Woodblock-Printing Mongolian Words Recognition by Bi-LSTM with Attention Mechanism
作者： Yanke Kang , Hongxi Wei , Hui Zhang , Guanglai Gao
关键词:
摘要: Woodblock-printing Mongolian documents are seriously degraded due to aging. Therefore, it is difficult segment woodblock-printing words into individual glyphs. In this paper, a holistic recognition approach based on sequence model has been proposed for the words. The input of frames wood-block printing word. order generating corresponding frames, each word image should be normalized same sizes in advance. And then, segmented several fragments with equal size along writing direction. output letters. To specific, contains three parts: an encoder, decoder and attention network. encoder consists deep neural network bi-directional Long Short-Term Memory (Bi-LSTM). (LSTM) softmax layer. connected by network, which can map multiple one letter. Experimental results demonstrate that outperforms segmentation method.
sci-hub.se 参考文章(13)
Xiangdong Su, Guanglai Gao, Weihua Wang, Feilong Bao, Hongxi Wei, Character Segmentation for Classical Mongolian Words in Historical Documents Communications in Computer and Information Science. pp. 464- 473 ,(2014) , 10.1007/978-3-662-45643-9_49
Hongxi Wei, Guanglai Gao, A keyword retrieval system for historical Mongolian document images International Journal on Document Analysis and Recognition. ,vol. 17, pp. 33- 45 ,(2014) , 10.1007/S10032-013-0203-6
Guanglai Gao, Xiangdong Su, Hongxi Wei, Yeyun Gong, Classical Mongolian Words Recognition in Historical Document international conference on document analysis and recognition. pp. 692- 697 ,(2011) , 10.1109/ICDAR.2011.145
Hongxi Wei, Guanglai Gao, Yulai Bao, Yali Wang, An efficient binarization method for ancient Mongolian document images international conference on advanced computer theory and engineering. ,vol. 2, ,(2010) , 10.1109/ICACTE.2010.5579111
Chen-Yu Lee, Simon Osindero, Recursive Recurrent Nets with Attention Modeling for OCR in the Wild computer vision and pattern recognition. pp. 2231- 2239 ,(2016) , 10.1109/CVPR.2016.245
Xiangdong Su, Guanglai Gao, Hongxi Wei, Feilong Bao, A knowledge-based recognition system for historical Mongolian documents International Journal on Document Analysis and Recognition. ,vol. 19, pp. 221- 235 ,(2016) , 10.1007/S10032-016-0267-1
Xiangdong Su, Guanglai Gao, Hongxi Wei, Feilong Bao, Enhancing the Mongolian Historical Document Recognition System with Multiple Knowledge-Based Strategies international conference on neural information processing. pp. 536- 544 ,(2015) , 10.1007/978-3-319-26535-3_61
Yonghui Wu, Mike Schuster, Zhifeng Chen, Quoc V Le, Mohammad Norouzi, Wolfgang Macherey, Maxim Krikun, Yuan Cao, Qin Gao, Klaus Macherey, Jeff Klingner, Apurva Shah, Melvin Johnson, Xiaobing Liu, Łukasz Kaiser, Stephan Gouws, Yoshikiyo Kato, Taku Kudo, Hideto Kazawa, Keith Stevens, George Kurian, Nishant Patil, Wei Wang, Cliff Young, Jason Smith, Jason Riesa, Alex Rudnick, Oriol Vinyals, Greg Corrado, Macduff Hughes, Jeffrey Dean, None, Google's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation arXiv: Computation and Language. ,(2016)
Riaz Ahmad, M. Zeshan Afzal, S. Faisal Rashid, Marcus Liwicki, Thomas Breuel, Andreas Dengel, KPTI: Katib's Pashto Text Imagebase and Deep Learning Benchmark international conference on frontiers in handwriting recognition. pp. 453- 458 ,(2016) , 10.1109/ICFHR.2016.0090
Hongxi Wei, Hui Zhang, Guanglai Gao, Representing word image using visual word embeddings and RNN for keyword spotting on historical document images international conference on multimedia and expo. pp. 1368- 1373 ,(2017) , 10.1109/ICME.2017.8019403