A Neural Approach to Cursive Handwritten Character Recognition Using Features Extracted from Binarization Technique - Complex System Modelling and Control Through Intelligent Soft Computations - page 733

Noise (small dots or blobs) may easily be introduced into an image during image
acquisition (Verma and Blumenstein 2008 ). A common appearance of noise in
binary images takes the form of isolated pixels, salt-and-pepper noise or speckle
noise, thus; the processing of removing this type of noise is called

filling, where

each isolated pixel salt-and-pepper “ island ” is filled in by the surrounding “ sea ”

(O Gorman et al. 2008 ; Alginahi 2010 ).

The neural network accepted areas between the upper and lower baselines of
each word as input. This area, called the core, must be of

'

fixed height to be used in
conjunction with the neural net. Therefore it was necessary to scale the words so
that all cores are of an identical height (Verma and Blumenstein 2008 ).
De-skewing is the process of first detecting whether the handwritten word has

been written on a slope, and then rotating the word if the slope is angle is too high so

that the baseline of the word is horizontal (Verma and Blumenstein 2008 ). Some
degree of skew is unavoidable either a paper is scanned manually or mechanically
(Sarfraz and Rasheed 2008 ; Sadri and Cheriet 2009 ; Saba et al. 2011 ).
Thinning is a data reduction process that erodes an object until it is one-pixel
wide, producing a skeleton of the object making it easier to recognize objects such
as characters. Thinning erodes an object over and over again (without breaking it)
until it is one-pixel wide. On the other hand, the medial axis transform

'

nds the
points in an object that form lines down its center (Davies 2005 ). The medial axis
transform is similar to measuring the Euclidean distance of any pixel in an object to
the edge of the object, hence, it consists of all points in an object that are minimally
distant to more than one edge of the object (Russ 2007 ; Alginahi 2010 ).
The purpose of feature extraction is to achieve most relevant and discriminative
features to identify a symbol uniquely (Blumenstein et al. 2007 ). Many feature
extraction technique are proposed and investigated in the literature that may be used
for numeral and character recognition. Consequently, recent techniques show very
promising results for separated handwritten numerals recognition (Wang et al.
2005 ), however the same accuracy has not been attained for cursive character
classi cation (Blumenstein et al. 2007 ). It is mainly due to ambiguity of the

character without context of the entire word (Cavalin et al. 2006 ). Second problem
is the illegibility of some characters due to nature of cursive handwriting, distorted
and broken characters (Blumenstein et al. 2003 ).
Recently, neural network classi

ers are proved to be powerful and successful for
character/word recognition (Verma et al. 2004 ; Blumenstein et al. 2007 ). However,
to improve the intelligence of these ANNs, huge iterations, complex computations,
and learning algorithms are needed, which also lead to consume the processor time.
Therefore, if the recognition accuracy is improved, the consumed learning time will
increase and vice versa. Which is the main drawback of ANN based approaches
(Aburas and Rehiel 2008 ).