A Neural Approach to Cursive Handwritten Character Recognition Using Features Extracted from Binarization Technique - Complex System Modelling and Control Through Intelligent Soft Computations - page 739

Fig. 5 Handwritten character
image in binary format

column of this table represents the corresponding value of threshold intensity level
for binarization process. The threshold parameter along with the grayscale image is
made an input to the binarization method

' im2bw ' designed in MATLAB. The

output is a binary image as shown in Fig. 5 .

5.2.2 Adaptive Thresholding

It has been found that the global thresholding approach gives excellent result when
the text is written by a single pen with same intensity on a background of
throughout uniform intensity but considerably different from the text intensity. The
results start degrading when the pens of different intensities are used on various
noisy backgrounds. The global thresholding method fails because of drastic vari-
ation in the text intensity and variation in contrast illumination between text and
background. In such cases, the adaptive thresholding methods by Bernsen ( 1986 ),
Niblack ( 1986 ), Sauvola et al. ( 1997 ) are employed and the document image is
divided into small blocks and the threshold values are computed pixel by pixel or
region by region. For each region or block, a local threshold value is determined for
the binarization of that particular block and the binarized images of each block are
combined to get the

final binarized document image.
In the proposed binarization technique, the threshold value is decided based on
the intensity of the text as explained in Table 1 . It is assumed that the intensity of
the text is less than that of background i.e. the input image has black foreground
pixels and white background pixels. The colors can be inverted if the input image
has text intensity more than that of background. Also, the background intensity
remains almost uniform throughout the image and does not change drastically
anywhere in the input image. Hence, in the proposed binarization technique, global
intensity thresholding is employed and the resulting handwritten character image
after background noise removal and binarization using global gray scale intensity
threshold is shown in Fig. 5 .