Automatic generation of personalized Chinese handwriting characters

This paper presents a novel algorithmic method for automatically generating personal handwriting styles of Chinese characters through an example-based approach. The method first splits a whole Chinese character into multiple constituent parts, such as strokes, radicals, and frequent character components. The algorithm then analyzes and learns the characteristics of character handwriting styles both defined in the Chinese national font standard and those exhibited in a person's own handwriting records. In such an analysis process, we adopt a parametric representation of character shapes and also examine the spatial relationships between multiple constituent components of a character. By imitating shapes of individual character components as well as the spatial relationships between them, the proposed method can automatically generate personalized handwritings following an example-based approach. To explore the quality of our automatic generation algorithm, we compare the computer generated results with the authentic human handwriting samples, which appear satisfying for entertainment or mobile applications as agreed by Chinese subjects in our user study.