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

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

3.7 Area difference between genuine and impostor distribution with their<br />

ideal counterparts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55<br />

3.8 ZeroFRR and TZeroF RR are illustrated using hypothetical curves <strong>of</strong><br />

FAR(t) and FRR(t). . . . . . . . . . . . . . . . . . . . . . . . . . . . 57<br />

3.9 Training samples for three databases: first, second and third row shows<br />

the training set for Yale, PIE, and ORL databases, respectively. . . . 64<br />

4.1 A typical eigenvalue spectrum with its division into null space and<br />

range space. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75<br />

4.2 Stepwise feature subset selection . . . . . . . . . . . . . . . . . . . . . 80<br />

4.3 Architecture <strong>of</strong> our proposed method <strong>of</strong> feature fusion. . . . . . . . . 82<br />

4.4 The fusion architecture <strong>of</strong> the proposed method <strong>of</strong> decision fusion. . . 87<br />

4.5 Examples where LDA based on the criteria tr(S −1<br />

w Sb) fails to discrim-<br />

inate between classes. . . . . . . . . . . . . . . . . . . . . . . . . . . . 89<br />

4.6 Unique decomposition <strong>of</strong> faces into null space and range space <strong>of</strong> within-<br />

class scatter drawn from Yale (first row), ORL (second row) and PIE<br />

(third row) databases. . . . . . . . . . . . . . . . . . . . . . . . . . . 97<br />

4.7 Training samples for three databases: First, second and third row<br />

shows the training set for Yale, ORL, PIE databases, respectively. . 98<br />

5.1 Operation <strong>of</strong> proposed method for classifier fusion. . . . . . . . . . . . 108<br />

A.1 (a) and (b) A local window containing ridge valley structures; (c) and<br />

(d) Curve (i.e. x-signature) obtained along the direction normal to the<br />

ridge orientation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124<br />

A.2 Input fingerprint images. . . . . . . . . . . . . . . . . . . . . . . . . . 125<br />

A.3 Normalized images <strong>of</strong> two input fingerprints shown in Fig. A.2. . . . 125<br />

A.4 Orientation images <strong>of</strong> two input fingerprints shown in Fig. A.2. . . . . 127<br />

A.5 A X-signature with w = W and l = H in our case. . . . . . . . . . . . 127<br />

A.6 Frequency images (after filtering) <strong>of</strong> two input fingerprints shown in<br />

Fig. A.2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130<br />

A.7 Enhanced images <strong>of</strong> two input fingerprints shown in Fig. A.2. . . . . 131<br />

xvi

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

Saved successfully!

Ooh no, something went wrong!