Image Processing Reference
In-Depth Information
O
Fig. 11.8.
” represent the delta and core (parabolalike) points that have
been automatically extracted in several fingerprint images.
Top, left
and
bottom, left
are two
fingerprints of the same finger that differ significantly in quality. The certainties are 0.84 (+)
and 0.64 ()in
top, left
. The certainties are 0.73 (), 0.22 (“+”) and 0.10 (O) in
bottom, left
The “+” and the “
g
(
z
)=
z
2
n
=
−
1
⇒
(11.108)
were used to model core, whereas
g
(
z
)=
z
2
n
=1
⇒
(11.109)
were used for delta, see also Figs. 11.6 and 11.8.
Example 11.9. Figure 11.9 shows the result of GST to locate letters for an optical
character recognition system [186], in Sinhala script. Letters of a script encoding
a natural language can be more effectively modeled by GST than direct gray values
becauseocclusionandnoiseaffectsthedirectionalinformationlessthangrayvalues.
In this use of GST the analytic expressions of the pattern isocurves are not known,
buttheyaredeductedfromthestructuretensorvariationsofthereferencecharacters,
