手寫字風格辨識研究

Abstract

本研究主要以手寫硬筆字為研究範圍，探討手寫字風格的辨識因素。分為二階段進行研究，首先收集充足的手寫字風格代表字樣，整理分析是否具有共同架構。經過疊合演算後而得到平均手寫字樣，顯示每個手寫字樣都存有一近似正楷的骨架，為手寫字的範型。再以語意差異法的問卷調查方式，對一般人進行風格認知調查。問卷經過因素分析，萃取出人們辨識手寫字風格的主要因素為(1)結構上的空間性因子，其次為(2)筆劃態勢的動態性因子和(3)感性的感官性因子。此項初步分析結果也符合了其它不同對象的風格研究結論，要達成風格辨識得透過形狀特徵及結構關係。而所分析出的3項因子也可做為設計手寫字體風格參考方向。

The aim of this study is to use handwriting materials as a medium to investigate the psychological dimensions underlying style judgment. Handwritten samples were collected from college students. For the purpose of building the prototypical exemplar of a given handwritten character, samples of the same characters were averaged. A custom image processing protocol was developed to extract the linear version of an average character. All writing samples were assorted into 15 distinct style groups. Each group was then evaluated on 12 descriptive features. The results of the principal component analysis performed on the feature matrix indicate that there are three main factors underlying style judgment: the spatial, the dynamic, and the gender/sensational factor. The placement of the prototypical exemplars in the factor space, and its theoretical importance are also discussed in this thesis.