Topic based language models for OCR correction

ABSTRACT

Despite several decades of research in document analysis, recognition of unconstrained handwritten documents is still considered a challenging task. Previous research in this area has shown that word recognizers produce reasonably clean output when used with a restricted lexicon. But in absence of such a restricted lexicon, the output of an unconstrained handwritten word recognizer is noisy. The objective of this research is to process noisy recognizer output and eliminate spurious recognition choices using a topic based language model. We construct a topic based language model for every document using a training data which is manually categorized. A topic categorization sub-system based on Maximum Entropy model is also trained which is used to generate the topic distribution of a test document. A given test word image is processed by the recognizer and its word recognition likelihood is refined by incorporating topic distribution of the document and topic based language model probability. The proposed method is evaluated on a publicly available IAM dataset and experimental results show significant improvement in the word recognition accuracy from 32% to 40% over a test set consisting of 4033 word images extracted from 70 handwritten document images.

REFERENCES

• 1: Stochastic Error-Correcting Parsing for OCR Post-Processing, Proceedings of the International Conference on Pattern Recognition, p.4405, September 03-08, 2000
• 2: F. Farooq, D. Jose and V. Govindaraju, Phrase Based Direct Model for Improving Handwriting Recognition Accuracies, To appear in International Conference on Frontiers in Handwriting Recognition, 2008, Montreal, Canada.
• 3: F. Farooq, G. Chandalia and V. Govindaraju. Lexicon Reduction in Handwriting Recognition Using Topic Categorization. Under Review - In Eight International Workshop on Document Analysis Systems. Nara, Japan, 2008.
• 4: S. Madhvanath , V. Govindaraju , V. Ramanaprasad , D. S. Lee , S. N. Srihari, Reading handwritten US census forms, Proceedings of the Third International Conference on Document Analysis and Recognition (Volume 1), p.82, August 14-15, 1995
• 5: N. D. Guillevic D and Y. K. Word lexicon reduction by character spotting. Proceedings of the Seventh International Workshop on Frontiers in Handwriting Recognition. pages 373--382, 2000.
• 6: S. Impedovo, P. Wang, and H. Bunke. Automatic bankcheck processing. Machine Perception and Artificial Intelligence, 28, 1997.
• 7: Guido Kaufmann , Horst Bunke , M. Hadorn, Lexicon Reduction in an HMM-Framework Based on Quantized Feature Vectors, Proceedings of the 4th International Conference on Document Analysis and Recognition, p.1097-1101, August 18-20, 1997
• 8: Gyeonghwan Kim , Venu Govindaraju, A Lexicon Driven Approach to Handwritten Word Recognition for Real-Time Applications, IEEE Transactions on Pattern Analysis and Machine Intelligence, v.19 n.4, p.366-379, April 1997 [doi>10.1109/34.588017]
• 9: G. Kim, V. Govindaraju, and S. Srihari. Architecture for handwriting recognition systems. International Journal of Document Analysis and Recognition, 2(1):37--44, 1999.
• 10: A. Koerich, R. Sabourin, and C. Suen. Large vocabulary offline handwriting recognition using a constrained level building algorithm. Pattern Analysis and Applications, 6(2):97--121, 2003.
• 11: Karen Kukich, Technique for automatically correcting words in text, ACM Computing Surveys (CSUR), v.24 n.4, p.377-439, Dec. 1992 [doi>10.1145/146370.146380]
• 12: S. Madhvanath and V. Govindaraju. Holistic lexicon reduction for handwritten word recognition. In Proceedings of the SPIE - Document Recognition III, pages 224--234, San Jose, CA, 1996.
• 13: S. Madhvanath and V. Govindaraju. Syntatic methodology of pruning large lexicons in cursive script recognition. Pattern Recognition, 34(1):37--46, January 2001.
• 14: Christopher D. Manning , Prabhakar Raghavan , Hinrich Schtze, Introduction to Information Retrieval, Cambridge University Press, New York, NY, 2008
• 15: U. Marti and H. Bunke. The iam-database: an english sentence database for off-line handwriting recognition. International Journal on Document Analysis and Recognition, 5:39--46, 2002.
• 16: U. Pal, P. Kundu and B. Chaudhuri, OCR error correction of an inflectional Indian language using morphological parsing, Journal of Information Science and Engineering, 16(6):903--922, 2000.
• 17: N. S. R. K. Powalka and R. J. Whitrow. Word shape analysis for a hybrid recognition system. Pattern Recognition, 30(3):421--445, March 1997.
• 18: Sargur N. Srihari , Edward J. Kuebert, Integration of hand-written address interpretation technology into the United States Postal Service Remote Computer Reader system, Proceedings of the 4th International Conference on Document Analysis and Recognition, p.892-896, August 18-20, 1997
• 19: K. Taghva and E. Stofsky. 2001. OCRSpell: an interactive spelling correction system for OCR errors in text. International Journal on Document Analysis and Recognition, 3(3):125--137.
• 20: Alessandro Vinciarelli , Samy Bengio , Horst Bunke, Offline Recognition of Unconstrained Handwritten Texts Using HMMs and Statistical Language Models, IEEE Transactions on Pattern Analysis and Machine Intelligence, v.26 n.6, p.709-720, June 2004 [doi>10.1109/TPAMI.2004.14]

INDEX TERMS

Primary Classification:
  I. Computing Methodologies
  I.7 DOCUMENT AND TEXT PROCESSING
      I.7.5 Document Capture
          Subjects: Optical character recognition (OCR)

Keywords:
OCR correction, document analysis, dynamic lexicon, language models, topic categorization