This Article 
 Bibliographic References 
 Add to: 
Keyword Spotting in Poorly Printed Documents using Pseudo 2-D Hidden Markov Models
August 1994 (vol. 16 no. 8)
pp. 842-848

An algorithm for robust machine recognition of keywords embedded in a poorly printed document is presented. For each keyword, two statistical models, called pseudo 2-D hidden Markov models, are created for representing the actual keyword and all the other extraneous words, respectively. Dynamic programming is then used for matching an unknown input word with the two models and for making a maximum likelihood decision. Although the models are pseudo 2-D in the sense that they are not fully connected 2-D networks, they are shown to be general enough in characterizing printed words efficiently. These models facilitate a nice "elastic matching" property in both horizontal and vertical directions, which makes the recognizer not only independent of size and slant but also tolerant of highly deformed and noisy words. The system is evaluated on a synthetically created database that contains about 26000 words. Currently, the authors achieve a recognition accuracy of 99% when words in testing and training sets are of the same font size, and 96% when they are in different sizes. In the latter case, the conventional 1-D HMM achieves only a 70% accuracy rate.

[1] C. Bose and S. Kuo, "Recognition of degraded and connected text using Hidden Markov Models," inProc. 11th Int. Conf. Part. Recognition, 1992.
[2] L. Rabiner, "A tutorial on hidden Markov models and selected applications in speech recognition,"Proc. IEEE, vol. 77, pp. 257-286, Feb. 1989.
[3] Y. He, M.-Y. Chen, and A. Kundu, "Handwritten word recognition using HMM with adaptive length Viterbi algorithm," inProc. IEEE Int. Conf. Acoust., Speech, Signal Processing, 1992, pp. III. 153-156.
[4] A. Belaid and J. Anigbogu, "Text recognition using stochastic models," inProc. 5th Int. Symp. Applied Stochastic Models and Data Analysis, Granada, Spain, 1991.
[5] S. Kuo and O. Agazzi, "Automatic keyword recognition using hidden markov models," to be published inJ. Visual Commun., Image Representation, vol. 5, no. 3, Sept. 1994.
[6] E. Levin and R. Pieraccini, "Dynamic planar warping for optical character recognition," inProc. ICASSP'92, 1992, pp. III149-III152.
[7] H.S. Baird, "Anatomy of a versatile page reader,"Proc. IEEE, vol. 80, pp. 1059-1065, July 1992.
[8] F. Chen, L. Wilcox, and D. Bloomberg, "Word spotting in scanned images using hidden Markov models," inProc. 1993 IEEE Int. Conf. Acoust., Speech and Signal Processing, Minneapolis, MN, vol. V, Apr. 27-30, 1993, pp. 1-4.
[9] D. S. Batory, "A model of transactions on physical databases,"ACM Trans. Database Syst., vol. 7, no. 4, pp. 509-539, Dec. 1982.
[10] S. Levinson, L. Rabiner, and M. Sondhi, "An introduction to the application of the theory of probabilistic functions of a Markov process to automatic speech recognition,"Bell Syst. Tech. J., vol. 62, pp. 1035-1074, Apr. 1983.
[11] O. E. Agazzi and S. Kuo, "Hidden Markov Model based optical character recognition in the presence of deterministic transformations,"Pattern Recognit., vol. 26, no. 12, pp. 1813-1826, Dec. 1993.

Index Terms:
hidden Markov models; dynamic programming; optical character recognition; statistical models; maximum likelihood estimation; decision theory; document image processing; keyword spotting; poorly printed documents; pseudo 2-D hidden Markov models; robust machine recognition; statistical models; dynamic programming; maximum likelihood decision; elastic matching; recognition accuracy; testing; training sets
S.S. Kuo, O.E. Agazzi, "Keyword Spotting in Poorly Printed Documents using Pseudo 2-D Hidden Markov Models," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 16, no. 8, pp. 842-848, Aug. 1994, doi:10.1109/34.308482
Usage of this product signifies your acceptance of the Terms of Use.