臺灣博碩士論文加值系統

本研究提出一個基於生理訊號的專注辨識系統，探討使用光體積變化描記圖(Photoplethysmography, PPG)辨識專注程度的方法。分別探討三種不同專注程度的組合方式，第一種組合分成專注、次等專注、不專注三個程度；第二種組合將專注與次等專注合併，分成兩個程度；第三種組合直接以專注與不專注作為兩個程度。
本研究的系統分成為生理訊號擷取、特徵擷取、特徵選取、特徵正規化與分類五個部分。在生理訊號的擷取，共有10受測者，分別為8名男性與2名女性，專注是利用改良過後視覺的持續性的專注測驗CPT(Continuous Performance Test)來誘發，並記錄受測者的回饋。特徵擷取方面，從每60秒光體積變化描記圖的訊號擷取出4大類的特徵，為了減少個體差異的影響，再將這4大類的特徵與基線作運算擷取個體差異特徵。計算出特徵後利用正規化將特徵規範在同一範圍內。特徵挑選方面，利用基因演算法(GA)找出能夠使辨識率提高的特徵集。分類時則是採用倒傳遞類神經網路與支持向量機兩種分類器，驗證方法採用留一驗證法，並且視資料量的大小考慮資料不對等的情況。使用倒傳遞類神經網路分辨第一種組合的正確率為45%，第二種組合的正確率為89%，第三種組合的正確率為65.84%。無挑選時，使用支持向量機分辨第一種組合的正確率為45.6%，第二種組合的正確率為74.07%，第三種組合的正確率為67.5%；加入GA後，能使三種組合的正確率分別提升為45.6%、80.7%、80%。由此可見使用個體差異特徵後再經過特徵挑選可有效提升辨識率。
最後，在Android平台上實現利用光體積變化描記圖即時辨識專注的系統，包含了感測器Pulse Sensor，微處理機Arduino Nano，並利用OTG傳輸至手機上分辨專注與不專注，其模擬的正確率為83.33%。

In this research, we proposed an attention recognition system based on physiological signals. We used Photoplethysmography (PPG) to recognize different attention level so three attention level combination settings were analyzed. The first combination included attention, minor attention, and non-attention. The second combination included two levels, one was non-attention and the other was the combination of attention and minor attention. The third combination contained attention and non-attention levels.
In our study, the attention recognition system was composed of data acquisition, features calculation, features normalization, features selection, and classification. First, in the data acquisition part, 10 subjects, including 8 males and 2 females participated in this study. In order to induce attention, we used visual test called the Continuous Performance Test (CPT), and recorded the participants’ responses. Second, in the data acquisition part, we calculated 4 categories of features from every 60 seconds of PPG signals segments. In order to reduce the influence of individual difference, we calculated the difference features from the four categories of features from the induced and baseline PPG signals. Third, we normalized our feature set to the same level. Fourth, in order to increase the recognition rate, we used Genetic Algorithm (GA) to select the most effective feature set. Finally, we used BPNN and SVM to classify attention by using leave-one-out cross validation. The problem of unequal sample numbers in each class was discussed. By using BPNN, the accuracy of the first combination achieved 45%. The accuracy of the second combination achieved 89%. The accuracy of the third combination achieved 89%. By using SVM, the accuracy of the first combination achieved 45.6%. The accuracy of the second combination achieved 74.07%. The accuracy of the third combination achieved 67.5%. With feature selection using GA, the accuracies of the three combination rose to 45.6%, 80.7% and 80% respectively. The results demonstrated the capability of using individual differences features and GA feature selector to promote the recognition rate.
As on the android platform, we implemented a real-time attention recognition system base on PPG. This system included a Pulse Sensor, an Arduino Nano micro-controller and transferred to the cellphone by OTG. The system can distinguish between attention and non-attention, with an accuracy of 83.33%.

致謝 I
摘要 II
Abstract III
目錄 V
圖目錄 X
表目錄 XIII
第一章 緒論 1
1.1前言 1
1.2研究動機 1
1.3相關文獻回顧 2
1.4研究目標 2
1.5論文架構 2
第二章 研究背景 4
2.1專注介紹 4
2.1.1專注的定義 4
2.1.2專注的分類 4
2.2專注辨識系統的類別 5
2.2.1使用者相依的專注辨識系統 5
2.2.2使用者獨立的專注辨識系統 5
2.3生理訊號與專注的關係 5
2.3.1自律神經系統(Autonomic Nerve System, ANS) 5
2.3.2心率變異分析(Heart Rate Variability, HRV) 6
2.4生理訊號專注辨識系統 7
2.4.1專注辨識系統系統簡介 7
2.4.2生理訊號介紹 8
第三章 研究方法 11
3.1生理訊號擷取 12
3.1.1量測生理訊號的方法 12
3.1.2生理訊號擷取 12
3.1.3實驗流程 13
3.1.4 CPT實驗 16
3.2 PPG擷取HRV特徵之可行性 18
3.3特徵擷取 19
3.3.1統計特徵 20
3.3.2頻域特徵 21
3.3.3非線性特徵 22
3.3.4 Poincare Plot特徵 23
3.3.5個體差異特徵 26
3.4特徵正規化 27
3.5特徵選取 28
3.5.1基因演算法 28
3.6分類 30
3.6.1支持向量機 30
3.6.2倒傳遞類神經網路 33
3.6.3資料量不對等 35
3.6.4交叉驗證 36
第四章 Android系統架構 37
4.1作業系統Android簡介 38
4.1.1 Android的優勢 38
4.1.2 Android的基本設置 38
4.1.3 Android版本介紹 39
4.2光體積變化描記圖硬體簡介 42
4.2.1 Pulse Sensor簡介 42
4.2.2 Pulse Sensor硬體規格與架構 42
4.2.3 Pulse Sensor的可行性 43
4.3OTG傳輸 46
4.3.1OTG簡介 46
4.3.2OTG架構 46
4.4Arduino硬體平台 47
4.4.1Arduino Nano 47
4.5Android平台 49
4.5.1手機接收訊號 49
4.5.2濾波器設計 50
4.5.3 P點偵測 52
4.5.4Android系統軟體介面 53
第五章 實驗結果與討論 59
5.1分四類結果與討論 60
5.2分三類結果與討論 61
5.2.1使用倒傳遞類神經網路分三類 61
5.2.2使用支持向量機分三類 63
5.3三類合併為兩類結果與討論 70
5.3.1實驗評估指標 70
5.3.2使用倒傳遞類神經網路分類結果 71
5.3.3使用支持向量機分類結果 75
5.4兩類結果與討論 83
5.4.1使用倒傳遞類神經網路分兩類結果 83
5.4.2使用支持向量機分兩類結果 85
5.5特徵討論 91
5.6在Android手機上的結果 92
5.7相關文獻比較 93
第六章 結論與未來展望 95
6.1結論 95
6.2未來展望 97
參考文獻 98
附錄 101

[1] 科技新報，http://technews.tw/
[2] S. M. Yang, C. M. Chen, C. M. Yu, “Assessing the Attention Levels of Students by Using a Novel Attention Aware System based on Brainwave Signals” International Congress on Advanced Applied Informatics(IIAI-AAI), pp. 379 – 384, 2015.
[3] N. Lutsyuk, E. Eismont, V. Pavlenko, “Correlation of the characteristics of EEG potentials with the indices of attention in 12-to 13-year-old children,” Neurophysiology, vol. 38, no. 3, pp. 209-216, 2006.
[4] C. Y. Chen, C. J. Wang, E. L. Chen, C. K. Wu, Y. K. Yang, J. S. Wang, P. C. Chung, “Detecting Sustained Attention during Congitive Work using Heart Rate Variability,” International Conference on Intelligent Hiding and Multimedia Signal Processing, 2010
[5] M. H. Shah, S. A. Kazmi, K. A. Sidek, S. Khan and F. Z. Iqbal, “Photoplethysmographic based heart rate variability for different physiological conditions, ” Research and Development (SCOReD),2014 IEEE Student Conference on, pp.1-6, 16-17 Dec. 2014.
[6] 有關專注力，http://uniedutw.blogspot.tw/
[7] 數位學習注意力對學習成效影響之研究，http://www.nacs.gov.tw/
[8] 廖繼薇，“情境教學式運動遊戲對幼兒專注力之影響”，國立體育大學體育學院，104年
[9] R. W. Picard, E. Vyzas, J. Healey, "Toward Machine Emotional Intelligence: Analysis of Affective Physiological State", IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 23, No. 10, October 2001.
[10] P. J. Lang, "Behavioral Treatment and Bio-Behavioral Assessment: Computer Appliances", In J. H. Johnson J. B. Sidowski, & T. A. Williams (Eds.), Technology in Mental Health Care Delivery Systems, 1980, Page(s): 119-137.
[11] 心律變動性分析，http://www.tma.tw/
[12] 憂鬱、焦慮與自律神經失調，http://wwwu.tsgh.ndmctsgh.edu.tw/
[13] A. B. Hertzman, and C. R. Spielman, “Observations on the finger volume pulse recorded photoelectrically”, Am. J. Physiol., 1937, vol.119, pp.334-5.
[14] B. Khanoka, Y. Slovik, D. Landau, M. Nitzan, “Sympathetically Induced
Spontaneous Fluctuations of the Photoplethysmographic Signal”, Medical &
Biological Engineering & Computing, 2004, 42(1):80-85.
[15] J. G. Webster, “Design of Pulse Oximeters”，IOP 1997.
[16] Conners Continuous Performance Test 3rd Edition，http://www.mhs.com/
[17] 認知功能的評估-注意力，http://openchangjk.blogspot.tw/
[18] 戴欣浩，“基於短時間多生理訊號辨識情緒的特徵選取與特徵萃取方法研究”，國立中正大學電機所，104年
[19] H. William, A. Saul, T. William, P. Brian, “The Art of Scientific Computing Second Edition” ,Page(s): 113-116.
[20] I. Yalcin, K. Mehmet, “Combining classical HRV indices with wavelet entropy measures improves to performance in diagnosing congestive heart failure”, Computers in Biology and Medicine 37, 2007, Page(s): 1502-1510.
[21] S. M. Pincus, “Heart rate control in normal and aborted-SIDS infants”, American Journal of Physiology, Vol. 33, 1991, Page(s): 638-646.
[22] A. Metin, “Approximate Entropy and Its Application in Biosignal Analysis “, Nonlinear Biomedical Signal Processing：Dynamic Analysis and Modeling, IEEE Press, 2001, Vol. 2, Page(s): 72-91.
[23] M. Brennan, M. Palaniswami, P. Kamen,“Do Existing Measures of Poincare Plot Geometry Reflect Nonlinear Features of Heart Rate Variability? “, IEEE Transactions on Biomedical Engineering, 2001, Vol. 48, No. 11, Page(s): 1342-1346.
[24] J. Holland, “Adaptation in Natural an Artificial System,” University of Michigan Press, 1975.
[25] V. Vapnik, “Statistical Learning Theory”, New York: Wiley, 1998.
[26] N. Cristianini, J. S. Tayloy“An Introduction to Support Vector Machines and Other Kernel-based Learning Methods”, Cambridge University Press, 2000.
[27] C. C. Chang, C. J. Lin, “LIBSVM -- A Library for Support Vector Machines”, http://www.csie.ntu.edu.tw/~cjlin/libsvm/
[28] C.-W. Hsu, C.-C. Chang, C.-J. Lin, “A Practical Guide to Support Vector Classification”, Department of Computer Science, National Taiwan University, April 2010.
[29] M. T. Hagan, H. B. Demuth and M. H. Beale, “Neural Network Design”, First
published , 1996, PWS Publishing Company.
[30] Hopfield神經網路，https://zh.wikipedia.org/
[31] S. Haykin, “Neural Networks:A Comprehensive Foundation,1999, Prentice Hall Publishing Company.
[32] 蔡博全，一個使用智慧型手機的即時心肌缺血事件偵測系統”，中正大學電機所，104年
[33] 羅華強，“類神經網路：MATLAB的應用”，高立圖書出版
[34] Pulse Sensor，http://pulsesensor.com/
[35] USB On-The-Go，https://zh.wikipedia.org/
[36] Arduino Nano，https://www.arduino.cc/