臺灣博碩士論文加值系統

本論文旨在實現一套有效且運算快速的車牌自動辨識架構。車牌自動辨識可廣泛使用於停車場管理、自動收費及查緝贓車等生活實務應用上。但由於適用環境的差異性大以及受辨識車輛之條件不易控制，所以要研究出一套能廣泛的應用於各種狀況的辨識系統，正是本論文努力的目標。
在論文中，我們利用車牌之明暗對比與寬高比特性並配合所提之兩段式滑動視窗技巧快速的找出車牌字元之邊界。在影像增強部分，為了消除因照度不均所造成對比不明顯的現象，我們引入局部線性灰階轉換的技巧；此外，在實際的環境中，多數車牌均會面臨污損或退色的現象，而車牌之明暗對比增強雖可減少此一方面的影響，但其效果卻有一定的限度。有鑑於此，本計劃提出一遞迴式Ｋ-mean二元化法將鎖定之字元區域進行二值化處理以利於圖像之切割與辨識。最後，本系統採用了方向性特徵非線性的邊緣強化演算法來萃取特徵，系統經過特徵萃取後，再將此80維向量輸入一倒傳遞類神經網路辨識架構，即可獲得系統最後之車牌辨識結果。
本系統的實驗資料分別於早晨、中午、傍晚等戶外及地下停車場內拍攝，取樣圖樣共600張，無外加任何光源，資料為灰階影像，解析度為 像素。實驗結果顯示，我們所使用的方法具有相當良好的應用效果，且每一圖像的辨識時間約為0.45秒。

The main idea of the thesis is to realize a system of vehicle license plate recognition which is effective and fast-calculating. It can be widely used in the management of parking lot , automatic fee charge and checking on stolen vehicle. Because of the variation on environment is large and the recognized vehicle is not easy to be controlled, so we have to look into a recognition system widely suiting for each state.
In this thesis, we use two steps of sliding window to locate speedily the demarcation of the Vehicle Identification Number (VIN) according to the regulation of license plate, then use the partial linear gray level transformation to enhance the contrast and the recursive K-mean binary method in combination with projection technique to segment the VIN. Finally, the directional features of 80 dimensions are extracted and the back-propagation neural networks is used for obtaining the final solution.
The experimental data of the system is to take a picture outside in the morning, noon, and evening respectively, and also in an underground parking lot. There are 600 pieces of pattern in which the resolution is pixel without any additional illumination. The experimental result shows that our method has great effect in application and the recognition time of each image is about 0.45 second.

第一章 緒論 1
1.1 引言 1
1.2 研究動機 2
1.3 論文架構 3
第二章 影像訊號處理 4
2.1 影像處理的步驟 4
2.2 前級處理 6
2.2.1 區域處理 6
2.2.2 褶合演算法 8
2.2.3 梯度運算子 8
2.2.4 Laplacian 運算子 9
2.3 灰階轉換 11
2.4 臨界值法 15
2.4.1 以K-mean演算法為基礎的臨界值選擇 15
2.4.2 Otsu的二元化方法 17
2.5 表示與描述 19
2.5.1 文字特徵向量演算法簡介 19
2.5.2 常用的文字特徵向量演算法 19
2.5.3 方向性特徵向量探討 23
2.6 識別架構 31
第三章 車牌自動辨識系統的設計 32
3.1 車牌搜尋 33
3.2 影像增強與遞迴式K-mean二元化法 37
3.2.1 影像增強 37
3.2.2 遞迴式K-mean二元化法 38
3.3 字元切割 42
3.4 字元修補 46
3.5 字元辨識 46
第四章 實驗結果 47
4.1 灰階轉換前後之比較 47
4.2 遞迴式K-mean二元化法與Otsu法之比較 48
4.3 系統評估 50
第五章 結論與展望 66
5.1 結論 66
5.2 未來展望 67
參考文獻 68
附錄A 褶合核心 71
附錄B An Effective Method for Optical
Recognition of Vehicle License Plates 77

[1] S. K. Kim, D. W. Kim and H. J. Kim, "A recognition of vehicle license plate using a genetic algorithm based segmentation", Image Processing, IEEE Proceedings International Conference, vol. 2, pp. 661- 664, 1996.
[2] J. R. Cowell, "Syntactic pattern recognizer for vehicle identification numbers", Image and Vision Computing, Vol. 13, no. 1, pp.13-19, 1995.
[3] S. Draghici, "A Neural Network based Artificial Vision System for License Plate Recognition", International Journal of Neural Systems, Vol. 8, No.1, pp.113-126, February 1997.
[4] P. Comelli, P. Ferragina, M. N. Granieri and F. Stabile, "Optical recognition of motor vehicle licence plates," IEEE Trans.
Vehicular Technology, Vol.44, no.4, pp.790-799, November 1995.
[5] N. Otsu, "A Threshold selection method from grey-level histograms," IEEE Trans. Syst., Man Cybern., vol. SMC-8, pp. 62-66, 1978.
[6] O. D. Trier, A. K. Jain and T. Taxt, "Feature extraction methods for character recognition - A Survey", Pattern Recognition, vol.29, no.4, pp.641-662, 1996.
[7] S.B.Cho,"Neural-network classifiers for recognizing totally unconstrained handwritten numerals", IEEE Trans. Neural Network, vol.8, no.1, January 1997.
[8] D.E. Rumelhart, G. E. Hilton, and R. J. Williams, "Learning internal representations by error propagation", Parallel Distributed Processing, vol. 1:Fundations, D. E. Rumelhart and J. L. McClelland, Eds., MIT Press, pp.318-362, 1986.
[9] Y. Y. Tang et al, "Document analysis and understanding: a brief syrvey, " Proc, Of the First International Conference on Document Analysis and Recognition, 1991, PP. 17-31.
[10] Mitsu Yoshimura, Tatsuro Shimizu and Isao Yoshimura, "A Recognition System for Japanese Zip Code Using Arc Features," IEICE Trans. Inf. & Syst., vol E77-D, No.7, July, 1994.
[11] R.C. Gonzalez and R. E. woods, "Digital Image Processing," Addsion-Wesley Publishing Co., Reading, a Mass., 1992.
[12] Y. J. Kim and S. W. Lee, "Off-line recognition of unconstrained handwritten digits using multilayer backpropagation neural network combined with genetic algorithm," (in Korean), in Proc. 6th Wkshp. Image Processing Understanding, PP.186-193, 1994.
[13] S. Knerr, L. Personnaz, and G. Dreyfus, "Handwritten digit recognition by neural networks with single-layer training," IEEE Trans. Neural Networks, vol. 3, PP. 962-968, 1992.
[14] Sung-Bae Cho, "Neural-Network Classifiers for Recognizing Totally Unconstrained Handwritten Numerals," IEEE Trans, on Neural Networks, Vol. 8, NO.1, PP. 43-53, 1997.
[15] Hirokazu ONOUE and Mitsuru SHIONO, "A Licence Number Plate Extraction in a Car Image and Recognition of All Characters Containing KANJI", 日本國,電子情報通信學會論文誌 D-II Vol. J77-D-II No.3 pp.483-492, March ,1994.
[16] 吳孟璁, "車輛牌照自動辨識系統", 淡江大學資訊工程學系, 1998.
[17] 高橋裕子, 橋本昌明, 鹽 昭夫, 石井健一郎, "在變動的環境中將文字穩定抽出之二元化法", 信學技報, PRU91-49, 1991.
[18] Garl G. Looney, "Pattern Recognition Using Neural Networks", Oxford University Press, 1997.
[19] Anil K. Jain and Richard C. Dubes, "Algorithms for Clustering Data", Prentice-Hall, 1988.
[20] Rafael C. Gonzalez and Richard E. Woods, "Digital Image Processing", Addison Wesley, 1992.
[21] Craig A. Lindley, "Practical Image Processing in C", John Wiley and Sons, 1991.
[22] J. R. Parker, "Algorithms for Image Processing and Computer Vision", Wiley, 1997.
[23] 連國珍，數位影像處理，儒林圖書有限公司，民國八十一年。
[24] 林宸生，影像與語音處理數位信號，全華科技圖書股份有限公司，民國八十六年。
[25] 葉怡成，應用類神經網路，儒林圖書有限公司，民國八十六年。