臺灣博碩士論文加值系統

隨著科技的日新月異造成的進步，電子產品的功能也越來越強大。在過去需要透過掃描機才能取得的高解析度文字影像，在現代卻往往只要透過一個普通的數位相機便能夠取得，相較起來更具有可攜性與便利性。受此影響，文字辨識的範圍也不再侷限於過往的平面文字影像，而延伸到了更生活化的三維立體影像中。
而在文字影像不再屬於單純的平面以後，對於文字影像辨識也帶來了更多問題。其中最主要的問題是文字所屬平面帶來的文字變形。在圓柱影像中，位於圓柱兩側的文字在影像中會因相機並非完全水平面對文字而呈現傾斜，除了文字變形外，也造成原來水平的文字行變形成一條曲線，也進而導致了文字辨識的困難。
本研究提供了一個有效正確的圓柱文字影像校正方法，利用影像前處理包含全域二值化、文字區塊標記來擷取欲分析之影像資料。其次將利用本研究中所提出之連通元件將文字字元進行正確的外型標記。而後將連通元件連結成曲線文字行，在以迴歸分析分析出曲線方程式後，將曲線文字行校正為直線，並可將此正確之校正結果提供後續的切割與辨識系統之用。

Due to the rapid development of scientific technology, electronic products becomes more advanced functions. It used to use scanner to get high resolution 2D image, but now we can do that by using ordinary digital cameras. Portability and convenience of using digital cameras. Character recognition is no longer limited by recognizing 2D characters. It can also be extended to recognize 3D characters because of the inherent characteristic of digital cameras.
Since the images and texts are no longer planes, the recognition of them has brings in many problems. The major problem is that the plane to which the texts belong causes the sphere change of texts. When it formed the images of cylinder, the texts locate on both sides of the cylinder will slant because the camera is not totally horizontal in the images. In addition to the change of texts, the texts horizontal will be transformed into a curve which increases the difficulty in recognizing the texts.
In this thesis, we present an effective method to correct the text images of cylinder sphere. Firstly, image preprocessing is performed including global binarization and connected-component labeling to extract the image information. Next, a modified connected-component labeling is employed to will correct the labeling characters, and link these components to a curved text line. After the using of regression analysis to analyze the curve function, we will correct the curved text line to a horizontal line. The result can be used to facilitate later segmentation and recognition system.

Abstract ii
目錄 iv
第一章 緒論 1
1.1 研究動機與目的 1
1.2 相關研究 3
1.3 系統流程 5
1.4 論文架構 8
第二章 前處理 9
2.1 彩色影像轉換灰階影像 11
2.2 全域二值化 12
2.3 連通元件偵測 14
2.4 雜訊去除與圖文分離 17
2.5 文字行串連 17
2.6 區域二值化 19
第三章 曲線文字行繪製 21
3.1 連通元件修正 21
3.1.1 文字分類 23
3.1.2 連通元件繪製 26
3.2 連結曲線文字行 29
3.2.1 橢圓逼近與拋物線逼近 29
3.2.2 拋物線回歸分析 30
第四章 曲線文字校正 36
4.1 曲線Typoline繪製 37
4.1.1 K-means演算法介紹 38
4.1.2 Typoline演算法介紹 38
4.1.3 以Typoline修正連通元件 39
4.2 曲線文字行校正 40
4.3 文字寬度校正 41
第五章 實驗結果與討論 44
5.1 曲線文字行偵測結果 44
5.2 曲線文字行校正及其結果 48
5.3 圓柱寬度校正及其結果 52
5.4 討論 57
第六章 結論與未來工作 62
6.1 結論 62
6.2 未來工作 62
參考文獻 64

[1]. H. Yan, “Fuzzy Curve-Tracing Algorithm”, IEEE Transactions on Systems, Msn, and Cybernetics-PartB: CyberNetics, Vol.31, No.5, October 2001.
[2]. H. Yan, “Detection of Curved Text Path Based on the Fuzzy Curve-tracing (FCT) Algorithm”, International Journal on Document Analysis and Recognition, 2001.
[3]. B. S. Y Lum and H. Yan, “Complex Curve Tracing Based on a Minimum Spanning Tree Model and Regulrized Fuzzy Clustering”, International Conference on Image Processing, 2004.
[4] G. Toussaint, “Solving Geometric Problems with the Rotating Calipers”, IEEE proceedings In MELECON, 1983.
[5]. A. Zramdini and R. Ingold, “Optical Font Recognition Using Typographical Features”, IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 20, No. 8, August 1998.
[6]. T. Yamaguchi, Y. Nakano, M. Maruyama, H. Miyao and T. Hananoi, “Digit Classification on Signboards for Telephone Number Recognition”, conference on the Seventh International Document Analysis and Recognition, 2003.
[7]. H. Goto, K. Aoba, H. Aso, “A Framework for Detecting and Selecting Text Line Candidates of Correct Orientation”, IEEE conference on 14th International Pattern Recognition, icpr, vol. 2, pp.1074, 1998.
[8]. H. Goto, H. Aso, “Extracting curved text lines using local linearity of the text line”, International Journal on Document Analysis and Recognition, 1999.
[9]. Z. Zhang, C. L. Tan, "Correcting Document Image Warping Based on Regression of Curve Text Lines ", IEEE Conference on the Seventh International Document Analysis and Recognition, 2003.
[10]. H. Hase, M. Yoneda, T. Shinokawa, C. Y. Suen, “Alignment of Free Layout Color Texts for Character Recognition”, Sixth International Conference on Document Analysis and Recognition, pp.0932, 2001.
[11]. Din-Chang Tseng, “Image Processing”, Institute of Compute. Sciences & Information Engineering National Central University.
[12]. N. Otsu, "A Threshold Selection Method from gray-level Histograms." IEEE International Conference on System, Man, and Cybernetics. vol. 9, pp. 62–66, 1979..
[13]. J. T. Tou and R. C. Gonzalez, “Pattern Recognition Principles.” Addision-Wesley Publishing Company, 1974.
[14]. S. S. Khan and A. Ahmad, “Cluster center initialization algorithm for K-means Clustering.” IEEE Pattern Recognition Letters, vol. 25, pp. 1293-1302, 2004..