fingerprint image may thwart the feature extraction process; hence, it needs to be
removed. However, preprocessing methods should be selected carefully as they remove
fingerprint properties [4].
Image size is one of the most important characteristics. If the image size is large,
then the algorithm takes a bit more time to throw the outputs and more time is consumed
during the preprocessing stage too. Therefore, a standard size of 2 inch by 2 inch at 300
dpi scanned as a grayscale image serves as input. The preprocessing stage has two levels.
One is filtration and the other is normalization.
The filtration process is done to distinguish the fingerprint from the background.
This is done by analyzing the image. The image is read for the intensity value of each
pixel and later, the intensity value is change to either 1 (black – fingerprint) or 255 (white
– background). In this way, the fingerprint can be easily distinguished from the
background.
The next step after filtration is normalization. One of the images should be
normalized with respect to the other image in comparison so that the fingerprints can be
aligned with each other exactly. Normalization plays a very important role because this
phase helps in maximizing the mutual information value. If an image is normalized
perfectly and both the fingerprints are genuine, then the mutual information value is
maximized.
Normalization is done as follows: