Information Technology Reference
In-Depth Information
Fig. 6 Noisy handwritten
character image
Fig. 7 Noise detection in
handwritten character image
5.3 Noise Removal
Noise (small dots or foreground components) may be introduced easily into an
image while scanning the handwritten character image during image acquisition. It
is very necessary to eliminate the noise from the handwritten character images as
shown in Fig.
6
, so as to make this image
'
fit for further processing. MATLAB
s
'
is used to morphologically open the binary image by removing small
objects that have less than a particular number (user speci
bwareaopen
'
ed) of pixels and pro-
ducing another binary image. The small noise dots were removed by using
'
bwareaopen
'
but some small portions of the characters e.g. dots
'
.
'
of characters
'
of MATLAB
were used to highlight the pixels that were removed as shown in Fig.
7
. A logical
i, j
'
etc. were also lost. The methods
'
bwlabel
'
and
'
regionprops
'
Search WWH ::
Custom Search