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.

(a) (b)<br />

Figure A.14: Thinned images <strong>of</strong> two input fingerprints shown in Fig. A.2.<br />

ditions (c’) and (d’) are used to remove north-west boundary points and south-east<br />

corner points. The iteration process is repeated until no more pixels can be removed.<br />

The skeleton images <strong>of</strong> the input fingerprint images are shown in Fig. A.14.<br />

A.5 Minutiae Extraction<br />

Once a binary skeleton has been obtained, a simple image scan allows the pixel<br />

corresponding to minutiae to be detected. This stage consists <strong>of</strong> two stages [100],<br />

• Detection <strong>of</strong> ridge endings and bifurcations from thinned images, by using<br />

neighborhood constraints and<br />

• Post-processing to remove undesired features.<br />

For minutiae detection the concept <strong>of</strong> Crossing number is used. Pixels corre-<br />

sponding to minutiae are characterized by a crossing number different from 2. The<br />

crossing number [6] cn(p) <strong>of</strong> a pixel p in a binary image is defined as half the sum <strong>of</strong><br />

the differences between pairs <strong>of</strong> adjacent pixels in the 8-neighborhood <strong>of</strong> p:<br />

cn(p) = 1<br />

2<br />

�<br />

i=1,..,8<br />

|val(pi mod 8) − val(pi−1)|, (A.34)<br />

where p0, p1, ..., p7 are the pixels belonging to an ordered sequence <strong>of</strong> pixels defin-<br />

ing the 8-neighborhood <strong>of</strong> p and val(p) ∈ {0, 1} is the pixel value. It is simple to note<br />

(Fig. A.15) that a pixel p with val(p) = 1:<br />

137

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

Saved successfully!

Ooh no, something went wrong!