車牌辨識系統的應用相當廣泛，例如，電子停車位管理系統、交通違規偵測系統以及被盜車輛系統。大多數的解決方法為使用典型的車牌辨識演算法，透過影像分析技術來處理，主要為三個階段，包括車牌偵測、字元切割，以及字元辨識。這些方法發展了許多年，並且不斷的改進與優化其辨識率。但是都必須著重在兩個前提情況下：一、車牌必須清晰，且不能存在汙損，光源必須均勻；二、車牌不能過於傾斜，使的拍攝角度往往需要固定其位置，否則在字元分割上將會受到影響，造成辨識不易。再者，目前的車牌辨識都採用GPU(Graphics Processing Unit)運算與高階的硬體設備，使成本過於昂貴。為了解決上述三點問題，本研究使用嵌入式系統以及採用一模型"Tiny YOLOv3"(You Only Look Once)，該模型是一種機器學習(machine learning)，基於深度學習(deep learning)的卷積神經網路(convolutional neural network)，利用卷積層(convolution layer)來擷取目標物的特徵，進而達到物件偵測效果。整個辨識過程使用兩組神經網路，第一組從影像中偵測車牌，第二組從偵測到的車牌進行影像處理並字元分割，將分割到的字元送進Tesseract-OCR進行字元辨識。實驗結果顯示，本研究所提出的方法，無須全部滿足上述兩點情況以及使用高階的硬體設備，也能夠將車牌及其字元成功辨識。

License plate recognition systems are widely used, such as electronic parking management systems, traffic violation monitoring systems, and stolen vehicle systems. Most of the solutions are used typical license plate recognition algorithms, which are processed through image analysis techniques, which are mainly in three stages, including license plate localization, character segmentation, and character recognition. These methods have been developed for many years, and their recognition rate has been continuously improved and optimized. But all of them have two important prerequisites. First, the license plate must be clear, and there must be no fouling, and the light source must be uniform. Second, the license plate can’t be too skewed so that the angle and position of shooting view are usually the same. Otherwise the license plate detection and character segmentation will be seriously affected and then cause recognition difficulty. Furthermore, current license plate recognition uses GPU (Graphics Processing Unit) operation and high-end hardware equipment, making the cost too expensive. In order to solve these three points, our research use a model “Tiny YOLOv3” (You Only Look Once). This model is a convolutional neural network based on the deep learning in machine learning. It uses convolutional layers to get the features of object and then achieve the effect of recognition. We use a total of two models in the research. Detecting license plate from the image in the first model, and then use performs image processing and character segmentation from the detected license plates, and sends the segmented characters to Tesseract OCR to do character recognition in the second model. The results show that our research can successfully recognize the license plate and its characters without fully satisfying the above two points and using high-end hardware equipment. 

[1] Y. Zhang, Image Understanding. Walter de Gruyter GmbH & Co KG, 2017.
[2] C.-H. Lin, Y.-S. Lin, and W.-C. Liu, "An efficient license plate recognition system using convolution neural networks," in 2018 IEEE International Conference on Applied System Invention (ICASI), 2018, pp. 224-227: IEEE.
[3] Chuin-Mu Wang and Jian-Hong Liu, "License plate recognition system," in 2015 12th International Conference on Fuzzy Systems and Knowledge Discovery (FSKD), pp. 1708-1710, 2015.
[4] Ying Dia, Nanning Zheng, Xining Zhang and Guorong Xuan, "Automatic recognition of province name on the license plate of moving vehicle," in 9th International Conference on Pattern Recognition, pp. 927-929 vol.2, 1988.
[5] C. Lin and C. Wu, "A Lightweight, High-Performance Multi-Angle License Plate Recognition Model," in 2019 International Conference on Advanced Mechatronic Systems (ICAMechS), pp. 235-240, 2019.
[6] Y. Wen, Y. Lu, J. Yan, Z. Zhou, K. M. von Deneen and P. Shi, "An Algorithm for License Plate Recognition Applied to Intelligent Transportation System," IEEE Transactions on Intelligent Transportation Systems, vol. 12, no. 3, pp. 830-845, Sept. 2011.
[7] C. Lin and Y. Li, "A License Plate Recognition System for Severe Tilt Angles Using Mask R-CNN," in 2019 International Conference on Advanced Mechatronic Systems (ICAMechS), pp. 229-234, 2019.
[8] N. Wang, X. Zhu and J. Zhang, "License Plate Segmentation and Recognition of Chinese Vehicle Based on BPNN," in 2016 12th International Conference on Computational Intelligence and Security (CIS), pp. 403-406, 2016.
[9] B. Pechiammal and J. A. Renjith, "An efficient approach for automatic license plate recognition system," in 2017 Third International Conference on Science Technology Engineering & Management (ICONSTEM), pp. 121-129, 2017.
[10] R. Laroca, et al., "A Robust Real-Time Automatic License Plate Recognition Based on the YOLO Detector," in 2018 International Joint Conference on Neural Networks (IJCNN), pp. 1-10, 2018.
[11] S. Yonetsu, Y. Iwamoto and Y. W. Chen, "Two-Stage YOLOv2 for Accurate License-Plate Detection in Complex Scenes," in 2019 IEEE International Conference on Consumer Electronics (ICCE), pp. 1-4, 2019.
[12] J. P. D. Dalida, A. N. Galiza, A. G. O. Godoy, M. Q. Nakaegawa, J. L. M. Vallester and A. R. d. Cruz, "Development of intelligent transportation system for Philippine license plate recognition," in 2016 IEEE Region 10 Conference (TENCON), pp. 3762-3766, 2016.
[13] S. Farid and F. Ahmed, "Application of Niblack's method on images," in Emerging Technologies, 2009. ICET 2009. International Conference on, pp. 280-286, 2009.
[14] C. Anagnostopoulos, I. Anagnostopoulos, V. Loumos, and E. Kayafas, "A license plate-recognition algorithm for intelligent transportation system applications," IEEE Trans. Intell. Transp. Syst., vol. 7, no. 3, pp. 377–392, Sep. 2006.
[15] H. Bai and C. Liu, "A hybrid license plate extraction method based on edge statistics and morphology," in Proc. Int. Conf. Pattern Recognit., vol. 2. 2004, pp. 831–834.
[16] Satadal Saha, Subhadip Basu and Mita Nasipuri, "Automatic Localization and Recognition of License Plate Characters for Indian Vehicles," International Journal of Computer Science & Emerging Technologies, IJCSET, E-ISSN: 2044 - 6004Vol-2 No 4 August, 2011
[17] John G. Daugman, "Uncertainty relation for resolution in space, spatial frequency, and orientation optimized by two-dimensional visual cortical filters," Journal of the Optical Society of America A, vol. 2, no. 7, pp. 1160–1169, 1985.
[18] H. Fujisawa, Y. Nakano, and K. Kurino, "Segmentation methods for character recognition: from segmentation to document structure analysis," Proceedings of the IEEE, vol. 80, no. 7, pp. 1079–1092, Jul 1992.
[19] E. Tuba and N. Bacanin, "An algorithm for handwritten digit recognition using projection histograms and SVM classifier," in 23rd Telecommunications Forum Telfor (TELFOR), Nov 2015, pp. 464–467.
[20] P. Comelli, P. Ferragina, M. N. Granieri, and F. Stabile, "Optical recognition of motor vehicle license plates," IEEE Trans. Veh. Technol., vol. 44, no. 4, pp. 790–799, Nov. 1995.
[21] J. Redmon and A. Farhadi. (2016). "YOLO9000: Better, faster, stronger," [Online]. Available: https://arxiv.org/abs/1612.08242
[22] X. Ma and E. Hovy, "End-to-end sequence labeling via bidirectional lstm-cnns-crf," in ACL, 2016, pp. 1064–1074.
[23] S. M. Silva,C. R. Jung, "Real-time brazilian license plate detection and recognition using deep convolutional neural networks," in 2017 30th SIBGRAPI Conference on Graphics, Patterns and Images (SIBGRAPI), pp. 55-62, 2017.
[24] Z. Qin, Q. Li, H. Li, X. Dong and Z. Ren, "Advanced Intersection over Union Loss for Visual Tracking," in 2019 Chinese Automation Congress (CAC), pp. 5869-5873, 2019.
[25] J. Redmon, S. K. Divvala, R. B. Girshick, and A. Farhadi, "You only look once: Unified, real-time object detection," CoRR, vol. abs/1506.02640, 2015.
[26] Joseph Redmon and Ali Farhadi, "Yolov3: An incremental improvement," CoRR, abs/1804.02767, 2018.
[27] B. V. Kakani, D. Gandhi, and S. Jani, "Improved OCR based automatic vehicle number plate recognition using features trained neural network," in Proceedings of the 2017 8th International Conference on Computing, Communication and Networking Technologies (ICCCNT’17), 2017, pp. 1-6.
[28] Vijay, P.P. and Patil, N.C., "Gray Scale Image Segmentation using OTSU Thresholding Optimal Approach," Journal for Research vol. 2, pp. 20-24, 2016.
[29] R. M. Haralick, S. R. Sternberg and X. Zhuang, "Image analysis using mathematical morphology," IEEE Trans. Pattern Anal. Mach. Intell., (4), pp. 532-550, 1987.
[30] C. Phromlikhit, F. Cheevasuvit, and S. Yimman, "Tablet counting machine base on image processing," in Biomedical Engineering International Conference (BMEiCON), 2012. IEEE, 2012, pp. 1–5.
[31] H. Kang, S.-H. Lee, J.-Y. Lee, "Image segmentation based on fuzzy flood fill mean shift algorithm," Annual Meeting of the North American Fuzzy Information Processing Society, pp. 1-6, 2010.
[32] A. Beibut, K. Magzhan and K. Chingiz, "Effective algorithms and methods for automatic number plate recognition," 2014 IEEE 8th International Conference on Application of Information and Communication Technologies (AICT), Astana, 2014, pp. 1-4.
[33] R. Smith, D. Antonova, and D. Lee, "Adapting the Tesseract open source OCR engine for multilingual OCR," in Proc. Joint Workshop Multilingual OCR Anal. Noisy Unstruct. Text Data, 2011.
[34] R. W. Smith, "Hybrid page layout analysis via tab-stop detection," in Proc. 10th International Conference on Document Analysis and Recognition (ICDAR). IEEE, Jul. 2009, pp. 241–245.
[35] F. Simistira, A. Ul-Hassan, V. Papavassiliou, B. Gatos, V. Katsouros, and M. Liwicki. "Recognition of historical greek polytonic scripts using lstm networks, " In 2015 13th International Conference on Document Analysis and Recognition (ICDAR), pp. 766–770, Aug 2015.
[36] R. Smith. (2016). Tesseract Blends Old and New OCR Technology. [Online]. Available: https://github.com/tesseract-ocr/docs/tree/master/ das_tutorial2016.
[37] R. Smith, "An Overview of the Tesseract OCR Engine," International Conference on Document Analysis and Recognition (ICDAR), Vol. 2, pp. 629-633, 2007.
[38] C. Clausner, A. Antonacopoulos, and S. Pletschacher, "Efficient and effective OCR engine training," Int. J. Document Anal. Recognit., to be published.