Character recognition in degraded document images using morphological and phase-only filtering

Nowadays, new optical character recognition (OCR) algorithms for mobile devices are actively developed. Most of these algorithms work well with documents digitized under perfect conditions. When using mobile devices, document images are subject to various distortions, such as nonuniform and poor illumination, geometric distortions, low resolution, and sensor noise. In this paper, we propose a reliable method for recognition of contextually unrelated characters in degraded document images with the use of a bank of adaptive morphological and phase-only filters. With the help of computer simulation, the results of operation of traditional and proposed approaches to detection and classification of characters of the Latin alphabet in distorted document images are compared.

[1]  L. P. Yaroslavsky,et al.  III The Theory of Optimal Methods for Localization of Objects in Pictures , 1993 .

[2]  Chirag I. Patel,et al.  Optical Character Recognition by Open source OCR Tool Tesseract: A Case Study , 2012 .

[3]  Vitaly Kober,et al.  Adaptive composite filters for pattern recognition in nonoverlapping scenes using noisy training images , 2014, Pattern Recognit. Lett..

[4]  Vitaly Kober.  Robust and efficient algorithm of image enhancement , 2006, IEEE Transactions on Consumer Electronics.

[5]  Vitaly Kober,et al.  Adaptive synthetic discriminant function filters for pattern recognition , 2006 .

[6]  Petros Maragos,et al.  Optimal Morphological Approaches To Image Matching And Object Detection , 1988, [1988 Proceedings] Second International Conference on Computer Vision.

[7]  Yoshua Bengio,et al.  Gradient-based learning applied to document recognition , 1998, Proc. IEEE.

[8]  I. A. Ovseyevich,et al.  Unsharp Masking by the Rank-Order Filters with Spatially Adaptive Neighborhoods , 2002 .

[9]  P. Erhan Eren,et al.  Text recognition and correction for automated data collection by mobile devices , 2014, Electronic Imaging.

[10]  Vitaly Kober,et al.  Nonlinear synthetic discriminant function filters for illumination-invariant pattern recognition , 2008 .

[11]  Vitaly Kober,et al.  Nonlinear filters with spatially-connected neighborhoods , 2001 .

[12]  Petrica C. Pop,et al.  Optical character recognition in real environments using neural networks and k-nearest neighbor , 2013, Applied Intelligence.

[13]  Jun Guo,et al.  Text extraction from natural scene image: A survey , 2013, Neurocomputing.

[14]  Alireza Behrad,et al.  Cluster based weighted SVM for the recognition of Farsi handwritten digits , 2010, 10th Symposium on Neural Network Applications in Electrical Engineering.

[15]  Cemil Oz,et al.  The Application of optical character recognition for mobile device via artificial neural networks with negative correlation learning algorithm , 2013, 2013 International Conference on Electronics, Computer and Computation (ICECCO).

[16]  J. Fitch,et al.  Median filtering by threshold decomposition , 1984 .

[17]  Christoph H. Lampert,et al.  Document image dewarping using robust estimation of curled text lines , 2005, Eighth International Conference on Document Analysis and Recognition (ICDAR'05).

[18]  A. Lohmann,et al.  The wigner distribution function and its optical production , 1980 .

[19]  Daniel Schuster,et al.  Information extraction efficiency of business documents captured with smartphones and tablets , 2013, ACM Symposium on Document Engineering.

[20]  Petros Maragos.  Morphological correlation and mean absolute error criteria , 1989, International Conference on Acoustics, Speech, and Signal Processing,.

[21]  Di Ma,et al.  A super resolution framework for low resolution document image OCR , 2013, Electronic Imaging.

[22]  Saúl Martínez-Díaz,et al.  Distortion-invariant pattern recognition with local correlations , 2011, Pattern Recognition and Image Analysis.

[23]  I. A. Ovseyevich,et al.  Rank Image Processing Using Spatially Adaptive Neighborhoods 1 , 2001 .

[24]  Eric K. Ringger,et al.  Combining multiple thresholding binarization values to improve OCR output , 2013, Electronic Imaging.

[25]  Vitaly Kober,et al.  Target tracking in nonuniform illumination conditions using locally adaptive correlation filters , 2014 .

[26]  Ioannis Pratikakis,et al.  Adaptive degraded document image binarization , 2006, Pattern Recognit..

[27]  Tae-Sun Choi,et al.  Improved motion stereo matching based on a modified dynamic programming , 2001 .