Pattern recognition has been an important area in
computer vision applications. In the case of a planar image,
there are four basic forms of geometric distortion caused by the
change in camera location: translation, rotation, scaling and
skew. So far, a number of methods have been developed to solve
these distortions, such as moment invariants’, Fourier descriptor,
Hough transformation, shape matrix’ and the principle axis
method. All of the above methods can be made invariant to
translation, rotation and scaling. However, they become useless
when pattern is skewed: when the direction of the camera or
scanner is not vertical to the planar image or the sampling
intervals in the x and y directions are not equal, the image is
skewed. Authors present a moment based normalization process
for character images for the purpose of enhancing the
performance of character recognize for isolated characters of
Gujarati. This paper focuses on character pre-processing and
normalization stage of handwritten character recognition for
Gujarati.
Jayashree Rajesh Prasad : graduated in Computer
Science and Engineering from North Maharashtra University in 1996
and completed M.E. in Computer Engineering from Pune University
in 2004. She pursued Ph.D. in Computer Science and engineering
from Swami Ramananda Tirtha University, Nanded in 2014. She has
a research project “Conversion of Gujarati Script to Speech”, funded
by BCUD (University of Pune) to her credit. She works with Sinhgad
College of Engineering, Pune. Her research interests are in the field
of Soft Computing, pattern recognition and image processing. She is
Life member of Computer Society of India, Life Member of Indian
Society for Technical Education, Member of IAENG (International
Association of Engineers) and Member of IACSIT (International
Association of Computer Science and Information Technology).
Affine transformation
distortion
moments
rotation
scaling
translation
Author describes a normalization process that achieves
invariance under affine geometric distortions. In this
application authors apply a normalization procedure to
the image so that it meets a set of predefined moment
criteria. A recognition rate of 86.6 % [1] for the isolated
characters of Gujarati is achieved. This paper does not
focus on the classification details of this application.
Author has focused on the moment based image
normalization applied in the pre-prepossessing stage of a
character recognition task.
[1] Jayashree Prasad, Uday Kulkarni, ‘Gujarati Character
Recognition using weighted k-NN with mean chi
square Distance Measure’, DOI 10.1007/s13042-013-
0187-z International Journal of Machine Learning and
Cybernetics, Springer, ISSN 1868-8071.
[2] M. Maloo and Kale, “Gujarati Script Recognition: A
Review,” In: International Journal on Computer
Science and Engineering (IJCSE), 2011. [3] A. Name,
"Dissertation Title", M.S.(or Ph.D.) thesis,
Department, University, City, Country, Year.
[3] S. Antani and Lalitha Agnihotri, “Gujarati Character
Recognition,” In: ICDAR, pp. 418-421, 1999.
[4] Mohamed Cheriet, Nawwaf Harma, Cheng-Lin Liu,
Ching Y. Sen, “Character Recognition Systems A
Guide for Students and Practitioners,” A John Wiley
& Sons, Inc., Publication, 2007.
[5] Marius Bulacu, Lambert Schomaker and Axel Brink,
“Text-Independent Writer Identification and
Verification on Offline Arabic Handwriting,” In:
ICDAR, IEEE Computer Society, Volume. II, pp. 769-
773, 2007.
[6] Lam, L., Seong-Whan Lee, and Ching Y. Suen,
Thinning Methodologies- A Comprehensive Survey,
In: IEEE Transactions on Pattern Analysis and
Machine Intelligence, Volume 14, No. 9, page 879,
September 1992.