擇(自己量測之心電圖) 57
5.2.5、 資料與特徵的選擇結果討論(自己量測之心電圖) 59
5.2.6、 雜訊對系統的影響(自己量測之心電圖) 61
5.2.7、 不同時間、電極位置對系統的影響(自己量測之心電圖) 68
5.2.8、 資料庫以外的人的辨識能力 69
5.3 相關文獻比較 71
第六章 討論與結論 73
參考文獻 75

[1]A. K. Jain, H. Lin, S. Pankanti et al., “An identity-authentication system using fingerprints,” Proceedings of the IEEE, vol. 85, no. 9, pp. 1365-1388, 1997.
[2]S. Pankanti, S. Prabhakar, and A. K. Jain, "On the individuality fingerprints." pp. I-805-I-812 vol.1.
[3]M. K. Khan, J. Zhang, and S. J. Horng, "An effective iris recognition system for identification of humans." pp. 114-117.
[4]S. Pankanti, R. M. Bolle, and A. Jain, “Biometrics:The future of identification,” Computer, vol. 33, no. 2, pp. 46-49, 2000.
[5]T. W. Shen, W. J. Tompkins, and Y. H. Hu, "One-lead ECG for identity verification." pp. 62-63 vol.1.
[6]S. A. Israel, W. T. Scruggs, W. J. Worek et al., "Fusing face and ECG for personal identification." pp. 226-231.
[7]S. A. Israel, J. M. Irvine, A. Cheng et al., “ECG to identify individuals,” Pattern Recognition, vol. 38, no. 1, pp. 133-142, 2005.
[8]K. Kyeong-Seop, Y. Tae-Ho, L. Jeong-Whan et al., "A Robust Human Identification by Normalized Time-Domain Features of Electrocardiogram." pp. 1114-1117.
[9]K. N. Plataniotis, D. Hatzinakos, and J. K. M. Lee, "ECG Biometric Recognition Without Fiducial Detection." pp. 1-6.
[10]W. Yongjin, K. N. Plataniotis, and D. Hatzinakos, "Integrating Analytic and Appearance Attributes for Human Identification from ECG Signals." pp. 1-6.
[11]C. Y. Suen, K. Jinho, K. Kyekyung et al., "Handwriting recognition-the last frontiers." pp. 1-10 vol.4.
[12]J. P. Campbell, Jr., “Speaker recognition: a tutorial,” Proceedings of the IEEE, vol. 85, no. 9, pp. 1437-1462, 1997.
[13]"k-nearest neighbor algorithm," http://en.wikipedia.org/wiki/K-nearest_neighbor_algorithm.
[14]周嘉陞, 心電圖學必備: 合記圖書出版社, 2001.
[15]邱豔芬, 簡易心電圖讀本: 華杏出版股份有限公司, 1998.
[16]J. G., and Webster, Medical Instrumentation: Application and Design.: John Wiley & Sons, Inc., 1998.
[17]A. Grossmann, and J. Morlet, “Decomposition of Hardy Functions into Square Integrable Wavelets of Constant Shape,” SIAM Journal on Mathematical Analysis, vol. 15, pp. 14, 1984.
[18]"Heisenberg Uncertainty Principle," http://www.mtmi.vu.lt/pfk/funkc_dariniai/quant_mech/uncertainty_principle.htm.
[19]"Fourier transform," http://en.wikipedia.org/wiki/Fourier_transform.
[20]D. Gabor, “Theory of communications,” J. Inst. Elect. Eng., vol. 93, 1943.
[21]S. Mallat, A Wavelet Tour of Signal Processing, 1999.
[22]S. G. Mallat, “A theory for multiresolution signal decomposition: the wavelet representation,” Pattern Analysis and Machine Intelligence, IEEE Transactions on, vol. 11, no. 7, pp. 674-693, 1989.
[23]M. J. Shensa, “The discrete wavelet transform: wedding the a trous and mallat algorithms,” IEEE TRANSACTIONS ON SIGNAL PROCESSING, vol. 40, pp. 19, 1992.
[24]"Physionet," http://www.physionet.org.
[25]"BIOPAC," http://www.biopac.com/.
[26]"National Instruments," http://www.ni.com/.
[27]C. Ying-Hsiang, and Y. Sung-Nien, “Comparison of different wavelet subband features in the classification of ECG beats using probabilistic neural network,” in Engineering in Medicine and Biology Society, New York City, USA, 2006, pp. 4.
[28]"The MIT-BIH Noise Stress Test Database," http://www.physionet.org/physiobank/database/nstdb/.