This Article 
   
 Share 
   
 Bibliographic References 
   
 Add to: 
 
Digg
Furl
Spurl
Blink
Simpy
Google
Del.icio.us
Y!MyWeb
 
 Search 
   
Validation of Image Defect Models for Optical Character Recognition
February 1996 (vol. 18 no. 2)
pp. 99-108

Abstract—In this paper, we consider the problem of evaluating character image generators that model distortions encountered in optical character recognition (OCR). While a number of such defect models have been proposed, the contention that they produce the desired result is typically argued in an ad hoc and informal way. We introduce a rigorous and more pragmatic definition of when a model is accurate: we say a defect model is validated if the OCR errors induced by the model are indistinguishable from the errors encountered when using real scanned documents. We describe four measures to quantify this similarity, and compare and contrast them using over ten million scanned and synthesized characters in three fonts. The measures differentiate effectively between different fonts and different scans of the same font regardless of the underlying text.

[1] S.V. Rice, J. Kanai, and T. Nartker, "An evaluation of OCR accuracy," Annual Report of UNLV Information Science Research Institute, Las Vegas, Nev., pp. 9-34, 1993.
[2] R. Bradford, "Technical factors in the creation of large full-text databases," Proc. DOE Infotech Conf.,Oak Ridge, Tenn., May 1991.
[3] H.S. Baird, "Document Image Defect Models and Their Uses," Proc. Second Int'l Conf. Document Analysis and Recognition, pp. 62-67,Tsukuba Science City, Japan, Oct.20-22, 1993.
[4] Q.R. Wang, Y.X. Gu, and C.Y. Suen, "Applications of multi-layer decision tree in computer recognition of Chinese characters," IEEE Trans. Pattern Anal. Machine Intell., vol. 5, pp. 83-89, 1983.
[5] T. Pavlidis, "Effects of distortions on the recognition rate of a structural OCR system," Proc. Conf. Comp. Vision and Pattern Recog.,Washington, D.C., pp. 303-309, 1983.
[6] H.S. Baird and R. Fossey, "A 100-font classifier,", Proc. Int'l Conf. Doc. Anal. and Recog.,St. Malo, France, pp. 332-339, 1991.
[7] H.S. Baird, "Document Image Defect Models and Their Uses," Proc. Second Int'l Conf. Document Analysis and Recognition, pp. 62-67,Tsukuba Science City, Japan, Oct.20-22, 1993.
[8] R.M. Haralick, "English document database design and implementation methodology," Proc. Symp. Doc. Anal. and Info. Ret.,Las Vegas, Nev., pp. 65-104, Apr. 1993.
[9] F. Jenkins, J. Kanai, and T. Nartker, "Using ideal images to establish a baseline of OCR performance," Annual Report of UNLV Information Science Research Institute, Las Vegas, Nev., 1993.
[10] S. Khoubyari and J.J. Hull, "Keyword location in noisy document image," Proc. Symp. Doc. Anal. and Info. Ret.,"Las Vegas, Nev., pp. 217-232, Apr. 1993.
[11] T. Kanungo, R.M. Haralick, and I. Phillips, “Global and Local Document Degradation Models,” Proc. Second Int'l Conf. Document Analysis and Recognition, pp. 730-734, Oct. 1993.
[12] M. Buchman, "Distortion modeling for document images," Proc. DIMUND Work. on Page Decomp., Char. Recog., and Data Standards,Harper's Ferry, W.Va., Aug. 1993.
[13] G. Nagy, "On the auto-correlation function of noise in sampled typewritten characters," IEEE Region III Conv. Record,New Orleans, La., 1968.
[14] D. Lopresti, G. Nagy, P. Sarkar, and J. Zhou, "Spatial Sampling effects in optical character recognition," Proc. Int'l Conf. Doc. Anal. and Rec. (to appear), Montreal, Canada, Aug. 1995.
[15] K. Ishii, "Design of a recognition dictionary using artificially distorted characters," Systems and Computers in Japan, vol. 21, no. 9, pp. 35-44, 1990.
[16] M. Parizeau and R. Plamondon,“A handwriting model for syntactic recognition of cursive script,” Proc. 11th Int’l Conf. Pattern Recognition, The Hague, vol. II, pp. 308-312, 1992.
[17] T. Kanungo et al., "Document degradation models: Parameter estimation and model validation," Proc. IAPR Work. on Machine Vision Apps.,Kawasaki, Japan, pp. 552-557, Dec. 1994.
[18] R.A. Wilkinson et al., "The First Census Optical Character Recognition Systems Conference," National Inst. of Standards and Technology, Tech. Rep., NISTIR 1912, Aug. 1992.
[19] H.S. Baird, "Calibration of document image defect models," Proc. Symp. Doc. Anal. and Info. Ret.,Las Vegas, Nev., pp. 1-16, Apr. 1993.
[20] R.A. Wagner and M.J. Fischer, "The String-to-String Correction Problem," J. ACM, vol. 21, no. 1, pp. 168-78, 1974.
[21] J. Esakov, D.P. Lopresti, J.S. Sandberg, and J. Zhou, "Issues in automatic OCR error classification," Proc. Symp. Doc. Anal. and Info. Ret.,Las Vegas, Nev., pp. 401-412, Apr. 1994.
[22] P.A. Devijver and J. Kittler, Pattern Recognition: A Statistical Approach.Englewood Cliffs, N.J.: Prentice Hall, 1982.
[23] D. Harman, “Ranking Algorithms,” Information Retrieval: Data Structures and Algorithms, pp. 363–392, 1992.
[24] J.J. Hull and Y. Li, "Word recognition result interpretation using the vector space model for information retrieval," Proc. Symp. Doc. Anal. and Info. Ret.,Las Vegas, Nev., Apr. 1993.
[25] Y. Li, D. Lopresti, G. Nagy, and A. Tomkins, "Validation of image defect models for optical character recognition," Matsushita Information Technology Laboratory, Tech. Rep. 69-93R, Sept. 1994.
[26] Y. Li, D. Lopresti, and A. Tomkins, "Validation of document image defect models for optical character recognition," Proc. Symp. Doc. Anal. and Info. Ret.,Las Vegas, Nev., pp. 137-150, Apr. 1994.
[27] G. Nagy, "Validation of simulated OCR data sets," Proc. Symp. Doc. Anal. and Info. Ret.,Las Vegas, Nev., pp. 127-135, Apr. 1994.

Index Terms:
Optical character recognition, document image defect models, OCR error classification, defect model validation.
Citation:
Yanhong Li, Daniel Lopresti, George Nagy, Andrew Tomkins, "Validation of Image Defect Models for Optical Character Recognition," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 18, no. 2, pp. 99-108, Feb. 1996, doi:10.1109/34.481536
Usage of this product signifies your acceptance of the Terms of Use.