A Neural Approach to Cursive Handwritten Character Recognition Using Features Extracted from Binarization Technique - Complex System Modelling and Control Through Intelligent Soft Computations - page 738

with the image in an ef cient way. A grayscale image after binarization can be

classi ed into two categories: Globally Thresholded and Adaptive (Locally)

Thresholded images.

5.2.1 Global Thresholding

Global thresholding methods (Otsu 1979 ; Kapur et al. 1985 ; Cheng et al. 1998 ;Li
and Lee 1993 ) are most suitable in those documents in which there is uniform
contrast distribution between the foreground text and the background. In other
words, global Thresholding is used for binarization of those grayscale images in
which the written text is of almost same intensity value and is written by a single
pen on a background having almost uniform intensity but with quiet different
intensity value as compared to the intensity value of the text. In such type of
grayscale images, only one threshold value is used for the whole image to classify it
into two categories; text and background, and the thresholding is called Fixed
Global Threshold. It is used to compare the grayscale intensity of each pixel (I x )of
the image with a Global Threshold (say T = 0.5). The new intensity value of the
pixel (I y ) can be calculated by the following expression as:

I y ¼ 1f x T and I y ¼ if x \ T where 0 T 1

\

repre-
sents white pixels. Hence, the pixels having intensity value greater than the global
threshold will be white and the pixel having intensity value less than the global
threshold will appear as black. The grayscale intensity threshold value can be
assigned any value that lies between the intensity value of the foreground text and
the background intensity. Generally, its value should be closer to the intensity of
the foreground text as compared to the background intensity, so as to remove the
maximum undesirable background noise while preserving all the important details
of the foreground text.
The selection of the threshold parameter is based on the gray-scale intensity of
the text in the document. More intensity leads to the more threshold value. This
parameter is decided as shown in Table 1 .
The

Here, intensity value ' 0 ' represents black pixels and intensity value ' 1 '

first column of Table 1 represents the intensity of handwritten text present in
the document. This intensity is a gray-scale value when the document is converted
into grayscale format by using

' img2gray ' method of MATLAB. The second

Table 1 Intensity/threshold
comparison table

Gray-scale intensity of the text (I) Threshold value

0.00 - 0.20 0.19

0.21 0.40 0.39

-

0.41 0.60 0.59

-

0.61 - 0.80 0.79

0.81 1.00 0.99

-