nach oben

Erschienen in:

2012 | OriginalPaper | Buchkapitel

8. Handwritten Arabic Word Recognition Using the IFN/ENIT-database

verfasst von : Mario Pechwitz, Haikal El Abed, Volker Märgner

Erschienen in: Guide to OCR for Arabic Scripts

Verlag: Springer London

Einloggen

Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.

search-config

KI-gestützte Suche

Aus

Abstract

A well-structured and comprehensive dataset is the most important part in the development of a handwritten word recognizer. The IFN/ENIT-database, a well-organized set of images of Arabic handwritten town names, is widely used as a basis to develop handwritten Arabic word recognition systems. We describe in detail the IFN/ENIT-database, the form used to collect the data, the ground truth information, and the statistics of the data. The recognizer developed using this database is presented in the main part of this contribution. The pre-processing of the name images, e.g., baseline estimation, normalization, and feature extraction as well as the hidden Markov model (HMM)-based recognizer together with the results achieved are presented and discussed in detail in this chapter.

Sie haben noch keine Lizenz? Dann Informieren Sie sich jetzt über unsere Produkte:

Springer Professional "Wirtschaft+Technik"

Online-Abonnement

Mit Springer Professional "Wirtschaft+Technik" erhalten Sie Zugriff auf:

über 102.000 Bücher
über 537 Zeitschriften

aus folgenden Fachgebieten:

Automobil + Motoren
Bauwesen + Immobilien
Business IT + Informatik
Elektrotechnik + Elektronik
Energie + Nachhaltigkeit
Finance + Banking
Management + Führung
Marketing + Vertrieb
Maschinenbau + Werkstoffe
Versicherung + Risiko

Jetzt Wissensvorsprung sichern!

Jetzt informieren

Springer Professional "Technik"

Online-Abonnement

Mit Springer Professional "Technik" erhalten Sie Zugriff auf:

über 67.000 Bücher
über 390 Zeitschriften

aus folgenden Fachgebieten:

Automobil + Motoren
Bauwesen + Immobilien
Business IT + Informatik
Elektrotechnik + Elektronik
Energie + Nachhaltigkeit
Maschinenbau + Werkstoffe

Jetzt Wissensvorsprung sichern!

Jetzt informieren

Springer Professional "Wirtschaft"

Online-Abonnement

Mit Springer Professional "Wirtschaft" erhalten Sie Zugriff auf:

über 67.000 Bücher
über 340 Zeitschriften

aus folgenden Fachgebieten:

Bauwesen + Immobilien
Business IT + Informatik
Finance + Banking
Management + Führung
Marketing + Vertrieb
Versicherung + Risiko

Jetzt Wissensvorsprung sichern!

Jetzt informieren

Vorheriges Kapitel Printed Arabic Text Recognition

Nächstes Kapitel RWTH OCR: A Large Vocabulary Optical Character Recognition System for Arabic Scripts

From the 946 Tunisian town/village names we obtain only 937 really different words (some have only a different postcode). The classification task has to deal with a middle-sized lexicon.

The criteria for the selection are explained in Sect. 8.2.4.

The additional information like name, residence, age, and profession were vital to keep track of who filled out which form. This information is needed to build writer disjunct sets from the collected data. Otherwise, the same writer could probably contribute to the learn and to the test set; this would contort the tests and had to be avoided. To gather statistical parameters about the writers who contributed to the database, the additional information from the form was also very helpful.

For example, the writer is coded in the image file name, and the age, profession, and writing quality were used to arrange writers into groups and divide them uniformly over the sets (Sect. 8.2.4).

If it was not possible for whatever reason to give an “acceptable” baseline during the verification task, the quality flag for the baseline was set to “bad.” Also, the writing quality could be marked as “bad.”

All information refers to IFN/ENIT-database version 1.0p2 (www.ifnenit.com). The database is in ongoing development.

To optimize the database in relation to equally distributed town/village names was not an option, because the effort would be too huge. For example, if we only assume 100 times for each town/village name, we would need about 1600 writers!

The topline estimation is discussed separately in Sect. 8.3.4.

Some town/village names consist of several isolated words which differentiate themselves only slightly from “normal” PAWs.

Figure 8.7b shows examples where the baseline is marked.

If the baseline quality mark within the GT is set to “bad,” it signals that there was a problem in defining a sufficient baseline position, at least with a straight line.

Average word image height was approximately 100 pixels, 8.5 mm; for details see Sect. 8.2.4.

The results depend on some parameters (see Fig. 8.12 and Fig. 8.13).

Baseline error ≤7 pixels.

The parameter for thresholding and the clustering were set to values that result in not more than ten “local” maxima (baseline position candidates) in the Hough space.

Considering “local” maximum in the Hough space.

With regard to the vocabulary and considering the frequency and the approximate font size of the words; using the common Naskh style.

Based on global maximum detection within the filtered Hough space.

It is strongly dependent on the dataset.

We will call the feature sets A and B.

The IFN/ENIT-database version consists of four equally sized sets, a, b, c, d (cf. Sect. 8.2.4).

Abandah, G., Jamour, F.: Recognizing handwritten Arabic script through efficient skeleton-based grapheme segmentation algorithm. In: Proceedings of the 10th International Conference on Intelligent Systems Design and Applications (ISDA), pp. 977–982 (2010) CrossRef

Abandah, G., Malas, T.: Feature selection for recognizing handwritten Arabic letters. Dirasat Eng. Sci. 37(2), 242–256 (2010)

Al-Hajj, R., Likforman-Sulem, L., Mokbel, C.: Combining slanted-frame classifiers for improved HMM-based Arabic handwriting recognition. IEEE Trans. Pattern Anal. Mach. Intell. 31(7), 1165–1177 (2009) CrossRef

Capson, D.W.: An improved algorithm for the sequential extraction of boundaries from a raster scan. Comput. Vis. Graph. Image Process. 28, 109–125 (1984) CrossRef

Chen, J., Cao, H., Prasad, R., Bhardwaj, A., Natarajan, P.: Gabor features for offline Arabic handwriting recognition. In: Proceedings of the 9th International Workshop on Document Analysis Systems (DAS), pp. 53–58 (2010) CrossRef

Dehghan, M., Faez, K., Ahmadi, M., Shridhar, M.: Handwritten Farsi (Arabic) word recognition: a holistic approach using discrete HMM. Pattern Recognit. 34(5), 1057–1065 (2001) MATHCrossRef

Ding, X., Jin, J., Wang, H., Peng, L.: Printed Arabic document recognition system. In: Proceedings of SPIE—Document Recognition and Retrieval XII, vol. 5676, pp. 48–55 (2005)

Dreuw, P., Jonas, S., Ney, H.: White-space models for offline Arabic handwriting recognition. In: Proceedings of the International Conference on Pattern Recognition (ICPR), pp. 1–4 (2008)

Dreuw, P., Heigold, G., Ney, H.: Confidence-based discriminative training for model adaptation in offline Arabic handwriting recognition. In: Proceedings of the International Conference on Document Analysis and Recognition (ICDAR), pp. 596–600 (2009)

10.

El Abed, H., Märgner, V.: Comparison of different pre-processing methods for offline recognition of handwritten Arabic words. In: Proceedings of the 9th International Conference on Document Analysis and Recognition (ICDAR), vol. 2, pp. 974–978 (2007)

11.

El Abed, H., Märgner, V.: Improvement of Arabic handwriting recognition systems—combination and/or reject? In: Proceedings of the Document Recognition and Retrieval XVI. SPIE Proc., vol. 7247, pp. 10 (2009)

12.

El Abed, H., Märgner, V.: ICDAR 2009—Arabic handwriting recognition competition. Int. J. Doc. Anal. Recognit. 14(1), 3–13 (2011). Special Issue on Performance Evaluation CrossRef

13.

Elbaati, A., Kherallah, M., El Abed, H., Ennaji, A., Alimi, A.M.: Arabic handwriting recognition using restored stroke chronology. In: International Conference on Document Analysis and Recognition (ICDAR) (2009)

14.

Ferreira, A., Ubeda, S.: Ultra fast parallel contour tracking with application to thinning. Pattern Recognit. 27(7), 867–878 (1994) CrossRef

15.

Graves, A.: Supervised sequence labelling with recurrent neural networks. Ph.D. thesis, Fakultat für Informatik—Technische Universität München (2007)

16.

Graves, A., Schmidhuber, J.: Offline handwriting recognition with multidimensional recurrent neural networks. In: Advances in Neural Information Processing Systems, vol. 21 (2009)

17.

Gray, R.A.: Vector quantization. IEEE ASSP Mag. 1, 4–28 (1984) CrossRef

18.

Hamdani, M., El Abed, H., Kherallah, M., Alimi, A.M.: Combining multiple HMMs using on-line and off-line features for off-line Arabic handwriting recognition. In: Proceedings of the 10th International Conference on Document Analysis and Recognition (ICDAR), pp. 201–205 (2009) CrossRef

19.

Hamdani, M., El Abed, H., Hamdani, T.M., Märgner, V., Alimi, A.M.: Improving a HMM-based off-line handwriting recognition system using MME-PSO optimization. In: Proceedings of the Document Recognition and Retrieval XVIII (2011)

20.

Heigold, G., Deselaers, T., Schlüter, R., Ney, H.: Modified MMI/MPE: a direct evaluation of the margin in speech recognition. In: Proceedings of the International Conference on Machine Learning, pp. 384–391 (2008) CrossRef

21.

Huang, X.D., et al.: Hidden Markov Models for Speech Recognition. Edinburgh Universal Press, Edinburgh (1990)

22.

Impedovo, S., Ottiviano, L., Occhinegro, S.: Optical character recognition—a survey. Int. J. Pattern Recognit. Artif. Intell. 5(1), 1–24 (1991) CrossRef

23.

Märgner, V., Pechwitz, M., El-Abed, H.: ICDAR 2005 Arabic handwriting recognition competition. In: Eighth International Conference on Document Analysis and Recognition (ICDAR’05), vol. 1, pp. 70–74 (2005) CrossRef

24.

Pechwitz, M.: Automatische Erkennung handgeschriebener arabischer Wörter. Ph.D. thesis, Technische Universität Braunschweig, Germany (2005)

25.

Pechwitz, M., Maddouri, S.S., Märgner, V., Ellouze, N., Amiri, H.: IFN/ENIT—database of handwritten Arabic words. In: Proc. of Colloque International Francophone sur l’Ecrit et le Document, CIFED 2002, Hammamet, Tunisia, October 21–23, 2002, pp. 129–136 (2002)

26.

Plamondon, R., Srihari, S.N.: On-line and off-line handwriting recognition: a comprehensive survey. IEEE Trans. Pattern Anal. Mach. Intell. 22(1), 63–84 (2000) CrossRef

27.

Rabiner, L.R.: A tutorial on hidden Markov models and selected applications in speech recognition. In: Waibel, A., Lee, K.-F. (eds.) Readings in Speech Recognition, pp. 267–296. Morgan Kaufmann, San Mateo (1990)

28.

Zamperoni, P.: Methoden der Digitalen Bildsignalverarbeitung. Vieweg, Braunschweig (1989) CrossRef

29.

Zamperoni, P., Klette, R.: Handbuch der Operatoren der Bildverarbeitung. Vieweg, Braunschweig (1995)

Titel: Handwritten Arabic Word Recognition Using the IFN/ENIT-database
verfasst von: Mario Pechwitz
Haikal El Abed
Volker Märgner
Verlag: Springer London
Buch: Guide to OCR for Arabic Scripts
Print ISBN: 978-1-4471-4071-9

Electronic ISBN: 978-1-4471-4072-6

Copyright-Jahr: 2012
DOI: https://doi.org/10.1007/978-1-4471-4072-6_8

Springer Professional

Abstract

Bitte loggen Sie sich ein, um Zugang zu Ihrer Lizenz zu erhalten.

Sie haben noch keine Lizenz? Dann Informieren Sie sich jetzt über unsere Produkte:

Springer Professional "Wirtschaft+Technik"

Springer Professional "Technik"

Springer Professional "Wirtschaft"

Premium Partner