Persian Optical Character Recognition Using Deep Bidirectional Long Short-Term Memory
Source abstract
Abstract
Optical Character Recognition (OCR) is a system of converting images, including text,into editable text and is applied to various languages such as English, Arabic, and Persian. While these languages have similarities, their fundamental differences can create unique challenges. In Persian, continuity between Characters, the existence of semicircles, dots, oblique, and left-to-right characters such as English words in the context are some of the most important challenges in designing Persian OCR systems. Our proposed framework, Bina, is designed in a special way to address the issue of continuity by utilizing Convolution Neural Network (CNN) and deep bidirectional Long-Short Term Memory (BLSTM), a type of LSTM networks that has access to both past and future context. A huge and diverse dataset, including about 2M samples of both Persian and English contexts,consisting of various fonts and sizes, is also generated to train and test the performance of the proposed model. Various configurations are tested to find the optimal structure of CNN and BLSTM. The results show that Bina successfully outperformed state of the art baseline algorithm by achieving about 96% accuracy in the Persian and 88% accuracy in the Persian and English contexts.
Source: crossref · Quality: baseline
Research context
OpenAlex topics: Handwritten Text Recognition Techniques · Vehicle License Plate Recognition · Image Retrieval and Classification Techniques
SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.
Affiliations
School of Computer Science and Statistics, Trinity College Dublin, D02 YY50 Dublin, Ireland
Faculty of New Sciences and Technologies, University of Tehran, Tehran P.O. Box 14399-56191, Iran
School of Computing Science, Simon Fraser University, Burnaby, BC V5A 1S6, Canada
Computer Science Department, School of Computing and Analytics, Northern Kentucky University, Highland Heights, KY 41076, USA
Funding
Computer Science program in the School of Computing and Analytics at Northern Kentucky University
References (52)
Show source-provided references
1. Rejecting character recognition errors using CNN based confidence estimation10.1049/cje.2016.05.018 ↗
2. Saeed, S., Naz, S., and Razzak, M.I. (2019). An application of deep learning in character recognition: An overview. Handbook of Deep Learning Applications, Springer.10.1007/978-3-030-11479-4_3 ↗
3. Smith, R. (1995, January 14–16). A simple and efficient skew detection algorithm via text row accumulation. Proceedings of the 3rd International Conference on Document Analysis and Recognition, Montreal, QC, Canada.
4. A blackboard approach towards integrated Farsi OCR system10.1007/s10032-009-0079-7 ↗
5. Multi-dimensional long short-term memory networks for artificial Arabic text recognition in news video10.1049/iet-cvi.2017.0468 ↗
6. A rule-based post-processing approach to improve Persian OCR performance
7. Malik, S.A., Maqsood, M., Aadil, F., and Khan, M.F. (2019, January 14–15). An efficient segmentation technique for Urdu optical character recognizer (OCR). Proceedings of the Future of Information and Communication Conference, San Francisco, CA, USA.10.1007/978-3-030-12385-7_11 ↗
8. Segmentation free nastalique urdu ocr
9. Segmentation-and annotation-free license plate recognition with deep localization and failure identification10.1109/TITS.2016.2639020 ↗
10. Nastalique segmentation-based approach for Urdu OCR10.1007/s10032-015-0250-2 ↗
11. Deep learning10.1038/nature14539 ↗
12. Deep learning in neural networks: An overview10.1016/j.neunet.2014.09.003 ↗
13. IDM based on image classification with CNN
14. A novel separability objective function in CNN for feature extraction of SAR images10.1049/cje.2018.12.001 ↗
15. Using highway connections to enable deep small-footprint LSTM-RNNs for speech recognition10.1049/cje.2018.11.008 ↗
16. Märgner, V., and El Abed, H. (2012). Offline Arabic Handwriting Recognition with Multidimensional Recurrent Neural Networks. Guide to OCR for Arabic Scripts, Springer.10.1007/978-1-4471-4072-6 ↗
17. Automatic discretization of Arabic text using recurrent neural networks10.1007/s10032-015-0242-2 ↗
18. Smith, R. (2007, January 23–26). An overview of the Tesseract OCR engine. Proceedings of the Ninth International Conference on Document Analysis and Recognition (ICDAR 2007), Curitiba, Brazil.10.1109/ICDAR.2007.4376991 ↗
19. Sarraf, S. (2018). French word recognition through a quick survey on recurrent neural networks using long-short term memory RNN-LSTM. arXiv.
20. Urdu Nasta’liq Text Recognition Using Implicit Segmentation Based on Multi-Dimensional Long Short Term Memory Neural Networks10.1186/s40064-016-3442-4 ↗
21. Ahmad, R., Afzal, M.Z., Rashid, S.F., Liwicki, M., and Breuel, T. (2015, January 23–26). Scale and rotation invariant OCR for Pashto cursive script using MDLSTM network. Proceedings of the 2015 13th International Conference on Document Analysis and Recognition (ICDAR), Tunis, Tunisia.10.1109/ICDAR.2015.7333931 ↗
22. A novel connectionist system for unconstrained handwriting recognition10.1109/TPAMI.2008.137 ↗
23. Liwicki, M., Graves, A., Fernàndez, S., Bunke, H., and Schmidhuber, J. (2007, January 23–26). A novel approach to on-line handwriting recognition based on bidirectional long short-term memory networks. Proceedings of the 9th International Conference on Document Analysis and Recognition, Curitiba, Brazil.
24. Introducing a very large dataset of handwritten Farsi digits and a study on their varieties10.1016/j.patrec.2006.12.022 ↗
25. Alaei, A., Nagabhushan, P., and Pal, U. (2011, January 16–17). A new dataset of Persian handwritten documents and its segmentation. Proceedings of the 2011 7th Iranian Conference on Machine Vision and Image Processing, Tehran, Iran.10.1109/IranianMVIP.2011.6121553 ↗
26. Montazeri, M., and Kiani, R. (2017, January 22–23). PHCWT: A Persian Handwritten Characters, Words and Text dataset. Proceedings of the 2017 10th Iranian Conference on Machine Vision and Image Processing (MVIP), Isfahan, Iran.10.1109/IranianMVIP.2017.8342357 ↗
27. An overview of character recognition focused on off-line handwriting10.1109/5326.941845 ↗
28. Ul-Hasan, A., and Breuel, T.M. (2013, January 24). Can we build language-independent OCR using LSTM networks?. Proceedings of the 4th International Workshop on Multilingual OCR, Washington DC, USA.10.1145/2505377.2505394 ↗
29. Meta features-based scale invariant OCR decision making using LSTM-RNN10.1007/s10588-018-9265-9 ↗
30. Khosravi, H., and Kabir, E. (2006, January 14–15). Introduction of two fast and effective features for handwritten Farsi digit recognition. Proceedings of the 4th Iranian Conference on Machine Vision and Image Processing, Mashhad, Iran.
31. Pourreza, M., Derakhshan, R., Fayyazi, H., and Sabokrou, M. (2018, January 17–19). Sub-word based Persian OCR using auto-encoder features and cascade classifier. Proceedings of the 2018 9th International Symposium on Telecommunications (IST), Tehran, Iran.10.1109/ISTEL.2018.8661146 ↗
32. A Trainless Recognition of Handwritten Persian/Arabic Letters using Primitive Elements
33. Alizadehashraf, B., and Roohi, S. (2017, January 22–23). Persian handwritten character recognition using convolutional neural network. Proceedings of the 2017 10th Iranian Conference on Machine Vision and Image Processing (MVIP), Isfahan, Iran.10.1109/IranianMVIP.2017.8342359 ↗
34. Persian handwritten digit, character and word recognition using deep learning10.1007/s10032-021-00368-2 ↗
35. Huang, G., Liu, Z., Van Der Maaten, L., and Weinberger, K.Q. (2017, January 21–26). Densely connected convolutional networks. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, HI, USA.10.1109/CVPR.2017.243 ↗
36. He, K., Zhang, X., Ren, S., and Sun, J. (2016, January 27–30). Deep residual learning for image recognition. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, NV, USA.10.1109/CVPR.2016.90 ↗
37. Simonyan, K., and Zisserman, A. (2014). Very deep convolutional networks for large-scale image recognition. arXiv.
38. Chimp Optimization Algorithm to Optimize a Convolutional Neural Network for Recognizing Persian/Arabic Handwritten Words
39. Gradient-based learning applied to document recognition10.1109/5.726791 ↗
40. Tesseract-Ocr. Release 5.1.0 · TESSERACT-OCR/tesseract.
41. Ghanbari, N. (2019). A review of research studies on the recognition of Farsi alphabetic and numeric characters in the last decade. Fundamental Research in Electrical Engineering, Springer.10.1007/978-981-10-8672-4_13 ↗
42. Ul-Hasan, A. (2016). Generic Text Recognition Using Long Short-Term Memory Networks. [Ph.D. Thesis, University of Kaiserslautern].
43. Ul-Hasan, A., Shafaity, F., and Liwicki, M. (2015, January 23–26). Curriculum learning for printed text line recognition of ligature-based scripts. Proceedings of the 2015 13th International Conference on Document Analysis and Recognition (ICDAR), Tunis, Tunisia.10.1109/ICDAR.2015.7333912 ↗
44. CNNpred: CNN-based stock market prediction using a diverse set of variables10.1016/j.eswa.2019.03.029 ↗
45. Lee, C.Y., and Osindero, S. (2016, January 27–30). Recursive recurrent nets with attention modeling for ocr in the wild. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, NV, USA.10.1109/CVPR.2016.245 ↗
46. Convolutional recurrent neural networks with hidden Markov model bootstrap for scene text recognition10.1049/iet-cvi.2016.0417 ↗
47. Handwritten Hindi character recognition: A review10.1049/iet-ipr.2017.0184 ↗
48. Framewise phoneme classification with bidirectional LSTM and other neural network architectures10.1016/j.neunet.2005.06.042 ↗
49. Graves, A. (2012). Supervised sequence labelling. Supervised Sequence Labelling with Recurrent Neural Networks, Springer.10.1007/978-3-642-24797-2 ↗
50. Segmentation-free optical character recognition for printed Urdu text10.1186/s13640-017-0208-z ↗
51. Khosrobeigi, Z. Real dataset for Persain OCR-V1. 2020.
52. Data and Signal Processing Laboratory. The University of Tehran. Iran.