對於設計師而言，設計中文字型除了需要手寫超過 6000 個字以上的常用
字，還必須逐字調整，是件繁雜的工程。隨著人工智慧的發展，以生成對抗網
路進行字體生成的技術，能夠迅速的幫助設計師完成大半工作，設計師只需
要設計出基本的字型風格後，寫出少量的字，就可以藉由電腦得到其他的字
符。
在中文字的構成中，筆畫可以說是最基本的字符構成單位。本論文提出
一個模型架構，藉由將筆畫順序加入生成對抗網路模型，將標準字型轉換成
手寫特殊字型，完成更好的字型轉換；此外，更改現有的損失函數、加入自注
意力機制、雙鑑別器，進一步改進了現有模型。其實驗結果並分別與現有的
模型做比較，並達到現今先進模型之效能。為了增加本論文之真實應用面，
我們還製作了一個字型轉換模組，可將模型生成的字符圖片，完整轉換成一
組可供電腦使用的字體。
本論文主要有以下三項貢獻：提出手寫筆畫字型生成對抗網路模型，將
標準中文字體轉換成手寫風格字型，並將筆劃概念加入模型中，增加模型之
準確度；將許多最新的方法架構加入至模型中，提升模型之準確度；並將提
出之方法實際運用，將生成出的字體製作成字型檔案，並提供介面讓使用者
使用。

Designing Chinese calligraphy has been a tricky task for designers for a long time since there are more than 6000 characters in Chinese. Thanks to the development of artificial intelligence, the techniques of generating new fonts with the deep generative adversarial network now can promptly help designers save their efforts. Nowadays, designers only need to design the basic amount of style fonts then get the rest of the characters by computing.
As we know, strokes are the fundamental components of Chinese characters. In other words, once we understand the features of strokes, we can generate a new character easily. Accordingly, this research proposes a model based on a deep generative adversarial network that can transfer standard Chinese characters into stylized ones for better character transformation. We added a serial of stroke encodings into the generative adversarial network to enhance the quality of generated images. Besides, the model improves the existing model in several aspects, including loss functions, new training methods, dual discriminator, and new architectures. Compared with the previous research, the experiment result demonstrates high-quality stylized Chinese characters images against other state-of-the-art methods. To make the study more applicable, we developed a new dataset of Chinese hand-written characters and the system to transfer standard Chinese font into hand-written stylized one which can be used for typing.

[1] Yuchen Tian. zi2zi: Master Chinese Calligraphy with Conditional Adversarial Networks. https://github.com/kaonashi-tyc/zi2zi, 2017.
[2] Isola, P., Zhu, J. Y., Zhou, T., & Efros, A. A. (2017). Image-to-image translation with conditional adversarial networks. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 1125-1134).
[3] Zhu, J. Y., Park, T., Isola, P., & Efros, A. A. (2017). Unpaired image-to-image translation using cycle-consistent adversarial networks. In Proceedings of the IEEE international conference on computer vision (pp. 2223-2232).
[4] Liu, X., Meng, G., Xiang, S., & Pan, C. (2019). FontGAN: A Unified Generative Framework for Chinese Character Stylization and De-stylization. arXiv preprint arXiv:1910.12604.
[5] Wu, S. J., Yang, C. Y., & Hsu, J. Y. J. (2020). CalliGAN: style and structure-aware Chinese calligraphy character generator. arXiv preprint arXiv:2005.12500.
[6] Zeng, J., Chen, Q., Liu, Y., Wang, M., & Yao, Y. (2020). StrokeGAN: Reducing Mode Collapse in Chinese Font Generation via Stroke Encoding. arXiv preprint arXiv:2012.08687.
[7] Gatys, L. A., Ecker, A. S., & Bethge, M. (2015). A neural algorithm of artistic style. arXiv preprint arXiv:1508.06576.
[8] Goodfellow, I. J., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., … & Bengio, Y. (2014). Generative adversarial networks. arXiv preprint arXiv:1406.2661.
[9] Mirza, M., & Osindero, S. (2014). Conditional generative adversarial nets. arXiv preprint arXiv:1411.1784.
[10] Arjovsky, M., Chintala, S., & Bottou, L. (2017, July). Wasserstein generative adversarial networks. In International conference on machine learning (pp. 214-223). PMLR.
[11] Nguyen, T. D., Le, T., Vu, H., & Phung, D. (2017). Dual discriminator generative adversarial nets. arXiv preprint arXiv:1709.03831.
[12] Azadi, S., Fisher, M., Kim, V. G., Wang, Z., Shechtman, E., & Darrell, T. (2018). Multi-content gan for few-shot font style transfer. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 7564-7573).
[13] Gao, Y., Guo, Y., Lian, Z., Tang, Y., & Xiao, J. (2019). Artistic glyph image synthesis via one-stage few-shot learning. ACM Transactions on Graphics (TOG), 38(6), 1-12.
[14] Wang, Y., Gao, Y., & Lian, Z. (2020). Attribute2Font: creating fonts you want from attributes. ACM Transactions on Graphics (TOG), 39(4), 69-1.
[15] Li, C., Taniguchi, Y., Lu, M., & Konomi, S. I. (2021). Few-shot Font Style Transfer between Different Languages. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (pp. 433-442).
[16] Yuchen Tian. Rewrite: Neural Style Transfer For Chinese Fonts. https://github.com/kaonashi-tyc/Rewrite. 2017
[17] Wen, C., Pan, Y., Chang, J., Zhang, Y., Chen, S., Wang, Y., … & Tian, Q. (2021). Handwritten Chinese font generation with collaborative stroke refinement. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (pp. 3882-3891).
[18] Park, S., Chun, S., Cha, J., Lee, B., & Shim, H. (2020). Few-shot Font Generation with Localized Style Representations and Factorization. arXiv preprint arXiv:2009.11042.
[19] Gao, Y., & Wu, J. (2020, April). GAN-Based Unpaired Chinese Character Image Translation via Skeleton Transformation and Stroke Rendering. In Proceedings of the AAAI Conference on Artificial Intelligence (Vol. 34, No. 01, pp. 646-653).
[20] Huang, Y., He, M., Jin, L., & Wang, Y. (2020, August). RD-GAN: Few/Zero-Shot Chinese Character Style Transfer via Radical Decomposition and Rendering. In European Conference on Computer Vision (pp. 156-172). Springer, Cham.
[21] Mechrez, R., Talmi, I., & Zelnik-Manor, L. (2018). The contextual loss for image transformation with non-aligned data. In Proceedings of the European Conference on Computer Vision (ECCV) (pp. 768-783).
[22] Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., ... & Polosukhin, I. (2017). Attention is all you need. In Advances in neural information processing systems (pp. 5998-6008).
[23] Hu, J., Shen, L., & Sun, G. (2018). Squeeze-and-excitation networks. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 7132-7141).
[24] Suzuki, S. (1985). Topological structural analysis of digitized binary images by border following. Computer vision, graphics, and image processing, 30(1), 32-46.