Skip to content

xjnpark/CCFont

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 
 
 

Repository files navigation

CCFont : Component Based Chinese Font Generation Model

Park, J.; Hassan, A.U.; Choi,J. CCFont: Component-BasedChinese Font Generation ModelUsing Generative AdversarialNetworks (GANs).Appl. Sci.2022,12,8005. https://doi.org/10.3390/app12168005

MDPI Appl. Sci. 2022, 12(16), 8005; https://doi.org/10.3390/app12168005 PDF


Abstract

Font generation using deep learning has made considerable progress using image styletransfer, but the automatic conversion/generation of Chinese characters still remains a difficult taskowing to the complex character shape and large number of Chinese characters. Most known Chinesecharacter generation models use the image conversion method of the Chinese character shape itself;however, it is difficult to reproduce complex Chinese characters. Recent methods have utilizedcharacter compositionality by separating up to three or four components to improve the qualityof generated characters, but it is still difficult to generate high-quality results for complex Chinesecharacters with many components. In this study, we proposed the CCFont model (component-basedChinese font generation model using generative adversarial networks (GANs)) that automaticallygenerates all Chinese characters using Chinese character components (up to 17 components). TheCCFont model generates all Chinese characters in various styles using the components of Chinesecharacters based on conditional GAN. By acquiring local style information from the components,the information is more accurate and there is less information loss than when global informationis obtained from the image of the entire character, reducing the failure of style conversion andimproving quality to produce high-quality results. Additionally, the CCFont model generates high-quality results without any additional training (zero-shot font generation without any additionaltraining) for the first-seen characters and styles. For example, the CCFont model, which was trainedwith only traditional Chinese (TC) characters, generates high-quality results for languages that canbe divided into components, such as Korean and Thai, as well as simplified Chinese (SC) charactersthat are only seen during inference. CCFont can be adopted as a multi-lingual font-generation modelthat can be applied to all languages, which can be divided into components. To the best of ourknowledge, the proposed method is the first to generate a zero-shot multilingual generation modelusing components. Qualitative and quantitative experiments were conducted to demonstrate theeffectiveness of the proposed method.


Results

- Sample output images of the TC characters generated

- Sample output images of SC characters (left: output, right: target)

- Sample images of Korean (Hangul) left and Thai characters right generated by CCFont


Citation

Park, J.; Hassan, A.U.; Choi,J. CCFont: Component-BasedChinese Font Generation ModelUsing Generative AdversarialNetworks (GANs).Appl. Sci.2022,12,8005. https://doi.org/10.3390/app12168005


Copyright

The code and other helping modules are only allowed for PERSONAL and ACADEMIC usage.


About

Component-based Chinese Font Generation

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published