20.01.2013 Views

Master Thesis - Department of Computer Science

Master Thesis - Department of Computer Science

Master Thesis - Department of Computer Science

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

These stages are discussed in the following sections.<br />

A.1 Image Enhancement<br />

The quality <strong>of</strong> the ridge structures in a fingerprint image is an important characteris-<br />

tic, as the ridges carry the information <strong>of</strong> characteristic features required for minutiae<br />

extraction. Ideally, in a well-defined fingerprint image, the ridges and valleys should<br />

alternate and flow in a locally constant direction. This regularity facilitates the de-<br />

tection <strong>of</strong> ridges and consequently, allows minutiae to be precisely extracted from<br />

thinned ridges. However, in practice, a fingerprint image may not be well defined due<br />

to the noise that corrupt the clarity <strong>of</strong> the ridge structures. Thus image enhancement<br />

are <strong>of</strong>ten employed to reduce the noise and enhance the definition <strong>of</strong> ridges against<br />

valleys.<br />

The fingerprint image enhancement algorithm receives an input gray-level finger-<br />

print image and outputs the enhanced image after applying a set <strong>of</strong> intermediate steps<br />

which are: (a) Normalization, (b) Orientation Image Estimation, (c) Frequency Im-<br />

age Estimation and (d) Filtering using Gabor filter. This algorithm is mostly adopted<br />

from the technique proposed by Hong, Wan and Jain [49].<br />

A.1.1 Notation<br />

• A gray-level Fingerprint image (I): A gray-level fingerprint image, I, is<br />

defined as an N ×N matrix, where I(i, j) represents the intensity <strong>of</strong> the pixel at<br />

the i th row and j th column. The mean and variance <strong>of</strong> a gray-level fingerprint<br />

image, I, are defined as,<br />

M(I) = 1<br />

N 2<br />

V AR(I) = 1<br />

N 2<br />

N−1<br />

�<br />

i=0<br />

N−1 �<br />

i=0<br />

N−1<br />

�<br />

j=0<br />

N−1 �<br />

j=0<br />

I(i, j) (A.1)<br />

(I(i, j) − M(I)) 2<br />

(A.2)<br />

• An orientation image (O): An orientation image, O, is defined as an N ×N<br />

image, where O(i, j) represents the local ridge orientation at pixel (i, j). Local<br />

ridge orientation is usually specified for a block rather that at every pixel. An<br />

123

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!