中国科学院机构知识库网格
Chinese Academy of Sciences Institutional Repositories Grid
首页 机构 成果 学者
  
  1. CAS IR Grid
  2. 自动化研究所
  3. 中国科学院自动化研究所
  4. 复杂系统管理与控制国家重点实验室
  5. 影像分析与机器视觉团队
Deep Contextual Stroke Pooling for Scene Character Recognition

文献类型：期刊论文

作者Zhang, Zhong1,2; Wang, Hong1,2; Liu, Shuang1,2; Xiao, Baihua3
刊名IEEE ACCESS
出版日期2018
卷号6页码:16454-16463
关键词Scene Character Recognition Deep Contextual Stroke Pooling Contextual Factor
DOI10.1109/ACCESS.2018.2817342
文献子类Article
英文摘要Characters, as a kind of symbols carrying rich semantic information, are composed of strokes arranged in a certain structure and are of great significance in our daily life. In this paper, we are concerned with the problem of scene character recognition, and study the problem from the perspective of feature representation. We propose a novel pooling method termed deep contextual stroke pooling (DCSP) for scene character recognition. The proposed DCSP discovers the most prominent stroke information by using stroke detectors and captures the spatial context of discriminative strokes by learning contextual factor. Specifically, we first utilize the convolutional summing map in one convolutional layer to select discriminative strokes and use the convolutional activation features of discriminative strokes to train stroke detectors. Then, we propose the contextual factor to represent the co-occurrence probability of the stroke and its location. Finally, in the response regions, we incorporate the contextual factor into the detector scores and obtain the deep contextual confidence vectors of scene characters. Extensive experiments are conducted on three databases, i.e., ICDAR2003, Chars74k, and SVIIN, and the experimental results demonstrate that our method achieves higher accuracies than the state-of-the-art methods.
WOS关键词TEXT RECOGNITION ; IMAGE ; REPRESENTATION ; GESTURES
WOS研究方向Computer Science ; Engineering ; Telecommunications
语种英语
WOS记录号WOS:000429991600001
资助机构National Natural Science Foundation of China(61501327 ; Natural Science Foundation of Tianjin(17JCZDJC30600 ; Open Projects Program of the National Laboratory of Pattern Recognition(201700001 ; China Scholarship Council(201708120039 ; 61711530240) ; 15JCQNJC01700) ; 201800002) ; 201708120040)
源URL[http://ir.ia.ac.cn/handle/173211/21998]  
专题自动化研究所_复杂系统管理与控制国家重点实验室_影像分析与机器视觉团队
作者单位1.Tianjin Normal Univ, Tianjin Key Lab Wireless Mobile Commun & Power Tr, Tianjin 300387, Peoples R China
2.Tianjin Normal Univ, Coll Elect & Commun Engn, Tianjin 300387, Peoples R China
3.Chinese Acad Sci, Inst Automat, State Key Lab Management & Intelligent Control Co, Beijing 100190, Peoples R China
推荐引用方式
GB/T 7714		 Zhang, Zhong,Wang, Hong,Liu, Shuang,et al. Deep Contextual Stroke Pooling for Scene Character Recognition[J]. IEEE ACCESS,2018,6:16454-16463.
APA Zhang, Zhong,Wang, Hong,Liu, Shuang,&Xiao, Baihua.(2018).Deep Contextual Stroke Pooling for Scene Character Recognition.IEEE ACCESS,6,16454-16463.
MLA Zhang, Zhong,et al."Deep Contextual Stroke Pooling for Scene Character Recognition".IEEE ACCESS 6(2018):16454-16463.

入库方式： OAI收割

来源：自动化研究所

浏览0
下载0
收藏0
其他版本

除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。