車牌偵測系統近幾年無論在學術或商業上被廣泛的研究與發展，本論文依據文字的紋理特徵與車牌形狀特徵，提出一個改進的方式。

首先，以二維離散小波轉換萃取重要的字元紋理之後，再利用形態學的斷開與閉合達成車牌紋理群聚，並配合尺寸門檻機制與連通法將車牌完全的擷取出來。擷取出的車牌區域，先將其區域二值化後，再利用連通法把每個字元切割出來，每個字元一樣通過尺寸門檻值過濾之後，再送至OCR軟體來辨識字元，以驗證車牌偵測與字元切割是否成功。

本論文以FGPA實現小波離散轉換以及形態學硬體，提供一個更快速的讀取與演算硬體設計，並且達到多個車牌偵測與機車車牌偵測。利用上述的方法，我們不僅可以偵測汽車車牌，也可以偵測到機車車牌，並且達到多個車牌的偵測。經過實驗，系統可以有95% 的車牌偵測率，本論文系統實現上除了在應用面的研究之外，還提出改善離散小波轉換讀取模組與形態學演算的學術研究成果，經過實際硬體的實現驗證，該兩項模組均達到可觀的執行速度改善。

The field of license plate detection has been widely researched and developed in the region of academic and business. According to the feature of text and shape of license plate, this paper presents an improved method.

First, the vertical features are extracted via 2-D discrete wavelet transform (DWT). Second, the features will be grouped with the license plate shape by morphological process (Closing and Opening). Final, the candidate regions will be filtered via many dimension thresholds with 4-Connected Components, and then to obtain the license plate region. All the characters of license plate region are segmented through binary, 4-Connected Component and many dimension thresholds. Then we use OCR to recognize those characters for verifying the capability of license plate detection.

This paper presents two hardware modules that involve DWT and mathematical morphological (MM). Faster data access and multi-licenses detection are realized through the hardware accomplishment in FPGA. Not only the license plates of the automobile are located but also motorbike and multi-plates are both be located via the method we mention above. Beside the application field, this paper improves the DWT module into a simple design as well as speed-up the MM module with an advanced algorithm in academic field. In the experiment, the faster performance of these tow hardware modules is verified.

[1] “Method of License Plate Location Based on Corner Feature”, Zhong Qin, Shengli Shi, Jianmin Xu and Hui Fu, Intelligent Control and Automation, 2006. WCICA 2006. The Sixth World Congress on vol, 2, 21-23 June 2006 Page(s):8645 – 8649.

[2] “Multiple license plate detection for complex background”, Ching-Tang Hsieh, Yu-Shan Juan, Kuo-Ming Hung, Advanced Information Networking and Applications, 2005. AINA 2005. 19th International Conference on vol. 2, 28-30 March 2005 Page(s):389 - 392
.
[3] “An Approach to Korean License Plate Recongnition Based on Vertical Edge Matching”, Mei Yu and Yong Deak Kim, Systems, Man, and Cybernetics, 2000 IEEE International Conference on vol. 4, Page(s): 2975-2980 vol.4

[4] “Arabic Handwriting Texture Analysis for Writer Identification Using the DWT-Lifting Scheme”, Sami Gazzah and Najoua Essoukri Ben Amara, Document Analysis and Recognition, 2007. ICDAR 2007. Ninth International Conference on
vol. 2, 23-26 Sept. 2007 Page(s):1133 - 1137

[5] “Automatic Text Detection and Tracking in Digital Video”, Li H, Doerman D, Kia O. Image Processing, IEEE Transactions on Volume 9, Issue 1, Jan. 2000 Page(s):147 - 156

[6] “Segmentation of Text from Compound Imaged”, N.Krishnan, C. Nelson Kennedy Babu, S.Ravi and Josphine Thavamani, Conference on Computational Intelligence and Multimedia Applications, 2007. International Conference on vol. 3, 13-15 Dec. 2007 Page(s):526 - 528

[7] “JPEG2000 Standard for Image Compression Concepts, Algorithms and VLSI Architectures”, Tinku Acharya & Ping-Sing Tsai, Hoboken, N.J.: Wiley - Interscience, c2005.

[8] 吳炳飛、胡亦強、瞿忠正、蘇崇彥，JPEG2000 影像壓縮技術，全華科技圖書

[9] “JPEG2000 Still Image Coding System”, Charilaos Christopoulos, Athanassios Skodras, and Touradj Ebrahim, Consumer Electronics, IEEE Transactions on vol. 46, Issue 4, Nov. 2000 Page(s):1103 - 1127

[10] “The lifting scheme: A custom-design construction of biorthogonal wavelets”, W. Sweldens, Appl. Comput. Harmon. Anal., vol. 3, no. 2, 1996 Page(s). 186–200.

[11] “Fast Integer-to-Integer Reversible Lifting Transform with Reduced Memory Consumption”, Jose Oliver, Elena Oliver, Manuel P. Malumbres, Signal Processing and Information Technology, 2005. Proceedings of the Fifth IEEE International Symposium on18-21 Dec. 2005 Page(s):289 - 294

[12] ADSP-BF533 Blackfin Processor Hardware Reference

[13] “Digital Image Processing Second Edition”, Rafael C. Gonzalez, Richard E. Woods, Upper Saddle River, N.J.: Prentice-Hall, c2002.

[14] “Threshold Selection Method from Gray-Level Histogram “, N. Otsu, Systems, Man and Cybernetics, IEEE Transactions on vol. 9, Issue 1, Jan. 1979 Page(s):62 - 66

[15] 鍾國亮，影像處理與電腦視覺，東華書局

[16] Altera Document, Avalon Memory-Mapped Interface Specification.pdf

[17] “A VLSI Architecture for Lifting-Based Forward and Inverse Wavelet Transform “, Kishore Andra, Chaitali Chakrabarti, Tinku Acharya, Signal Processing, IEEE Transactions on vol. 50, Issue 4, April 2002 Page(s):966 - 977

[18] “Low-Complexity Binary Morphology Architectures with Flat Rectangular Structuring Elements”, Hugo Hedberg. Fredrik Kristensen, Viktor O¨ wall. Circuits and Systems I: Regular Papers, IEEE Transactions on vol. PP, Forthcoming, 2003 Page(s):1 - 1

[19] “Speeding-up Mathematical Morphology Computations with Special-Purpose Array Processors”, Alberto Broggi. System Sciences, 1994. vol. I: Architecture, Proceedings of the Twenty-Seventh Hawaii Internation Conference on vol. 1, 4-7 Jan. 1994 Page(s):321 - 330

[20] “Partial-Result-Reuse Architecture and Its Design Technique for Morphological Operations with Flat Structuring Elements”, Shao-Yi Chien, Shyh-Yih Ma, and Liang-Gee Chen. Circuits and Systems for Video Technology, IEEE Transactions on
vol. 15, Issue 9, Sept. 2005 Page(s):1156 - 1169

[21] Altera Document, mnl_nios2_board_stratix_1s10.pdf

[22] “A hybrid License Plate Extraction Method Based On Edge Statistics and Morphology”, Bai Hongliang and Liu Changping, Pattern Recognition, 2004. ICPR 2004. Proceedings of the 17th International Conference on vol. 2, 23-26 Aug. 2004 Page(s):831 - 834