基於深度學習的中文手寫字辨識

以作者查詢圖書館館藏

、以作者查詢臺灣博碩士

、以作者查詢全國書目

、勘誤回報

、線上人數：15

、訪客IP：3.133.128.253

姓名

李宣霈(Hsuan-Pei Lee) 查詢紙本館藏

畢業系所

資訊工程學系

論文名稱

基於深度學習的中文手寫字辨識
(Handwritten Chinese Characters Recognition Based on Deep Learning)

相關論文

★ Single and Multi-Label Environmental Sound Recognition with Gaussian Process	★ 波束形成與音訊前處理之嵌入式系統實現
★ 語音合成及語者轉換之應用與設計	★ 基於語意之輿情分析系統
★ 高品質口述系統之設計與應用	★ 深度學習及加速強健特徵之CT影像跟骨骨折辨識及偵測
★ 基於風格向量空間之個性化協同過濾服裝推薦系統	★ RetinaNet應用於人臉偵測
★ 金融商品走勢預測	★ 整合深度學習方法預測年齡以及衰老基因之研究
★ 漢語之端到端語音合成研究	★ 基於 ARM 架構上的 ORB-SLAM2 的應用與改進
★ 基於深度學習之指數股票型基金趨勢預測	★ 探討財經新聞與金融趨勢的相關性
★ 基於卷積神經網路的情緒語音分析	★ 運用深度學習方法預測阿茲海默症惡化與腦中風手術存活

檔案

[Endnote RIS 格式]

[Bibtex 格式]

[相關文章]

[文章引用]

[完整記錄]

[館藏目錄]

至系統瀏覽論文 ( 永不開放)

摘要(中)

在機器學習和深度學習還沒這麼熱門時，文字影像辨識這個領域就已經有不少研究和討論，像OCR（Optical Character Recognition）的技術發展已經算是很成熟。隨著近年來深度學習飛快地發展，文字影像辨識也同樣獲益，越來越多搭配深度學習方法來做文字影像辨識的研究一一出爐。英數字的辨識在近年來已經逐漸成熟，但中文字礙於本身文字結構比較複雜，加上中文字庫之龐大，使得中文字辨識的技術即使搭配了深度學習，其成熟度仍比不上英數字的辨識。
除了中文字本身辨識難度比英數字高，不同人的手寫風格又不一樣，如果同一份文件有來自不同人的手寫字體，文字辨認的難度就又更高了。因此本篇論文研究的重點在於，如果使用生成網路，大量生成不同風格的字體，加到中文手寫字的資料庫，和未加入多種不同風格的手寫字體的資料庫相比，是否能夠有更好的辨認效果？

摘要(英)

Character recognition has already been a popular research field even when machine learning and deep learning haven’t been discussed frequently. For example, the technique of OCR(Optical Character Recognition) has already been quite mature. Along with the development of machine learning and deep learning these years, the research of character recognition has also made a great leap by using deep learning. English characters and digit recognition has already been quite mature. However, Chinese characters recognition hasn’t been as mature as English characters and digit recognition even if many researches were based on deep learning since the structure of Chinese characters is more complexed.
In addition that the Chinese characters recognition is more difficult than English characters and digit recognition, due to the variance of the style of handwritten characters from one person to another person, handwritten characters is even more difficult to be detected or recognized if there are more than one style of handwritten characters on a piece of paper. Therefore, the purpose of this research is to find out whether the multi-style handwritten Chinese characters dataset can do better job on character detection and recognition compared to one-style or few-style handwritten Chinese characters dataset.

關鍵字(中)

★ 深度學習
★ 中文手寫字
★ 手寫字辨識

關鍵字(英)

★ Deep Learning
★ Handwritten Chinese Chracters
★ Handwritten Characters Recognition

論文目次

中文摘要 i
Abstract ii
目錄 iii
圖目錄 v
表目錄 vi
第一章緒論 1
1-1 研究背景與動機 1
1-2 研究方法與章節概要 2
第二章深度學習 4
2-1 類神經網路 5
2-2 倒傳遞類神經網路 7
2-3 深層神經網路 9
第三章相關研究及文獻探討 12
3-1 CASIA中文手寫字數據集 12
3-2 GAN 13
3-2-1 pix2pix 13
3-2-2 zi2zi GAN架構 15
3-3 Faster R-CNN 16
3-3-1 目標檢測簡介 16
3-3-2 R-CNN[21] 17
3-3-3 Fast R-CNN[22] 18
3-3-4 Faster R-CNN[6] 19
第四章實驗架構與方法 22
4-1 生成資料庫階段 22
4-2 偵測階段 24
第五章實驗結果 26
5-1 實驗環境 26
5-2 實驗結果 26
第六章結論與未來的研究方向 28
第七章參考文獻 29

參考文獻

[1] J.Leimer, ‘Design Factors in the Development of an Optical Character Recognition Machine’, IRE Transactions on Information Theory, Special Issue on Sensory Information Processing, vol.IT-8, pp.167-171. February 1962.
[2] Anthony Kay(2007). ‘Tesseract: an Open-Source Optical Character Recognition Engine’[online], Linux Journal. Available from: https://www.linuxjournal.com/article/9676. [Accessed: 12-July-2019]
[3] Xiang Bai, Cong Yao, Yingying Zhu, ‘Scene text detection and recognition: recent advances and future trends’, Frontiers of Computer Science. June 2015.
[4] 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, IEEE. November 2011.
[5] I.Goodfellow et al., ‘Generative adversarial nets’, Advances in Neural Information Processing Systems, pp.2672-2680. 2014.
[6] Shaoqing Ren, Kaiming He, Ross Girshick, Jian Sun, ‘Faster R-CNN: Towards Real-Time Object Detection with Regional Proposal Networks’, IEEE Transactions on Pattern Analysis and Machine Intelligence, vol.39, No. 6,pp.1137-1149. June 2017.
[7] Rob Robinson(2017). ‘Convolutional Neural Networks – Basics’[online]. MLNotebook. Available from: https://mlnotebook.github.io/post/CNN1/. [Accessed: 11-July-2019]
[8] Adit Deshpande(2016). ‘A Beginner’s Guide To Understanding Convolutional Neural Networks Part 2’[online]. Available from: https://adeshpande3.github.io/adeshpande3.github.io/A-Beginner′s-Guide-To-Understanding-Convolutional-Neural-Networks-Part-2/.[Accessed: 11-July-2019]
[9] SuperDataScience Team(2018). ‘Convolutional Neural Networks(CNN): Step 4 – Full Connection’[online]. Available from: https://www.superdatascience.com/blogs/convolutional-neural-networks-cnn-step-4-full-connection. [Accessed: 11-July-2019]
[10] Yuchen Tian(2017). ‘zi2zi: Master Chinese Calligraphy with Conditional Adversarial Networks’[online]. Available from: https://kaonashi-tyc.github.io/2017/04/06/zi2zi.html. [Accessed: 11-July-2019]
[11] Augustus Odena et al., ‘Conditional Image Synthesis With Auxiliary Classifier GANs’, Machine Learning, October 2016. arXiv: 1610.09585.
[12] S.S.Farfade, M.J.Saberian, Li-Jia Li, ‘Multi-view Face Detection Using Deep Convolutional Neural Networks’, International Conference on Multimedia Retrieval 2015(ICMR), April 2015. arXiv:1502.02766
[13] Dollar.P et al., ‘Fast Feature Pyramdis for Object Detection’, IEEE Transactions on Pattern Analysis and Machine Intelligence, vol.36, No. 8, pp.1532-1545. August 2014.
[14] Lienhart.R., J.Maydt, ‘An extended set of Haar-like features for rapid object detection’, Proceedings. International Conference on Image Processing. September 2002.
[15] David G.Lowe, ‘Distinctive Image Features from Scale-Invariant Keypoints’, International Journal Computer Vision, vol.60, No. 2, pp.91-110. November 2004.
[16] Dalal.N., B.Triggs, ‘Histogram of oriented gradients for human detection’, IEEE Computer Society Conference on Computer Vision and Pattern Recognition(CVPR’05). June 2005.
[17] Andrzej Maćkiewicz, Waldemar Ratajczak, ‘Principal components analysis(PCA)’, Computers & Geosciences, vol.19, No. 3, pp.303-342. 1993.
[18] Felzenszwalb.P.F. et al., ‘Object Detection with Discriminatively Trained Part-Based Models’, IEEE Transactions on Pattern Analysis and Machine Intelligence, vol.32, No.9, pp.1627-1645. 2010.
[19] Cortes.C., V.Vapnaik., ‘Support-Vector Networks’, Machine Learning, vol.20, No.3, pp.273-297. 1995.
[20] Sermanet.P. et al., ‘OverFeat: Integrated Recognition, Localization and Detection Using Convolutional Networks’, CoRR. 2013. arXiv: 1312.6229.
[21] Ross Girshick et al., ‘Rich feature hierarchies for accurate object detection and semantic segmentation’, 2014 IEEE Conference on Computer Vision and Pattern Recognition. June 2014.
[22] Ross Girshick, ‘Fast R-CNN’, ICCV 2015. April 2015. arXiv: 1504.08083.

指導教授

王家慶(Jia-Ching Wang)

審核日期

2019-8-22

推文