Information Technology Reference
In-Depth Information
3 Motivation and Various Challenges During Recognition
Now a days, English is the most commonly used language over the internet, in the
entire banking system in the world, postal departments across the countries,
insurance companies, space research projects, software multinational companies,
business houses and research organizations. As English Language is used by a
much higher percentage of the world
ted all over
the globe if an automation system is designed for off-line handwriting recognition.
The off-line handwriting recognition system can enable the automatic reading and
processing of a large amount of data printed or handwritten in English script.
Although such automated systems for recognizing off-line handwriting already
exist, the scope of further improvement is always there. As a result, in spite of a
dramatic boost in this
'
s population, people will be bene
field of research, the development of an intelligent and robust
off-line handwritten words segmentation and recognition remains an open problem
and continues to be an active area for research towards building an fully automated
system by exploring new techniques and methodologies that would improve seg-
mentation and recognition performance in terms of accuracy and speed.
In the process of handwritten character
recognition, various challenges
encountered can be described as follows:
There can be variation in shapes and writing styles of different writers.
Different sizes of character images written by different writers.
Since handwriting depends on the writer and even a single writer cannot always
write the same character in exactly the same way under different conditions.
Characters may be written on a paper with colored or noisy background.
Characters may be written by a pen having ink of different colors.
4 Overall OCR System Design
Various steps involved in the proposed handwritten character recognition system
are illustrated in Fig.
1
.
4.1 Character Image Acquisition
During image acquisition, the handwritten character images were acquired through
a scanner or a digital camera. The input character images were saved in JPEG or
BMP formats for further processing. Some of those character image samples were
written on white paper with colored ink and others on a colored or a noisy
background.
Search WWH ::
Custom Search