本研究目的在於發展一個高辨識率且低複雜度的嵌入式車牌辨識系統。本論文同時在桌上型PC以及ARM-cortex A8上面進行實驗，系統包含三個主要模組: 車牌定位、車牌字元旋轉與切割、車牌字元辨識。車牌定位是先使用小波轉換後，做邊緣偵測來找出所有可能的車牌區域；車牌字元切割是將定位出來的車牌區域，依據連通物件以及車牌字元的特性，加以分析處理，將切割出來的車牌字元旋轉至水平，並且切割出車牌字元；最後，我們提出一個兩階段式神經網路分類器(two-stage neural network classifier)來進行字元辨識。
採用適應性的車牌定位方法，在複雜和持續變動的真實環境下，我們的車牌辨識系統仍擁有高度分割出車牌字/以及背景的正確率；另外，在真實應用中，車牌影像資料總是隨著時間不斷加入，新增的車牌影像品質和特性無法以事前的批次學習方式掌握，本研究所設計的兩階段式神經網路分類器，可改善一般的分類器訓練過後無法繼續學習的缺點，得以在變動的環境下，動態調整分類器架構和參數最佳化，以維持最佳的辨識性能。
我們首先於PC進行演算法的驗證和實驗，繼而將系統移植到異質雙核心處理器的嵌入式平台。我們將分析車牌辨識系統各模組的複雜度，將計算複雜度較高的模組移到DSP處理器上處理並與ARM處理器進行協同運算，最後實現基於雙核心架構的車牌辨識嵌入式軟體。實驗顯示，與其他車牌辨識研究比較，我們的系統具有優異的辨識性能和效率。

This paper aims to establish a high recognition rate and low complexity embedded license plate recognition system. Our system consists of three main parts, including license plate location, segmentation of characters and characters recognition. The license plate locations in image are identified by wavelet transform and edge detection. The segmentation of characters uses connected-component and the feature of license plate of characters to segment each one. Character recognition is achieved with two stage neural network classifier.
We use adaptive method in License plate location. Our system still has a high correct rate of dividing the license plate and the background in complex and always changing environment. Besides, in real applications, license plate images are always being added over time, the new license plate image quality and features can’t be controlled in previous training. In this research, the two stage neural network classifier can improve the defects of the general classifiers that can’t continue learning after learning. It can dynamic adjustment of classifier structure and parameters optimization to maintain the best recognition performance in the changing environment.
At first, we have verified our algorithm and experimented on PC; after that, our license plate recognition system has migrated to the heterogeneous dual-core processors embedded system. We have analyzed each module’s complexity in our license plate recognition system, moved higher computation functions to DSP processor and collaborative computing with the ARM processor. Finally we implemented the embedded software based on dual-core architecture. Experiments show that our system has excellent performance and efficiency of identification than other license plate recognition researches.

[1] Chien-Hung Chen. 2008. “The Study of Checking the Annual Inspection Status of Motorcycles Based on License Plate Recognition”, Taipei, Taiwan: Tatung University, Department of Computer Science and Engineering, page 25-77.
[2] D.F. Specht, 1990. “Probabilistic neural networks (original contribution) ” Neural Networks .vol. 3, no.1, page 109-118.
[3] Hsin-Te Lue. 2002. “Recognition System of License Plate Using Multi-Experts”, Zhongli, Taiwan: National Central University, Department of Computer Science and Engineering, page 55-72.
[4] H. J. Lee, Y. C. Lin. 1998. “sing confusing characters to improve character recognition rate”, IEEE International Conference on Systems, Man, and Cybernetics, Vol. 5, 1998, page 4195 -4200.
[5] Chih-Kuo Chiou. 2006. ”Implementation of A License Plate Recognition System“, Taipei, Taiwan: National Taiwan University of Science and Technology, Department of Electrical Engineering, page 45-68.
[6] Lin, Tai-Liang. 2000. “Intelligent License Plate Searching and Content Segmentation in Image Processing”, Taipei, Taiwan: National Taiwan University, Department of Electrical Engineering, page 31-46.
[7] Chang, Ming Hau. 1995. “Alphanumeric Recognition by Parts”, Kaohsiung, Taiwan, National Sun Yat-sen University, Department of Computer Science and Engineering, page 41-51.
[8] J.Barroso, J.Bulas-Cruz, E.L. Dagless, and A.Rafael. “Number Plate Reading Using Computer Vision”, Industrial Electronics, 1997. ISIE ’97, Proceedings of the IEEE international Symposium on, 1997, page 761-766 vol.3.
[9] Nabarun Bhattacharyya, Bipan Tudu, Animesh Metla, Animesh Metla and Rajib Bandyopadhyay. 2008. “Incremental PNN Classifier for a Versatile Electronic Nose”. IEEE International Conference on Sensing Technology, Nov. 30, 2008, Tainan, Taiwan, page 242-247.
[10] Hsiang-Chieh Chen. 2005. “Automatic vehicle license plate recognition system design”, Zhongli, Taiwan: National Central University, Department of Electrical Engineering, page 59-72.
[11] Tzung-Bau Guan, 2001. “DSP Embedded System for Automatic License Plate Recognition”, Taipei, Taiwan: National Taiwan University, Department of Electrical Engineering, page 60-69.
[12] Shir-Cheng Gau. 2005. “Development of Dual-Core-Processor based Real-Time Wireless Embedded Brain Signal Acquisition / Processing System and its Application on Driver''s Drowsiness Estimation”, Hsinchu, Taiwan: National Chiao Tung University, Department of Computer Science and Engineering, page 45-76.
[13] 盧怡仁、蔡偉和編著，數位訊號處理平台在嵌入式系統的應用，文魁出版社, page 60-68。
[14] [Online.] pixhawk website, “DSPLink Memory Map Tutorial”, “http://pixhawk.ethz.ch/tutorials/omap/dsplink/memorymap”
[15] [Online.] TI website, “OMAP3530 resource”, “http://focus.ti.com/docs/prod/folders/print/omap3530.html”
[16] [Online.] “DVSDK_3_00 3_00_02_44 Product Download Page” “http://software-dl.ti.com/dsps/dsps_public_sw/sdo_sb/targetcontent/dvsdk/DVSDK_3_00/3_00_02_44/index_FDS.html”
[17] Bei CHEN, Wenlun CAO, Hongcai ZHANG. 2008, “An Efficient Algorithm on Vehicle License Plate Location”, IEEE International Conference on Automation and Logistics Qingdao, China September, page 1386-1389
[18] Zhong Qin, Shengli Shi, Jianmin Xu and Hui Fu. 2006. “Method of License Plate Location Based on Corner Feature”, Proceedings of the 6th World Congress on Intelligent Control and Automation, June 21 - 23, 2006, Dalian, China. page 8645-8649
[19] Lin Wang, Renlong Pan, Mo Dai. 2009. “An efficient Method based on Orientation Field for Detection of License Plates”, Sixth International Conference on Fuzzy Systems and Knowledge Discovery, page 349-353.
[20] Zhen-Xue Chen, Cheng-Yun Liu, Fa-Liang Chang, and Guo-You Wang. 2009. “Automatic License-Plate Location and Recognition Based on Feature Salience”, IEEE Transactions on Vehicular Technology, Vol. 58, No. 7, page 3781-3785.
[21] Shyang-Lih Chang, Li-Shien Chen, Yun-Chung Chung, and Sei-Wan Chen. 2004. “Automatic License Plate Recognition”, IEEE Transactions on Intelligent Transportation Systems, Vol. 5, No. 1, page 42-53.
[22] Xin Fan, Guoliang Fan. 2009. “Graphical Models for Joint Segmentation and Recognition of License Plate Characters”, IEEE Signal Processing Letters, Vol. 16, No. 1, page 10-13.
[23] Y.-P. Huang, T.-W. Chang, Y.-R. Chen ,F.E. Sandnes. 2008. “A back propagation based real-time license plate recognition system”, Int. Journal of Pattern
Recognition and Artificial Intelligence, vol. 22, no. 2, page 233-251.
[24] M. Sarfraz, M.-J. Ahmed and S.-A. Ghazi. 2003. “Saudi Arabian license plate recognition system”, in Proc. of the Int. Conf. on Geometric Modeling and Graphics, page 36-41.
[25] J. Matas and K. Zimmermann. 2005. “Unconstrained license plate and text localization and recognition”, in Proc. of the IEEE Intelligent Transportation Systems, Vienna, Austria, page 225-230.
[26] R. Juntanasub and N. Sureerattanan. 2005. “Car license plate recognition through
Hausdorff distance technique”, in Proc. of the 17th IEEE Int. Conf. on Tools with
Artificial Intelligence, Hong Kong, China, page 647-651.