基本信息
张恒  男  硕导  中国科学院自动化研究所
电子邮件: heng.zhang@ia.ac.cn
通信地址: 北京市海淀区中关村东路95号自动化大厦
邮政编码:

研究领域

多模态文档图像智能分析和识别,包括自然场景图像、手写文本图像、平板在线手写笔迹、古籍等多种模态下的文档图像分析与理解,研究方向包括但不限于自然场景文字识别、手写文字识别、签名鉴定等

​基于国际领先的模式识别与机器学习技术,将图片上的文字内容,自动识别成可编辑的文本。支持手写体、印刷体以及自然场景中的文字识别,包括身份证、名片、工作证等卡片类,也支持货运单、票据、个人笔记、政府公文等特定场景的文档图像识别,还可对复杂自然场景图片中的文字进行检测和识别,相关科研成果广泛应用于手机终端、数字地图、智能教育和国家安全领域。相关科研成果发表在模式识别和文字识别重要期刊和会议论文中,包括ICFHR2022最佳论文一篇,入选中国科协——科技期刊双语传播工程一篇。


招生信息

   
招生专业
081104-模式识别与智能系统
招生方向
文字识别,文档图像分析与识别

教育背景

2007-09--2013-07   中国科学院大学——中科院自动化研究所   工学博士学位
2003-09--2007-07   中国科学技术大学——教改试点班   工学学士学位

工作经历

   
工作简历
2015-10~现在, 中国科学院自动化研究所, 副研究员
2013-07~2015-10,中国科学院自动化研究所, 助理研究员
社会兼职
2019-05-17-今,文档图像分析与识别专委会, 委员

专利与奖励

   
专利成果
( 1 ) 基于语义图谱的短文本特征扩展方法, 发明专利, 2015, 第 4 作者, 专利号: CN104391942A

( 2 ) 基于非监督关键二元词串提取的微博文本自动摘要方法, 发明专利, 2014, 第 3 作者, 专利号: CN104216875A

出版信息

   
发表论文
[1] 于明明, 张恒, 殷飞, 刘成林. An Approach for Handwritten Chinese Text Recognition Unifying Character Segmentation and Recognition. Pattern Recognition[J]. 2024, 
[2] 李天佐, 张恒, 李晓辉, 殷飞. Adaptive Scaling and Reffned Pyramid Feature Fusion Network for Scene Text Segmentation. ICDAR. 2024, 
[3] 凌禄蓉, 张恒, 殷飞, 刘成林. Deep Metric Learning with Cross-Writer Attention for Offfine Signature Veriffcation. ICDAR. 2024, 
[4] 陈懿, 张恒, 刘成林. Improved Learning for Online Handwritten Chinese Text Recognition with Convolutional Prototype Network. ICDAR2023[J]. 2023, 
[5] 何平, 张恒, 刘成林. 基于多尺度注意力机制的场景文本擦除. 模式识别与人工智能[J]. 2022, 
[6] 于明明, Heng Zhang, 刘成林. An efficient prototype-based model for handwritten text recognition with multi-loss fusion (best paper award). ICFHR. 2022, 
[7] 高立崑, 张恒, 刘成林. Regularing CTC in Expectation-Maximization Framework with Application to Handwritten Text Recognition. IJCNN2021[J]. 2021, 
[8] 高立崑, 张恒, 刘成林. Handwritten text recognition with convolutional prototype network and most aligned frame based CTC training. ICDAR[J]. 2021, 
[9] 张颐康, 张恒, 刘成林. 基于跨模态深度度量学习的甲骨文字识别. 自动化学报[J]. 2021, 
[10] Zhang Yikang, Zhang Heng, Liu Yongge, Yang Qing, Liu Chenglin. Oracle Character Recognition by Nearest Neighbor Classification with Deep Metric Learning. ICDAR. 2019, http://www.irgrid.ac.cn/handle/1471x/6862437.
[11] Zhang, Heng, Zhou, XiangDong, Liu, ChengLin. Keyword spotting in handwritten chinese documents using semi-markov conditional random fields. ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE[J]. 2017, 58: 49-61, http://dx.doi.org/10.1016/j.engappai.2016.11.006.
[12] Zhang, Heng, Zhong, Guoqiang. Improving short text classification by learning vector representations of both words and hidden topics. KNOWLEDGE-BASED SYSTEMS[J]. 2016, 102: 76-86, http://dx.doi.org/10.1016/j.knosys.2016.03.027.
[13] Wang Peng, Xu Jiaming, Xu Bo, Liu ChengLin, Zhang Heng, Wang Fangyuan, Hao Hongwei, Zong C, Strube M. Semantic Clustering and Convolutional Neural Network for Short Text Categorization. PROCEEDINGS OF THE 53RD ANNUAL MEETING OF THE ASSOCIATION FOR COMPUTATIONAL LINGUISTICS (ACL) AND THE 7TH INTERNATIONAL JOINT CONFERENCE ON NATURAL LANGUAGE PROCESSING (IJCNLP), VOL 2. 2015, 352-357, http://apps.webofknowledge.com/CitedFullRecord.do?product=UA&colName=WOS&SID=5CCFccWmJJRAuMzNPjj&search_mode=CitedFullRecord&isickref=WOS:000493810000058.
[14] Bo Xu, Peng Wang, Hongwei Hao, Fangyuan Wang, Heng Zhang, ChengLin Liu, Jiaming. Word embedding clustering and convolutional neural network for short text categorization. ACL-IJCNLP 2015. 2015, http://ir.ia.ac.cn/handle/173211/13510.
[15] Wang Peng, Hao Hongwei, Wang Fangyuan, Zhang Heng, Liu Chenglin, Xu Bo, Xu Jiaming. Semantic Clustering and Convolutional Neural Network for Short Text Categorization. 2015, http://ir.ia.ac.cn/handle/173211/21733.
[16] Peng Wang, Heng Zhang. A Robust Framework for Short Text Categorization based on Topic Model and Integrated Classifier. INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS (IJCNN). 2014, 3534-3539, http://ir.ia.ac.cn/handle/173211/5294.
[17] Xu Jiaming, Xu Bo, Zhao Jun, Tian Guanhua, Zhang Heng, Hao Hongwei. Short Text Hashing Improved by Integrating Topic Features and Tags. NEURAL INFORMATION PROCESSING. 2014, http://ir.ia.ac.cn/handle/173211/10563.
[18] Zhang, Heng, Wang, DaHan, Liu, ChengLin. Character confidence based on N-best list for keyword spotting in online Chinese handwritten documents. PATTERN RECOGNITION[J]. 2014, 47(5): 1880-1890, http://dx.doi.org/10.1016/j.patcog.2013.12.001.
[19] Wang Peng, Zhang Heng, Wu YuFang, Xu Bo, Hao HongWei, IEEE. A Robust Framework for Short Text Categorization based on Topic Model and Integrated Classifier. PROCEEDINGS OF THE 2014 INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS (IJCNN). 2014, 3534-3539, 
[20] Zhang Heng, Liu ChengLin. Keyword Spotting in Online Chinese Handwritten Documents with Candidate Scoring Based on Semi-CRF Model. INTERNATIONAL CONFERENCE ON DOCUMENT ANALYSIS AND RECOGNITION (ICDAR). 2013, 567-571, http://ir.ia.ac.cn/handle/173211/5290.
[21] Zhang, Heng, Wang, DaHan, Liu, ChengLin, Bunke, Horst. KEYWORD SPOTTING FROM ONLINE CHINESE HANDWRITTEN DOCUMENTS USING ONE-VERSUS-ALL CHARACTER CLASSIFICATION MODEL. INTERNATIONAL JOURNAL OF PATTERN RECOGNITION AND ARTIFICIAL INTELLIGENCE[J]. 2013, 27(3): http://dx.doi.org/10.1142/S0218001413530017.
[22] Zhang Heng, Zhou XiangDong, Liu ChengLin, IEEE. Keyword Spotting in Online Chinese Handwritten Documents with Candidate Scoring Based on Semi-CRF Model. 2013 12TH INTERNATIONAL CONFERENCE ON DOCUMENT ANALYSIS AND RECOGNITION (ICDAR)[J]. 2013, 567-571, 
[23] 玛依热依布拉音, 张恒, 刘成林, 艾斯卡尔艾木都拉. 联机手写***文字母识别方法. 模式识别与人工智能[J]. 2012, 25(6): 979-986, http://lib.cqvip.com/Qikan/Article/Detail?id=44473248.
[24] Mayire Ibrahim, Heng Zhang, Askar Hamdulla, ChengLin Liu. An effective character separation method for online cursive Ughur handwriting. CHINESE CONFERENCE ON PATTERN RECOGNITION (CCPR 2012). 2012, CCIS 321, Springer, 2012, pp.530-538., http://ir.ia.ac.cn/handle/173211/13522.
[25] ChengLin Liu, Heng Zhang, Mayire Ibrahim, Askar Hamdulla. A method for online handwritten Uyghur character recognition. CHINESE JOURNAL OF PATTERN RECOGNITION AND ARTIFICIAL INTELLIGENCE[J]. 2012, 25(6): 979-986, http://ir.ia.ac.cn/handle/173211/13431.
[26] Zhang Heng, Wang DaHan, Liu ChengLin, IEEE. A Confidence-Based Method for Keyword Spotting in Online Chinese Handwritten Documents. 2012 21ST INTERNATIONAL CONFERENCE ON PATTERN RECOGNITION (ICPR 2012)[J]. 2012, 525-528, 
[27] Ibrahim, Mayire, Zhang, Heng, Liu, ChengLin, Hamdulla, Askar, Liu, CL, Zhang, CS, Wang, L. An Effective Character Separation Method for Online Cursive Uyghur Handwriting. PATTERN RECOGNITION. 2012, 321: 530-+, 
[28] Zhang Heng, Liu ChengLin. A lattice-based method for keyword spotting in online Chinese handwriting. INTERNATIONAL CONFERENCE ON DOCUMENT ANALYSIS AND RECOGNITION (ICDAR). 2011, 1064-1068, http://ir.ia.ac.cn/handle/173211/4320.
[29] Zhang Heng, Liu ChengLin. Keyword spotting from online Chinese handwritten documents using one-vs-all trained character classifier. INTERNATIONAL CONFERENCE ON FRONTIERS IN HANDWRITING RECOGNITION (ICFHR). 2010, 271-276, http://ir.ia.ac.cn/handle/173211/4342.

科研活动

   
科研项目
( 1 ) 面向数字人文的中文古籍文档图像智能识别与理解——文本识别, 负责人, 国家任务, 2020-01--2024-12
( 2 ) 复杂版面手写图文识别及理解关键技术研究, 参与, 国家任务, 2021-01--2024-12
( 3 ) 城市场景三维感知和语义表达——场景文字识别, 负责人, 国家任务, 2022-01--2025-12
( 4 ) 大规模笔迹数据比对检索模型研究, 负责人, 国家任务, 2023-01--2025-12

指导学生

已指导学生

刘莹  硕士研究生  085211-计算机技术  

于明明  硕士研究生  081203-计算机应用技术  

现指导学生

李天佐  硕士研究生  085404-计算机技术  

凌禄蓉  硕士研究生  081104-模式识别与智能系统  

张敬尧  硕士研究生  085410-人工智能