Abstract
In the paper a new method of handwritten word recognition is described. In this method different probabilistic character language models (PCLMs) are used in order to improve the word recognition accuracy. These models consist of conditional probability distributions of characters given that preceding, succeeding or surrounding characters are known. The application of character succession and precedence in word recognition was explored in many earlier works. The novelty of the method proposed here consists in utilizing also the conditional probabilities based on character appearing of both sides of the letter being recognized and in combining such classifier with other ones based on character precedence and succession. This new classifier that uses two sided character neighborhood is realized as an iterative procedure which starts with character support factors evaluated independently for each character in the word by a soft character classifier. Then in each step of this iterative procedure the support factors are calculated as Bayes-like formula using needed probabilities. Next, the support factors obtained at the character level are used to calculate the support factors for complete words. Finally, the soft word recognition results obtained for classifiers based on three different PCLMs are combined. Experiments described in the paper show the superiority of the combined method over all its simpler components.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
Brakensiek A., Rottland J., Kosmala A., Rigoll G.: Off-Line Handwriting Recognition Using Various Hybrid Modeling Techniques and Character N-Grams, Proc. of the Seventh Int. Workshop on Frontiers in Handwriting Recognition, (2000) 343–352
Kuncheva L.: Combining Classifiers: Soft Computing Solutions, [in.] Pattern Recognition: from Classical to Modern Approaches, Pal S., Pal A. [eds.], World Scientific (2001) 427–451
Marti U.V., Bunke H.: Using a Statistical Language Model to Improve the Performance of an HMM-Based Cursive Handwritting Recognition System, Int. Journ. of Pattern Recognition and Artificial Intelligence, Vol. 15 (2001) 65–90
Liu C., Nakashima K., Sako H.: Handwritten Digit Recognition: Benchmarking of State-of-the-Art Techniques. Pattern Recognition, Vol. 36. (2003) 2271–2285
El-Nasan A., Nagy G., Veeramachaneni S.: Handwriting recognition using position sensitive n-gram matching, Proc. 7th Int. Conf. on Document Analysis and Recognition (2003) 577–582
Vinciarelli A., Bengio S., Bunke H.: Offline Recognition of Unconstrained Handwritten Text Using HMMs and Statistical Language Models, IEEE Trans, on PAMI, Vol. 26 (2004) 709–720
Sas J., Luzyna M.: Combining Character Classifier Using Member Classifiers Assessment, Proc. of 5th Int. Conf. on Intelligent Systems Design and Applications, ISDA 2005, IEEE Press (2005) 400–405
Sas J.: Application of Bidirectional Probabilistic Character Language Model in Handwritten Word Recognition, Proc. of 7th Int. Conf. on Intelligent Data Engineering and Automated Learning, Springer (LNCS) (2007) 679–687
Kurzynski M., Zolnierek A.: Sequential pattern recognition: naive Bayes versus fuzzy relation method, Proc. of Int. Conf. on Computational Intelligence for Modelling, Control and Automation CIMCA 2005, IEEE Press, (2006) 1165–1170
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2007 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Sas, J., Zolnierek, A. (2007). Handwritten Word Recognition with Combined Classifier Based on Tri-grams. In: Kurzynski, M., Puchala, E., Wozniak, M., Zolnierek, A. (eds) Computer Recognition Systems 2. Advances in Soft Computing, vol 45. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-540-75175-5_60
Download citation
DOI: https://doi.org/10.1007/978-3-540-75175-5_60
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-75174-8
Online ISBN: 978-3-540-75175-5
eBook Packages: EngineeringEngineering (R0)