個人化全真字型

Personalized TrueType Font

吳先忠(Hsien-Chung Wu)

指導教授 ： 洪文斌

淡江大學/工學院/資訊工程學系碩士班/碩士(2007年)

https://doi.org/10.6846/TKU.2007.00791

摘要

本篇論文主要探討如何對個人手寫字影像，在與原手寫字不失真且利用較少貝氏曲線的情況下，自動合成字形外框，並可將其應用在 TrueType字型檔上。
本文所提出的演算法，共包含四個步驟。首先，第一步驟，先以邊界搜尋演算法(Boundary Following Algorithm)擷取出手寫字的輪廓。第二步驟，再計算由這些邊界點所構成之支援區域(Region of Support)的共變異矩陣(Covariance Matrix)，並求得其最小特徵值(Eigenvalue)，作為檢測邊界點彎曲程度的指標，藉此定位輪廓上的轉折點。在第三步驟中，因為這些轉折點會將手寫字輪廓切割為數段的曲線，因此，利用二次貝氏曲線(Quadratic Bezier Curve)，來對每一輪廓片段進行曲線密合(Curve Fitting)。第四步驟，則進行參數的最佳化，藉此來減少合成手寫字形外框，所需的二次貝氏曲線個數。
在本論文中，使用者先透過Waltop WCK-C121的手寫板，搭配手寫軟體：神來e筆e-Calligraphy v1.6.4，藉此取得具有筆畫粗細變化的個人手寫字影像，而本系統在偵測字形輪廓之特徵點的步驟上，可以有效的去除手寫字所產生的雜訊，並且可使用較少的二次貝氏曲線來描繪字形外框，且合成的字形非常接近原始的個人手寫字，因此，本系統亦可在不失真的情況下，達到壓縮字形資料點的目的。

關鍵字

轉折點偵測 ； 曲線密合 ； 貝氏曲線 ； 字形描述

並列摘要

In this thesis, we propose a method to capture the outline from the bitmap image of personal handwriting character automatically and make use of fewer Bezier curves with minimum distortion and then it can apply to the TrueType font.
In the system binary handwriting images are used as input, the contour of handwriting character is found first by the boundary following algorithm. Then the corner detection method that based on the eigenvalues of the covariance matrix of boundary points over the support region and it divide the contour into several segments. Therefore, we could fit quadratic Bezier curve to each segment. However the Bezier curve is not always the best fitting and it needs to break the segment into two segments. So finally we apply the step of reparameterization on the segment, and the purpose of this step is finding the best parameter of quadratic Bezier curve to reduce the number of Bezier curves that describe the outline of personal handwriting character.
In this thesis, user through the tablet of Waltop WCK-C121 with handwriting software of Royal e-pen e-Calligraphy v 1.6.4 firstly, and then we can obtain the handwriting images by it. Experimental results have shown that the proposed methods of capturing outlines can robust in noisy contour of handwriting character and the outlines are described using of less quadratic Bezier curve and it is similar to the handwriting character.

並列關鍵字

Corner detection ； Curve fitting ； Bezier curve ； Outline capturing

參考文獻

[12] G. Agam and I. Dinstein, “Regulated morphological operations,” Pattern Recognition, Vol. 32, No. 6, pp. 947-971, 1999.

[13] H. Anton, Elementary linear algebra, Eighth Edition, Wiley, 2000.

[15] H.H. Chang and H. Yan, “Vectorization of hand-drawn image using piecewise cubic Bézier curves fitting,” Pattern Recognition, Vol. 31, No. 11, pp. 1747–1755, 1998.

[16] D. Chetrikov and S. Zsabo, “A simple and efficient algorithm for detection of high curvature points in planar curves,” Proceedings of the 23rd Workshop of the Austrailian Pattern Recognition Group, pp. 1751–2184, 1999.

[17] D.S. Guru and R. Dinesh, “Non-parametric adaptive region of support useful for corner detection: A novel approach,” Pattern Recognition, Vol. 37, No. 1, pp. 165–168, 2004.