臺灣博碩士論文加值系統

本文提出一個新的特徵擷取方法用來識別已分割文字，此方法特色在於特徵長度會隨著文字複雜度成正比，因此儲存一個簡易文字的特徵長度所佔空間較小、且在識別速度上較快，而繁難文字則反之，故能有效提升其識別效率。為進一步證實其可行，本文以一個實作中文識別系統，測試104個不同複雜度的中文字來驗證。其識別率高達99%，且此104個中文字所需儲存的特徵資料僅12.71KByte，此結果與我們預期一致。

In this paper, we proposed a new method for abstracting characteristics for identifying a well cut word. By this method, the length of characteristics will vary with the complexity of the identified word. Because of this, for a word of simple structure, we need small storage space and can get fast identify speed. On the other hand, for a word of complex structure, we adopt regular storage space and with regular identify speed. So that, we can expect to enhance the overall identify efficiency. For verifying the feasibility, in this study, a Chinese characteristic identify system a constructed, and 104 Chinese words with different complexity are tested by this system. The overall recognition rate of our system reaches 99.038%, and the data volume of 104 Chinese words is only 12.71Kbytes. The results are well agreed with our expectation.

中文摘要 Ⅰ
英文摘要 Ⅱ
目錄 Ⅲ
圖目錄 Ⅵ

第一章 緒論 1
1-1 前言 1
1-2 研究動機 4
1-3 章節概要 4
第二章 理論基礎 5
2-1 二值化 5
2-2 骨架化 7
第三章 研究方法 9
3-1 骨架化關鍵點定義 9
3-2 關鍵點搜尋 9
3-2.1 端點搜尋 9
3-2.2 交點搜尋 11
3-2.3 特例一 12
3-2.4 特例二 13
3-3 特徵向量建立 14
3-4 旋轉不變特徵 16
3-5 識別方法 16
第四章 模擬 21
4-1 系統架構 21
4-2 中文字樣本建立 22
4-3 待測中文字 24
4-4 前處理 25
4-4.1 二值化 25
4-4.2 骨架化 26
4-5 特徵值擷取 26
4-5.1 關鍵點搜尋 26
4-5.2 特徵向量建立 28
4-6 目標中文字識別 29
4-6.1 旋轉中文字識別 29
4-6.2 簡易圖形與繁雜圖形的比較 33
第五章 討論與結論 39
5-1 討論 39
5-2 結論 40
參考文獻 41
致謝 43
附錄A 影像識別程式碼 44
附錄B 樣本與測試圖片產生程式碼 54

1.D. J. Burr, Elastic matching of line drawings, IEEE Trans. Pattern Anal. Mach. Intell. 3, 708-713 (November 1981).
2.T. Wakahara, Toward robust handwritten character recognition, Pattern Recognition Lett. 14, 345-354 (April 1993).
3.T. Wakahara, Shape matching using LAT and its application to handwritten numeral recognition, IEEE Trans. Pattern Anal. Mach. Intell. 16, 618-629 (June 1994).
4.A. Kundu, Y. He and P. Bahl, Recognition of handwritten word: first and second order hidden Markov model based approach, Pattern Recognition 22(3), 283297 (1989).
5.E. Holbaek-Hanssen, K. Br~tthen and T. Taxt, A general software system for supervised statistical classification of symbols, Proc. 8th Intl Conf. Pattern Recognition pp. 144-149. Paris, France (October 1986).
6.J. S . Weszka, “A survey of threshold selection techniques,” Comput. Graph. Image Processing, vol. 7, no. 2, pp. 259-265. Apr. 1978.
7.N. Otsu, A threshold selection method from gray-level histogram, IEEE Trans. Systems Man Cybernet. SMC-8, 1978, 62-66.
8.D. Rutovitz, “Pattern recognition,” J. Roy. Stat. Soc., vol. 129, Series A, pp. 504-530,1966
9.R. Stefanelli and A. Rosenfeld, “Some parallel thinning algorithms for digital pictures,”J. ACM, vol. 18, no. 2, pp. 255-264,1971.
10.“Thinning algorithms on rectangular, hexagonal, and triangular arrays,” Comm. ACM, vol. 15, no. 9, p.827-837, 1972.
11.T. Y. Zhang and C. Y. Suen, “A fast parallel algorithm for thinning digital patterns,” Comm. ACM, vol. 27,no .3, pp. 236-239, 1984.
12.P. S. P. Wang, “An improved fast parallel thinning algorithm for digital patterns,” in Proc. Int. Conf. Comput. Vision Patt. Recogn.(San Francisco), 1985, pp. 364-367.
13.Tai-Ning Yang, Sheng-De Wang “A rotation invariant printed Chinese character recognition system” . * Department of Electrical Engineering, National Taiwan University, EE Building, Taipei 106, Taiwan Received 31 December 1998; received in revised form 30 August 2000