臺灣博碩士論文加值系統

目前，車牌辨識系統已有相當的發展，且已在多種應用環境上發揮功能。對於車牌在圖片中的位置，近年來常應用學習型的相關演算法來進行偵測，例如：以Adaboost演算法在大量的Haar-like feature中挑選出關鍵特徵，並以此做為車牌偵測的依據。本論文承襲層級式分類架構 (Cascade classifier) 的優點，進而提出以非線性分類器進行高精確車牌偵測之技術內容。此外，本文也介紹了如何運用Reduced Support Vector Machine (RSVM) 及其相關內容來使非線性分類器的運算量降低。本論文提出之方法，在每層分類器皆能有效排除89% 以上的非車牌樣本，使得進入下個層級的樣本數量大幅減少。對於車牌偵測的成功率，先前研究多以隨機的方式將圖片區分為兩個集合，其中一個進行訓練、另外一個用來測試。而本論文將採用Cross-validation來計算整體的車牌偵測成功率，此作法既能公平地讓每張圖片輪流成為訓練樣本和測試樣本，又能夠有效地評估模型對於未知圖片的偵測成功率。本論文提出之方法，其整體偵測成功率約為98.2±0.33%，對一張640X480圖片的處理時間約123.5毫秒。

License plate recognition (LPR) system has been successfully applied to many applications such as unattended parking, stolen vehicle verification and security control. LPR system is generally composed of three parts: location of the license plate region (LP detection), segmentation of the plate characters and recognition of each character. This work focuses on the LP detection because this part is very crucial in the LPR system. In recent years, learning-based algorithm has been applied widely into the LP detection. For example, train several simple classifiers based on the statistical features first. Then use the AdaBoost learning algorithm to build up the other classifiers based on the Haar-like features. Combining the classifiers using the statistical features and the Haar-like features, a cascade classifier is obtained. This paper proposes a highly accurate license plate detection algorithm using nonlinear classifier based on the architecture of the cascade classifier. For immediateness, thesis paper introduced the Reduced Support Vector Machine (RSVM) to
reduce the cost of training and testing procedures dramatically. License plate detection success rate of 98.2 percent, the processing time of 123.5 ms.

摘要 i
Abstract ii
致謝 iii
目錄 iv
表目錄 vi
圖目錄 vii
第一章 緒論與文獻探討 1
1.1 研究動機與目的 1
1.2 文獻探討 3
1.3 影像灰階化 7
1.4 邊緣偵測 8
1.5 動態閥值選取 10
1.6 論文架構 12
第二章 系統架構 14
2.1 訓練階段 (Training phase) 14
2.2 測試階段 (Testing phase) 16
第三章 資料特徵 17
3.1 第一層分類器之特徵定義 18
3.2 第二層與第三層分類器之特徵定義 23
第四章 SSVM與RSVM 25
4.1 簡介 25
4.2 典型的SVM與SSVM 29
4.3 RSVM 32

第五章 實驗結果與討論 32
第六章 結論與未來方向 46
參考文獻 47

[1]Aizerman, M., Braverman, E. and Rozonoer, L. (1964), “Theoretical foundations of the potential function method in pattern recognition learning,” Automations and Remote Control, 25, 821-837.
[2]Anagnostopoulos, C. E., Anagnostopoulos, I. E., Psoroulas, I. D., Loumos V. and Kayafas E. (2008), “License plate recognition from still images and video sequences: A survey,” IEEE Transactions on Intelligent Transportation System, 9(3), 377-391.
[3]Bai, H. and Liu, C. (2004), “A Hybrid License Plate Extraction Method Based on Edge Statistics and Morphology,” in Proc. of the International Conference on Pattern Recognition, 2, 831-834.
[4]Burges, C. J. C. (1998), “A tutorial on support vector machines for pattern recognition,” Data Mining and Knowledge Discovery, 2, 121-167.
[5]Chen, X. and Yuille, A. L. (2004), “Detecting and Reading Text in Natural Scenes,” in Proc. of the International Conference on Computer Vision and Pattern Recognition, 2, 366-373.
[6]G. H. Ming, A. L. Harvey, and P. Danelutti, “Car number plate detection with edge image improvement,” Proceedings of the 4th International Symposium on Signal Processing and Its Applications, 1996, vol. 2, pp. 597–600.
[7]J.A.G. Nijhuis, M.H. ter Brugge, K.A. Helmholt, J.P.W. Pluim, L. Spaanenburg, R.S. Venema, M.A. Westenberg, “Car license plate recognition with neural networks and fuzzy logic,” IEEE Int. Conf. on Neural Networks, vol. 5, pp. 2232-2236, 1995.
[8]Lee, Y.-J. and Mangasarian, O. L. (2001), “SSVM: A Smooth Support Vector Machine for Classification,” Computational Optimization and Applications, 20, 5-22.
[9]Lee, Y.-J. and Mangasarian, O. L. (2001), “RSVM: Reduced Support Vector Machines,” First SIAM International Conference on Data Mining, Chicago.
[10]Lee, Y.-J., Lo, H.-Y. and Huang, S.-Y. (2003), “Incremental Reduced Support Vector Machines,” International Conference on Informatics, Cybernetics and Systems (ICICS), Kaohsiung, Taiwan.
[11]Lee, Y.-J. and Huang, S.-Y. (2007), “Reduced Support Vector Machines: A Statistical Theory,” IEEE Transactions on Neural Networks, 18(1), 1-13.
[12]Mangasarian, O. L. (2000), “Generalized support vector machines,” Smola, A., Bartlett, P., Schölkopf, B. and Schuurmans, D. eds., Advances in Large Margin Classifiers, Cambridge, MA, MIT Press, 135-146.
[13]Medialab LPR database (2012), Multimedia Technology Laboratory, National Technical University of Athens. http://www.medialab.ntua.gr/research/LPRdatabase.html
[14]Tseng, Y.-C., Lee, Y.-J. (2011), “Rank-r Updating Techniques for Fast Exact Cross-Validation,” Industrial Conference on Data Mining - Poster and Industry Proceedings 2011 (ICDM 2011), New York, USA, 54-67.
[15]Vapnik, V. N. (1995), The Nature of Statistical Learning Theory, New York: Springer-Verlag.
[16]Viola, P. and Jones, M. J. (2004), “Robust Real-Time Face Detection,” International Journal of Computer Vision, 57(2), 137-154.
[17]Zhang, H., Jia, W., He, X. and Wu, Q. (2006), “Learning-based license plate detection using global and local features,” in Proc. 18th ICPR, Hong Kong, 2, 1102-1105.
[18]內政部警政署警政治安全球資訊網，“全般刑案、暴力犯罪及竊盜之發生數”，http://www.npa.gov.tw/NPAGip/wSite/ct?xItem=42040&ctNode=11394&mp=1，2010。
[19]陳正誠、黃景彰、陳英傑、林樹國，“影像辨識技術應用於警政系統-以臺北縣警察局『贓車自動辨識系統』為例,” 第四屆資訊管理暨警政資訊實務研討會，2000。
[20]陳瑞添、吳東衡、吳豐明、鄭勝方、許君光、魏世政、林德興、劉益源，台中縣警察局『機動查緝贓車系統』計畫書，2001。