本論文提出一個多車牌辨識系統的架構設計，在一張影像中搜尋一個以上的車牌正確位置並將車牌上的每個字元獨立切割出來，最後再利用類神經方法辨識出車牌上的正確資訊。

   在本文所提出的方法中，首先我們使用對比增強的前處理，使字元與車牌底色的對比度提高，因而增加定位車牌的準確性。並利用邊緣角度以及型態學的擴張等方法來快速去除複雜背景，並以車牌具有對稱特性找出車牌候選區，將具有車牌字元的候選區定義為車牌區域。然後再使用標示連通物件的方法找出每個字元所形成的斜率，並與水平線作為基準所形成的夾角，視為車牌傾斜的角度，並依照所傾斜的角度扭正車牌，之後將每個標示字元依次分割出來。

   在車牌字元辨識部份，我們利用類神經網路的方法來辨識字元。由於類神經網路是使用大量的神經元來模仿生物神經網路的能力，並且可以透過學習的方式，來解決資料分類的問題，因此可以使用類神經網路的方法來解決辨識車牌字元的問題。除此之外，類神經網路還具有高容錯性，這個特性有利於解決切割出的字元具有雜訊或者影像殘缺不全的問題。

   本研究分別以正面拍攝之單車輛與多車輛影像作為測試，拍攝環境為白天、晚上、晴天與陰天。其中單車輛拍攝了329張照片，而多車輛影像拍攝了141張287面的車牌。此外也對36部機車每部分別以7種不同角度、2種不同距離拍攝14張影像，共504張影像，以評估系統的效能。

In this paper we propose a structure and design of a multiple vehicle license plates recognition system. It can search more than one correct positions of plate in an image and cut out each single word on the plates and then utilize neural network to distinguish out the information on the plates.

  In the proposed method, we use preprocess of contrast enhancement to improve the accuracy of plate location. Then we use the edge angle and morphological method to get rid of the complicated background, and utilize the symmetrical characteristics to find out the plate and define the area including characters as plate locations.

  Then, we use connected-component analysis to find out the slope of plate and rotate the plate according to the angles sloped, later cut out each labeled character sequentially.

  In the recognition, we use the neural network because of the ability that a large number of neurons imitate the neural network of living beings. Besides, neural network has high and getting fault-tolerant, and that is helpful to against the noisy or incomplete cut characters.

  This research is tested with the inputted images conduct of single vehicle and multi vehicles, shot in the evening, fine and cloudy day. And there are 329 single vehicle images and 141 multi vehicles images, totaling 287 plates. In order to test system efficiency, we also add the plates which are shot under on viewable angles.

目錄

第一章 緒論................................1
1.1研究背景................................1
1.2研究動機................................3
1.3研究目的................................5
1.4論文架構................................7

第二章 相關研究探討........................8
2.1車牌定位相關研究........................8
2.2字元切割與辨識相關研究.................13
2.3車牌辨識系統相關研究...................17

第三章 多車牌定位技術.....................20
3.1車牌種類和規格.........................20
3.2系統架構流程...........................22
3.3影像前處理與車牌定位...................23
3.3.1輸入影像灰階處理.....................24
3.3.2對比增強.............................24
3.3.3邊緣偵測.............................28
3.3.4萃取強邊緣點.........................31
3.3.5擴張(Dilation)與區塊化...............32
3.3.6標示連通物件法.......................35
3.3.7車牌候選區...........................37
3.3.8車牌候選區二值化與反向處理...........37
3.3.8.1 二值化............................37
3.3.8.2 反向處理..........................39
3.3.9車牌定位.............................41

第四章 字元切割與辨識.....................44
4.1字元切割...............................44
4.1.1車牌傾斜角之估算.....................45
4.1.2車牌旋轉扭正.........................46
4.1.3字元正規化...........................46
4.2字元辨識...............................49
4.2.1訓練字元取得.........................49
4.2.2 BPNN網路架構........................49
4.2.3訓練神經網路.........................51
4.2.4文字識別.............................55
4.2.5字元再確認...........................57

第五章 系統評估...........................59
5.1系統效能測試...........................59
5.1.1單一車輛實驗結果.....................59
5.1.2多車輛實驗結果.......................60
5.2不同角度與距離拍攝的實驗結果...........65
5.3與其他定位系統比較結果.................71
5.4系統辨識速度的測試.....................73
5.5研究範圍與限制.........................74
5.6系統測試結果範例.......................76
5.7系統測試錯誤範例.......................81

第六章 結論與未來研究方向.................86
6.1研究成果...............................86
6.2未來研究方向...........................88

參考文獻..................................90
附錄A  台閩地區機動車輛登記數.............95
附錄B  邊緣偵測運算元.....................96


                      圖目錄

圖1.1單一車牌輸入影像.....................................4
圖1.2多車牌輸入影像.......................................4
圖1.3 貼有車輛保險標章、排氣檢驗合格標籤與通行證等的車牌..6
圖2.1 H. F. Zhang等人的系統流程...........................8
圖2.2 S. W. PARK等人的系統流程...........................10
圖2.3 H. L. Bai 等人的系統流程...........................12
圖2.4 利用垂直投影法切割字元.............................13
圖2.5 文字的幾何特徵.....................................14
圖2.6 SimNet類神經網路架構圖.............................15
圖2.7 倒傳遞類神經網路架構...............................16
圖3.1 系統架構流程圖.....................................22
圖3.2車牌定位流程圖......................................23
圖3.3四種對比增強的灰階轉換型態..........................26
圖3.4.1在白天的情況下，影像對比增強前後之比較............27
圖3.4.2在夜晚的情況下，影像對比增強前後之比較............28
圖3.5 Sobel運算元........................................29
圖3.6 Sobel垂直運算與邊緣角度計算後的差異................30
圖3.7萃取強烈邊緣點結果..................................31
圖3.8區塊化示意圖........................................33
圖3.9傾斜車牌定位區域補償前後............................34
圖3.10(a)未經過區塊化之特徵白點圖........................34
圖3.10(b)王[21]所提出的區塊化............................34
圖3.10(c)本論文所提出的區塊化............................34
圖3.11(a) 4-neighborhood.................................36
圖3.11(b) 8-neighborhood.................................36
圖3.12門檻值二值化流程圖.................................38
圖3.13(a)50c.c原始車牌影像...............................39
圖3.13(b)50c.c原始車牌影像經過垂直Sobel運算元............39
圖3.13(c) 50c.c原始車牌影像透過求得的門檻值做二值化處理..39
圖3.14(a)90c.c~150c.c機車車牌顏色配置....................40
圖3.14(b)90c.c~150c.c機車車牌經過二值化，字元成為黑色....41
圖3.15為車牌定位的結果...................................43
圖4.1車牌字元切割流程圖..................................44
圖4.2車牌傾斜圖示........................................45
圖4.3 Bilinear Interpolation.............................47
圖4.4(a)車牌扭正前.......................................48
圖4.4(b)車牌扭正後.......................................48
圖4.4(c)車牌字元切割.....................................48
圖4.4(d) 字元正規化前後..................................48
圖4.5倒傳遞神經網路模型..................................51
圖4.6倒傳遞演算法的學習過程..............................52
圖4.7類神經網路辨識車牌字元流程..........................56
圖4.8(a)字元D與0差異的部分...............................58
圖4.8(b)字元8與B差異的部分...............................58
圖5.1車牌上的貼紙造成字元切割錯誤........................62
圖5.2擋泥板的廣告字體造成字元切割錯誤....................63
圖5.3車牌字元斷字的情形..................................63
圖5.4 字元太小而造成字元再次判別時0與D誤判...............64
圖5.5不同角度與距離的相對位置示意圖......................65
圖5.6 字母A各種不同的型態................................66
圖5.7各種不同角度下拍攝(a)~(f)...........................68
圖5.7各種不同角度下拍攝(g)~(l)...........................69
圖5.7各種不同角度下拍攝(m)~(n)...........................70
圖5.8 Y. S. Juan系統定位錯誤範例.........................72
圖5.9本研究系統汽車車牌定位成功範例......................72
圖5.10車牌辨識系統介面...................................73


                   表目錄

表3.1 台灣地區車輛種類及其車牌規格.......................21
表5.1單車輛與多車輛影像，車牌定位結果....................61
表5.2單車輛與多車輛影像，字元切割結果....................61
表5.3單車輛與多車輛影像(測試樣本)，字元辨識率與車牌辨識率的結果.....................................................61
表5.4單車輛與多車輛影像(測試+訓練)，字元辨識率與車牌辨識率的結果...................................................61
表5.5各種不同角度的字元切割率............................66
表5.6各種不同角度的字元辨識率(測試樣本)..................67
表5.7各種不同角度的車牌辨識率(測試樣本)..................67
表5.8 Y. S. Juan[30]車牌定位成功率與本研究定位成功率之比較.......................................................71

[1]	X. F. Hermida, F. M. Rodriguez, and J. L. Fernandez, “A System for the Automatic and Real Time Recognition of V. L. P. (Vehicle License plates),” in proc. of International Conference on Image Analysis and Processing, Florencia, Italia. Sep. 1997.

[2]	P. Comelli, et al., “Optical Recognition of Motor Vehicle License Plates,” IEEE Trans. on Vehicular Technology, vol. 44, no. 4, Nov. 1995, pp.790-799.

[3]	S. Draghici, “A Neural Network Based Artificial Vision System for License Plate Recognition,” International Journal of Neural Systems, vol. 8, no. 1, 1997, pp. 113-126.

[4]	T. Naito, et al., “Robust License-Plate Recognition Method for Passing Vehicles under outside Environment,” IEEE Trans. on Vehicular Technology, vol. 49, no. 6, Nov. 2000, pp. 2309-2319.

[5]	J. A. Hegt, R. J. de la Haye and N. A. Khan, “A High Performance License Plate Recognition System,” in Proc. of IEEE International Conference on Systems, Man, and Cybernetics, vol. 5, Oct. 1998, pp. 4357-4362.

[6]	H. Bai, J. Zhu, and C. Liu., “A Fast License Plate Extraction Method on Complex Background,” in Proc. of IEEE International Conference on Intelligent Transportation Systems, vol. 2, 2003, pp.985-987.

[7]	S. K. Kim, D. W. Kim, and H. J. Kim, “A Recognition of Vehicle License Plate Using a Genetic Algorithm Based Segmentation,” in Proc. of International Conference on Image Processing, vol. 1, Sep. 1996, pp.661-664.

[8]	H. Zhang, W. Jia, X. He, and Q. Wu, “A Fast Algorithm for License Plate Detection in Various Conditions,” in IEEE International Conference on Systems, Man, and Cybernetics, manuscript received March 2006.

[9]	Y. Freund, “An Adaptive Version of the Boost by Majority Algorithm,” Machine Learning, 43(3), 2001, pp. 293-318.

[10]	S. W. Park, S. C. Hwang, and J. W. Park, “The Extraction of Vehicle License Plate Region Using Edge Directional Properties of Wavelet Subband,” in IEICE Trans. on Information & System, vol. E86-D, no. 3, March 2003, pp. 664-669.

[11]	Z. B. Huang, and Y. F. Guo, “Classifier Fusion-Based Vehicle License Plate Detection Algorithm,“ in IEEE International Conference on machine Learning and Cybernetics, vol. 5, Nov. 2003, pp. 2984-2989.

[12]	H. Bai, J. Zhu, and C. Liu, “A Hybrid License Plate Extraction Method Based on Edge Statistics and Morphology,” in Proc. of the 17th International Conference on Pattern Recognition, vol. 2, 2004, pp. 831-834.

[13]	S. Y. Zhang, M. G. Zhang, and X. Z. Ye, “Car Plate Character Extraction under Complicated Environment,” in IEEE International Conference on Systems, Man, and Cybernetics, vol. 5, Oct. 2004, pp. 4722-4726.

[14]	林欣平，“車牌字元萃取，” 國立交通大學，電機與控制工程研究所碩士論文，1999。

[15]	Y. T. Hsu, C. B. Lin, S. C. Mar, and S. F. Su, “High Noise Vehicle Plate Recognition Using Gray System,” Journal of Grey Systems, vol. 10, no. 3, 1998, pp. 193-208.

[16]	馬西聰， “利用灰階關聯度辨識車輛字母的研究，” 國立台灣科技大學，電機工程研究所碩士論文，1997。

[17]	J. R. Cowell, “Syntactic Pattern Recognizer for Vehicle Identification Numbers,” Image and Vision Computing, vol. 13, no. 1, Feb. 1995, pp. 13-19.

[18]	魏銪志， “動態多標的車牌辨識系統之研究，” 私立元智大學，資訊管理研究所碩士論文，2000。

[19]	溫福助， “類神經網路樣板比對法於車牌字元辨識之研究，” 國立台灣大學，電機工程研究所碩士論文，2000。

[20]	M. Yu, and Y. D. Kim, “An Approach to Korean License Plate Recognition Based on Vertical Edge Matching,” Ajou University, Suwon, Korea, 2000.

[21]	王振興， “多標的汽機車車牌辨識系統之研究，” 私立元智大學，資訊管理所碩士論文，2003。

[22]	黃錦溢， “車牌自動辨識系統之設計，” 私立淡江大學，電機工程研究所碩士論文，2000。

[23]	Y. J. Kim, and S. W. Lee, “Off-Line Recognition of Unconstrained Handwritten Digits Using Multilayer Back propagation Neural Network Combined with Genetic Algorithm,” in Proc. 6th Wkshp on Image Processing Understanding, 1994, pp. 962-968.

[24]	S. Knerr, L. Personnaz, and G. Dreyfus, “Handwritten Digit Recognition by Neural Networks with Single–Layer Training,” IEEE Trans. on Neural Networks, vol. 8, no.1, 1997, pp. 43-53.

[25]	H. A. Hegt, R. J. de la Haye, N. A. Khan, “A High Performance License Plate Recognition System,” in IEEE International Conference on Systems, Man, and Cybernetics, vol. 5, 1998, pp.4357-4362.

[26]	M. M. M. Fahmy, “Automatic Number-Plate recognition: Neural Network Approach,” Vehicle Naviqation and Information Systems Conference, Aug. 1994, pp. 99-101.

[27]	Y. Q. Yang, J. Bai, R. L. Tian, and N. Liu, “A Vehicle License Plate Recognition System Based on Fixed Color Collocation,” in IEEE International Conference on machine Learning and Cybernetics, vol. 9, Aug. 2005, pp. 5394-5397.

[28]	蘇木春，張孝德， “機器學習：類神經網路、模糊系統以及基因演算法則，” 全華科技圖書公司，2000.

[29]	葉怡成， “類神經網路模式應用與實作，” 儒林圖書有限公司，2002.

[30]	C. T. Hsieh, Y. S. Juan, K. M. Huang, “Multiple License Plates Detection for Complex Background,” in IEEE International Conference on Advanced Information Networking and Applications, vol. 2, March 2005, pp. 389-392.