請用此 Handle URI 來引用此文件：
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/80732完整後設資料紀錄
| DC 欄位 | 值 | 語言 |
|---|---|---|
| dc.contributor.advisor | 張智星(Jyh-Shing Jang) | |
| dc.contributor.author | Hsien-Yao Shui | en |
| dc.contributor.author | 稅顯堯 | zh_TW |
| dc.date.accessioned | 2022-11-24T03:14:31Z | - |
| dc.date.available | 2021-11-05 | |
| dc.date.available | 2022-11-24T03:14:31Z | - |
| dc.date.copyright | 2021-11-05 | |
| dc.date.issued | 2021 | |
| dc.date.submitted | 2021-10-15 | |
| dc.identifier.citation | [1] ShowYourWords, WRITES Co. https://www.writes.com.tw/. [2] Yuchen Tian, “Rewrite: Neural Style Transfer For Chinese Fonts.” Retrieved Nov 23, 2016 from https://github.com/kaonashi-tyc/Rewrite, 2016. [3] Yuchen Tian, “zi2zi: Master Chinese Calligraphy with Conditional Adversarial Networks.” Retrieved Jun 3, 2017 from https://github.com/kaonashi-tyc/zi2zi, 2017. [4] ZC119, “Generating handwritten Chinese characters using CycleGAN.” https://github.com/ZC119/Handwritten-CycleGAN. [5] Alex Zhong, “High Performance Offline Handwritten Chinese Character Recognition Using GoogLeNet and Feature Maps.” https:// github.com/ zhongzhuoyao/HCCR-GoogLeNet. [6] Baoyao Zhou, Weihong Wang, Zhanghui Chen, “Easy generation of personal Chinese handwritten fonts.” 2011 IEEE International Conference on Multimedia and Expo, 2011. [7] Alfred Zong , Yuke Zhu, “StrokeBank: automating personalized chinese handwriting generation.” Twenty-Sixth IAAI Conference, 2014. [8] Zhouhui Lian, Bo Zhao, Jianguo Xiao, “Automatic generation of large-scale handwriting fonts via style learning.” SIGGRAPH Asia 2016 Technical Briefs, 2016. [9] Phillip Isola, Jun-Yan Zhu, Tinghui Zhou, Alexei A. Efros, “Image-to-image translation with conditional adversarial networks.” Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2017. [10] Augustus Odena, Christopher Olah, Jonathon Shlens, “Conditional Image Synthesis With Auxiliary Classifier GANs.” arXiv:1610.09585, 2016. [11] Yaniv Taigman, Adam Polyak, Lior Wolf, “Unsupervised Cross-Domain Image Generation.” arXiv:1611.02200, 2016. [12] Jun-Yan Zhu, Taesung Park, Phillip Isola, Alexei A. Efros, “Unpaired image-to-image translation using cycle-consistent adversarial networks.” arXiv:1703.10593, 2017. [13] Bo Chang, Qiong Zhang, Shenyi Pan, Lili Meng, “Generating Handwritten Chinese Characters using CycleGAN.” 2018 IEEE Winter Conference on Applications of Computer Vision, 2018. [14] Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun, “Deep Residual Learning for Image Recognition.” Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016, pp. 770-778, 2016. [15] Jinshan Zeng, Qi Chen, Yunxin Liu, Mingwen Wang, Yuan Yao, “StrokeGAN: Reducing Mode Collapse in Chinese Font Generation via Stroke Encoding.” arXiv preprint arXiv:2012.08687, 2020. [16] Chung-Che Wang, Hsien-Yao Shui, JyhShing Roger Jang, “Chinese Handwritten Style Transfer Using Only 25 Target Examples.” The 34th IPPR Conference on Computer Vision, Graphics, and Image Processing (CVGIP), 2021. [17] Chine-culture.com, “Grids used in Chinese calligraphy.” https://www.chine-culture.com/en/chinese-calligraphy/grids.php. [18] Satoshi Suzuki, Keiichi Abe, “Topological structural analysis of digitized binary images by border following.” Computer vision, graphics, and image processing, 1985. [19] Yiming Gao, Jiangqin Wu, “GAN-Based Unpaired Chinese Character Image Translation via Skeleton Transformation and Stroke Rendering.” In Proc. the 34th AAAI Conference on Artificial Intelligence, 646–653. New York, USA, 2020. [20] Neeramitra Reddy, “Skeletonization in python using opencv.” https://medium.com/analytics-vidhya/skeletonization-in-python-using-opencv-b7fa16867331. [21] Yue Gao, Yuan Guo, Zhouhui Lian, Yingmin Tang, Jianguo Xiao, “Artistic Glyph Image Synthesis via One-Stage Few-Shot Learning.” arXiv:1910.04987, 2019. [22] Yangchen Xie, Xinyuan Chen, Li Sun, Yue Lu, “DG-Font: Deformable Generative Networks for Unsupervised Font Generation.” arXiv:2104.03064, 2021. [23] Kumaran Arulmani, “Using Generative Adversarial Networks to Create Dog Images.” https://medium.com/@kums220/using-generative-adversarial-networks-to-create-dog-images-7dece0572e23. [24] Pytorch Team, “Inception_v3.” https://pytorch.org/hub/pytorch_vision_inception_v3/. [25] Fei Wu, “Overview of CycleGAN architecture and training.” https://towardsdatascience.com/overview-of-cyclegan-architecture-and-training-afee31612a2f. [26] Dynacw.com.tw, “DynaFont.” https://www.dynacw.com.tw/. [27] Yu Ching Sung, “Jason Handwriting.” https://www.facebook.com/groups/549661292148791/. [28] Senty Workshop, “SentyFont.” https://www.sentyfont.com/. [29] Cheng-Lin Liu, Fei Yin, Da-Han Wang, Qiu-Feng Wang, “CASIA Online and Offline Chinese Handwriting Databases.” 2011 International Conference on Document Analysis and Recognition, 2011. [30] Beijing Founder Handwriting Digital Technology Co., Ltd., “Make Font App.” https://app.mi.com/details?id=com.handwriting.makefont, 2021. [31] Diederik P. Kingma, Jimmy Ba, “Adam: A Method for Stochastic Optimization.” arXiv preprint arXiv:1412.6980, 2014. [32] Martin Heusel, Hubert Ramsauer, Thomas Unterthiner, Bernhard Nessler, Sepp Hochreiter, “GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium.” arXiv:1706.08500, 2017. [33] tesseract-ocr,“Tesseract OCR.” https://github.com/tesseract-ocr/tesseract. [34] Zhuoyao Zhong, Lianwen Jin, Zecheng Xie, “High performance offline handwritten chinese character recognition using googlenet and feature maps,” in International Conference on Document Analysis and Recognition (ICDAR), 2015. [35] Joseph Yossi Gil, Ron Kimmel, “Efficient Dilation, Erosion, Opening and Closing Algorithms.” IEEE Transactions on Pattern Analysis and Machine Intelligence 24(12):1606-1617, 2002. [36] Yue Jiang, Zhouhui Lian , Yingmin Tang, Jianguo Xiao, “SCFont: Structure-Guided Chinese Font Generation via Deep Stacked Networks.” Proceedings of the AAAI Conference on Artificial Intelligence, 33(01), 4015-4022, 2019. | |
| dc.identifier.uri | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/80732 | - |
| dc.description.abstract | "由於存在大量字符，因此生成中文字體庫是一項耗時的工作。以繁體中文設計為主需要開發超過 50,000 個字符，即使是比較小的 BIG5 集也將由 13,053 個中文字組成。當手寫中文字體具有各種不同的筆劃和部首結構樣式時，開發個性化的中文字體庫會更加困難。大多數個性化的中文字體產品都要求使用者先提供超過 1,000 個以上的手寫字，或者只能選擇有限風格的字體，無法滿足使用者多樣化的需求。近年來，許多研究提出了自動生成中文字體的AI 和計算機圖形系統，但是在輸出質量上仍然存在缺陷。在一方面基於筆劃的字體合成方法在構建筆劃提取資料庫方面遇到挑戰，繁瑣的人工的微調仍然不可避免。在另外一方面是基於學習的方法出現筆劃模糊，不完整或不正確的筆劃等常見問題。大部份的學習方法都要求用戶輸入以上 500 個字符，仍然缺乏實用性。在這項工作中，我們提出了一個兩階段的中文字體生成系統，該系統僅需要用戶編寫 25 個手寫字符。我們的系統在第一階段結合了字體樣式分類網絡，並在第二階段結合 CycleGAN 和骨架轉換網絡。實驗結果證明，與其他基於深度學習的方法相比，我們的方法具有更好的效能，除了產生風格更相近的個性化中文字體，也完美地解決了常見的中文字形生成的品質問題。" | zh_TW |
| dc.description.provenance | Made available in DSpace on 2022-11-24T03:14:31Z (GMT). No. of bitstreams: 1 U0001-1510202103503300.pdf: 8950169 bytes, checksum: 46bff9db01326feedb891f3c5cdd31ef (MD5) Previous issue date: 2021 | en |
| dc.description.tableofcontents | 致謝 ii 摘要 iii Abstract iv Contents vi List of Figures ix List of Tables xi Chapter 1 緒論 1 1.1 主題簡介 1 1.2 研究貢獻 2 1.3 工具簡介 3 1.3.1 TensorFlow 3 1.3.2 PyTorch 4 1.3.3 OpenCV 4 1.3.4 visualkeras 4 1.3.5 PIL 4 1.3.6 HCCR-GoogLeNet 5 1.4 章節概述 5 Chapter 2 文獻探討 7 2.1 基於筆畫部首為基礎的方法 7 2.2 基於深度學習的方法 8 2.2.1 卷積神經網路 9 2.2.2 成對資料訓練的生成對抗網路 10 2.2.3 非成對資料訓練的生成對抗網路 12 Chapter 3 研究方法 15 3.1 系統概觀 15 3.2 手寫中文字的選擇 17 3.2.1 中文字筆劃編碼 17 3.2.2 選擇目標中文字的方法 18 3.3 訓練資料的前處理 19 3.4 字形的骨架 21 3.5 特徵距離指標 22 3.6 損失函數 24 3.6.1 交叉熵（crossentropy） 24 3.6.2 均方誤差（Mean square error，MSE） 25 3.6.3 平均絕對值誤差（Mean absolute error，MAE） 25 3.7 風格分類網路 26 3.8 風格轉換網路 27 Chapter 4 資料集介紹 31 4.1 手寫和非手寫風格的電腦字型 31 4.2 CASIA 手寫中文資料集 32 4.3 真實手寫字輸入 33 Chapter 5 實驗設計與結果 35 5.1 實驗流程 35 5.1.1 資料前處理 35 5.1.2 神經網路架構 36 5.1.3 實驗規格 36 5.1.4 實驗環境 36 5.1.5 效果評估方式 37 5.2 實驗結果 38 5.2.1 實驗一：風格分類網路的效果 38 5.2.2 實驗二 40 5.2.3 實驗三 42 5.2.4 實驗四 43 5.2.5 實驗五 48 5.2.6 實驗六 49 5.3 錯誤分析與討論 51 Chapter 6 結論與未來展望 53 6.1 結論 53 6.2 未來展望 54 References 55 | |
| dc.language.iso | zh-TW | |
| dc.subject | CycleGAN | zh_TW |
| dc.subject | 中文字形生成 | zh_TW |
| dc.subject | 風格轉換 | zh_TW |
| dc.subject | 字體合成 | zh_TW |
| dc.subject | Chinese font generation | en |
| dc.subject | CycleGAN | en |
| dc.subject | font synthesis | en |
| dc.subject | style transfer | en |
| dc.title | 基於深度學習的中文字型生成之研究與實作 | zh_TW |
| dc.title | A Study and Implementation of Chinese Font Generation Based on Deep Learning | en |
| dc.date.schoolyear | 109-2 | |
| dc.description.degree | 碩士 | |
| dc.contributor.oralexamcommittee | 傅楸善(Hsin-Tsai Liu),王崇喆(Chih-Yang Tseng) | |
| dc.subject.keyword | 中文字形生成,風格轉換,字體合成,CycleGAN, | zh_TW |
| dc.subject.keyword | Chinese font generation,style transfer,font synthesis,CycleGAN, | en |
| dc.relation.page | 59 | |
| dc.identifier.doi | 10.6342/NTU202103746 | |
| dc.rights.note | 同意授權(限校園內公開) | |
| dc.date.accepted | 2021-10-18 | |
| dc.contributor.author-college | 電機資訊學院 | zh_TW |
| dc.contributor.author-dept | 資訊工程學研究所 | zh_TW |
| 顯示於系所單位： | 資訊工程學系 | |
文件中的檔案：
| 檔案 | 大小 | 格式 | |
|---|---|---|---|
| U0001-1510202103503300.pdf 授權僅限NTU校內IP使用（校園外請利用VPN校外連線服務） | 8.74 MB | Adobe PDF |
系統中的文件，除了特別指名其著作權條款之外，均受到著作權保護，並且保留所有的權利。