研究生: |
林怡均 |
---|---|
論文名稱: |
以機器學習即時分析筆跡特徵及進行個人身份辨識之研究 A Study of Machine Learning Approach for Analyzing Personal Handwriting Features and Identity Identification |
指導教授: | 區國良 |
口試委員: | |
學位類別: |
碩士 Master |
系所名稱: |
電機資訊學院 - 資訊工程學系 Computer Science |
論文出版年: | 2009 |
畢業學年度: | 97 |
語文別: | 中文 |
中文關鍵詞: | 生物特徵 、生物認證 、筆跡學 、機器學習 |
外文關鍵詞: | biometrics, biometric authentication, graphology, machine learning |
相關次數: | 點閱:3 下載:0 |
分享至: |
查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報 |
所謂的生物特徵是根據生物的型態、構造、生理及生態等特徵,將生物分門別類能夠加以描述。而「生物認證」,簡單的說就是以使用者身上與生俱來的特徵,如眼球、臉型、指紋等,來辨識身分。即便科技發達,取得並分析生物特徵已非難事,但人權與隱私意識高漲的現在,個人特徵更應保密,因此提出不致侵擾個人資料且可達到準確辨識身分的方法更是刻不容緩。
本論文建立一套可以記錄手寫文字並分析文字特徵的系統,討論以手寫文字做為辨識身分之工具之可能性。此系統突破傳統以傳統紙筆作為測試的媒介時,需要訓練評析者先備有分析筆跡特徵的能力。本論文實驗結果顯示,建構文字的方法、時間、工具與一般書寫文字無異,卻可免去傳統式人工分析的冗長時間以及評析者的事前訓練,不但有統一的分析標準,並可減輕評析者的負擔,不夾帶個人主觀因素等優點,可得更有參考價值的結果。
本論文分析研究採用機器學習的技術,使用數學統計的方式分析並判斷使用者的筆跡特徵屬性,並以決策樹判斷使用者身分,令簡單的書寫動作即可達到身分辨識的作用,不僅省時方便,既保有隱私又無需記憶密碼的優點。
The human biometric is an advanced research area for categorizing different people by recognizing human morphology, physiology and ecology. The recent results of biometric authentication researches are successfully employed in the area of identification with the personal unique characteristic, such as iris, faces, and fingerprints. However, some human rights and privacy issues are widely discussed with the development of biometric data analysis and retrieve at the same time. Thus, a whole new method of authentication without violating personal data is on demand.
In this thesis, a system is constructed which is capable of recording hand writing text and analyzing the characteristics. People are writing in the pen-paper based environment without change any writing habits when identifying. An innovative digital handwriting device is used to avoid traditional long-time human analysis and subjective factors. Furthermore, it provides a unified standard to collected graphology data. After writing 95 specified words, the machine learning technology is used for analyzing the writing characteristics and construct personal unique hand-writing model, which is used to identify this person when re-writing another 10 words. The average accuracy rate of the hand-writing model is 86.93%, and the average accuracy rate of users’ identification is 93.33%, meanwhile, only 0.02 second is required for identifying each person. This thesis proposed a hand-writing identify system which is time-saving, high accuracy, without password required, nor privacy problems need to be considered.
1. 靳貴平 and 龐其昌, 紫外指紋檢測的研究. Optics and Precision Engineering, 2003. 11(2): p. 5.
2. 楊國樞 and 林碧峰, 中文筆跡與人格:一項探索性研究. 國立台灣大學心理學系研究報告, 1964. 6: p. 39-62.
3. Zhang, B., S.N. Srihari, and C. Huang, Word image retrieval using binary features, in SPIE. 2003.
4. handwritingcenter.com. HANDWRITING ANALYSIS SOFTWARE 6.3. [cited; Available from: http://www.handwritingcenter.com/software.htm.
5. Huang, Y.-S., Introduction to Biometrics Technology. CCL TECHNICAL, 2001. 96(6.25): p. 65-73.
6. 刑事警察局. 指紋科學小百科. [cited; Available from: http://www.cib.gov.tw/science/science0305.aspx.
7. Galton, F., Finger Prints. 1892.
8. 陳順東, 虹膜辨識系統之研究與實作. 2004, 中央大學: 台灣桃園. p. 69.
9. Flom, L. and A. Safir, Iris recognition system. U.S.Patent, 1987. 4641349.
10. 吉田公一, 筆跡印文鑑定. 內政部警政署刑事警察局刑事鑑識中心, 2003.
11. 楨田仁, 看筆跡在說話. 2001.
12. 張星海. 筆跡分析真能洞察性格?. 2004 [cited; Available from: http://www.people.com.cn/BIG5/keji/1058/2978300.html.
13. Brewer, J.F., Graphology. Complementary Therapies in Nursing and Midwifery, 1999. 5(1): p. 6-14.
14. Tett, R.P. and C.A. Palmer, The validity of handwriting elements in relation to self-report personality trait measures. Pwson. indkirl. Oiff, 1997. 22(1): p. 11-18.
15. Read, J.C., A study of the usability of handwriting recognition for text entry by children. science direct, 2007(Interacting with Computers): p. 13.
16. Bradley, N., Users of Graphology. Graphology, the Journal of the British Academy of Graphology, 2005(69): p. 55-57.
17. Ward, The Graphologist. Graphology enhances edcuation, 1985. 1.10(3): p. 10-17.
18. 許耀東, 黃英軒, and 林宜樺, 結合類神經網路與基因演算法應用於手寫簽名身份辨識系統, in 南臺灣資訊科技與應用研討會. 2007. p. 110-115.
19. 王建平 and 王曉雪, 基於字型特徵的手寫體漢字多分類識別研究. 計算機應用, 2007. 27(12): p. 5.
20. 王正群, et al., 基於筆畫平面和模糊隸屬度的手寫體漢字識別. 揚州大學學報, 2008. 11(4): p. 5.
21. 陳孟琪, 國民中學情緒障礙學生書寫筆跡特徵之研究. 中華民國特殊教育學會年刊, 2005. 94: p. 75-101.
22. 馮文堯, 刑事警察科學知識全書. 1948, 上海.
23. Turing, A.M., Computing Machinery and Intelligence. Mind. 58: p. 433-460.
24. Friedman-Hill, E., Jess in Action Rule-Based Systems in Java. 2003: Manning Publications Co. All rights reserved.
25. Ian H. Witten and E. Frank, Data Mining: Practical Machine Learning Tools and Techniques (Second Edition). 2005.
26. Alpaydin, E., Introduction to Machine Learning. 2004.