论文信息 - OCR and post-correction of historical Finnish texts

OCR and post-correction of historical Finnish texts

This paper presents experiments on Optical character recognition (OCR) as a combination of Ocropy software and data-driven spelling correction that uses Weighted Finite-State Methods. Both model training and testing were done on Finnish corpora of historical newspaper text and the best combination of OCR and post-processing models give 95.21% character recognition accuracy.

[1] Vladimir I. Levenshtein,et al. Binary codes capable of correcting deletions, insertions, and reversals , 1965 .

[2] Thomas M. Breuel,et al. The OCRopus open source OCR system , 2008, Electronic Imaging.

[3] Thomas Breuel,et al. Recent progress on the OCRopus OCR system , 2009, MOCR '09.

[4] Faisal Shafait,et al. Document image analysis with OCRopus , 2009, 2009 IEEE 13th International Multitopic Conference.

[5] Rafael Llobet,et al. OCR Post-processing Using Weighted Finite-State Transducers , 2010, 2010 20th International Conference on Pattern Recognition.

[6] Thomas M. Breuel,et al. High-Performance OCR for Printed English and Fraktur Using LSTM Networks , 2013, 2013 12th International Conference on Document Analysis and Recognition.

[7] Uwe Springmann,et al. OCR of historical printings of Latin texts: problems, prospects, progress , 2014, DATeCH '14.

[8] Miikka Silfverberg,et al. Can Morphological Analyzers Improve the Quality of Optical Character Recognition , 2015 .

[9] Miikka Silfverberg,et al. Data-Driven Spelling Correction using Weighted Finite-State Methods , 2016, ACL 2016.

[10] Alexander Mehler,et al. A Comparison of Four Character-Level String-to-String Translation Models for (OCR) Spelling Error Correction , 2016, Prague Bull. Math. Linguistics.