車牌辨識系統在近幾年無論在學術或商業上被廣泛的研究與發展，車牌辨識系統主要有三個階段：1. 車牌偵測、2. 字元分割、3. 字元辨識。車牌偵測會使用做以下的影像處理：灰階化、二值化、侵蝕膨脹、連通法等，這些影像處理的各個像素點都有關係，因此會非常頻繁的存取影像資料，和大量的數學運算。在FPGA上資料的傳遞只需要儲存在暫存器，不需存回記憶體，而且可以合成大量的運算單元，因此本論文利用FPGA將影像處理管線化，不管是讀取或是儲存記憶體的次數大量減少，加快整體的系統的速度。
車牌辨識系統的車牌定位通常是以灰階加上二值化，再以邊緣偵測或是尋找物件的方法來找尋車牌，但是灰階會將大量的資料捨去，台灣的車牌並非只有白色底，也有綠色、紅色和黃色，所以要是尋找其他顏色底的車牌會非常困難，因此本論文利用HSI(Hue Saturation Intensity)的方法利用色調來做二值化的閥值，使得二值化過後能精確的尋找到車牌。
車牌辨識系統的字元辨識，目前的方法有：類神經網路法、結構法、樣板比對法還有定點取樣法等。本論文比較過這些方法，就以類神經網路法用FPGA實做出來精準度比高且可以結構化設計，因為在FPGA上實現，可以讓系統同時比對10個數字和26個英文字母，以加快字元辨識這個過程。此系統在汽機車車牌定位成功率為98.18%，字元切割成功率為96.27%，辨識率為97.09%。

License plate recognition system has been widely researched and developed in the academic sphere and commercial area. The license plate recognition consists of three parts including license plate detection, character segmentation, and character recognition. License plate position uses image processing techniques which include grayscale, thresholding, mathematical morphology, and monnected-component labeling. These image processing have to accessed the neighboring pixels frequently to achieve intended mathematical formulas. To reduce the image data transfers, recently accessed pixels are stored in the register on FPGA. This leads to consume a large number of FPGA memory and logic resource in synthesizing our design, but image processing of proposed pipeline architecture on FPGA drastically decreases the amount of times of read and write to memory. The hardware based image processing system shows great speed up in the license plate recognition .
The proposed license plate detection applies Grayscale and Threshold. It uses important contrast features for plate object detection. Applying grayscale to the color input car image will filter the plate object. License plates in Taiwan are not only white background; they also can be red, green and yellow background. Except white, the other background colors of license plates are difficult to be located. The threshold method of this paper uses saturation, intensity and hue (HSI), to easily find attribute of the color plate and makes the plate location a precise task.
Character recognition of license plate recognition system uses neural network, syntactic structural, template matching and fixed-point sampling method. This paper applies neural network for character recognition. It is highly accurate and can be structurally designed on FPGA with great performance. Our experiment shows the proposed FPGA-based license plate recognition system achieves 98.18% license plate detection rate, 96.27% character segmentation rate, and 97.09% character recognition rate.

[1] Q. Gao, X. Wang and G. Xie, “License Plate Recognition Based on Prior Knowledge,” In: Proc. of the IEEE Int. Conf. on Automation and logistics, pp. 2964-2968, 2007.
[2] 吳成柯，戴善榮，程湘君，雲立實 譯， 數位影像處理，儒林圖書有限公司，1996
[3] R. C. Gonzalez and R. E. Woods, “Digital Image Processing,” 2nd Edition, Prentice-Hall, New Jersey, 2002.
[4] Y. Han and R. A. Wagner, “An efficient and fast parallel-connected component algorithm,” J. ACM, 1990.
[5] E. R. Davies, Machine Vision. Theroy, Algorithm ,Practicalities, 3rded., Morgan Kaufmann, San Francisco, 2005.
[6] A. Rosenfeld, “Connectivity in Digital Pictures,” Journal of the ACM(JACM), vol.17 no.1,pp.146-160, 1970.
[7] P. Comelli, P. Ferragina, M.N. Granieri, and F. Stabile, “Optical recognition of Motor Vehicle License Plates,” IEEE Trans. on Vehicular Tech, vol. 44, no.4, November 1995.
[8] R. Brunelli, Template Matching Techniques in Computer Vision: Theory and Practice, Wiley,2009.
[9] 陳翔傑，「自動化車牌辨識系統設計」碩士論文，國立中央大學，桃園，2005。
[10] K. Deb and K. H. Jo, “HSI Color based Vehicle License Plate Detection,” International Conference on Control, Automation and Systems 2008.
[11] E. Sackinger, B. Boser, J. Bromley, Y. LeCun, and L.D. Jackel, “Application of the ANNA neural network chip to high-speed character recognition,” IEEE Trans. Neural Networks, vol. 3, no. 3, pp. 498-505, May 1992.
[12] L. Carro, and D. T. Franco, “FPGA based systems with linear and non-linear signal processing capabilities,” Euromicro Conference, vol. 1, pp. 260-264, Sept. 2000.
[13] R. Parisi, E. D. Di Claudio, G. Lucarelli, and G. Orlandi, “Car plate recognition by neural networks and image processing,” Proceedings of the 1998 IEEE International Symposium on Circuits and Systems, vol. 3, pp. 195-198, 1998.
[14] 張簡子，「以小腦模型在FPGA 上作車牌辨識」碩士論文，國立台灣師範大學工業教育所，2003。
[15] L. Jianzhuang, L. Wenqing, and T. Yupeng, “Automatic threshoding of gray-level pictures using two-dimension otsu method,” International Conf. on Circuits and Systems, vol. 1, pp. 325-327, June 1991.
[16] H. Selvaraj, M. Venkatesan, “A reconfigurable printed character recognition systemusing a logic synthesis tool,” Euromicro Conference, Volume 1, pp. 24-29, Aug. 1998.
[17] H. Bai, J. Zhu, and C. Liu, “A fast license plate extraction method on complex background,” Intelligent Transportation Systems, 2003. Proceedings. 2003 IEEE, vol. 2, pp. 985-987, 2003.
[18] H. Bai and C. Liu, “A hybrid license plate extraction method based on edge statistics and morphology,” Pattern Recognition, 2004. ICPR 2004. Proceedings of the 17th International Conference on, vol. 2, pp. 831-834, 2004.
[19] Altera Document, Avalon Memory-Mapped Interface Specification.pdf
[20] N. K. Chern, P. A. Neow and M.H. Ang, “Practical issues in pixel-based autofocusing for machine vision,” Proceedings of the IEEE International Conference on Robotics and Automation, vol. 3, pp. 2791-2796, May 2001.
[21] N. K. Chern, N., P. Aun Neow, M. H. Ang, and Jr., “Practical issues in pixel–based autofocusing for machine vision,” IEEE International Conference on Robotics and Automation, vol. 3, pp. 2791-2796, 2001.
[22] 吳成柯，戴善榮，程湘君，雲立實 譯， 數位影像處理，儒林圖書有限公司，1996。
[23] N. Otsu, “A Threshold Selection Method from Gray-Level Historgams,” IEEE Transactions on System, Man, and Cybernetics, vol. SMC-9, pp. 62-66, 1979.
[24] 鍾國亮，「影像處理與電腦視覺」第五版，東華書局，2012。
[25] H. E. Kocer and K. K. Cevik, “Artificial neural networks based vehicle license plate recognition,” Proce. Comput. Sci., vol. 3, pp. 1033-1037, 2011.
[26] P. Tarabek, “Fast license plate detection based on edge density and integral edge image,” IEEE Applied Machine Intelligence and Informatics, vol. 10, pp. 37-40, International Symposium on, 2012
[27] N. Mani, and B. Srinivasan, “Application of Artificial Neural Network Model for Optical Character Recognition,” Proc. IEEE Int. Conf. on Systems, Man and Cybernetics, pp. 2517-2520, 1997.
[28] 何坤達，「嵌入式車牌辨識」碩士論文，國立台灣科技大學電子所，2006。
[29] 吳國修，「基於離散小波轉換與形態學影體實現多個車偵測」碩士論文，國立台灣科技大學電子所，2008。