HP iPAQ h5400 Series White Paper page 5

biometric security with the iPAQ Pocket PC h5400 series
. .
fingerprint image processing and feature extraction
. .
. .
Reliable fingerprint feature extraction is the foundation of accurate fingerprint matching.
. .
Algorithms and software are developed to analyze a fingerprint image and to extract the
. .
. .
information that can be used to uniquely identify a person. This is achieved in two steps. First, the
. .
fingerprint image is transformed into a form that facilitates feature detection. Second, the image is
. .
. .
further analyzed to detect local and global features like ridge flow, minutiae points, fingerprint
. .
pattern, ridge counts, etc. The results are passed on to fingerprint image quality control.
. .
. .
. .
Fingerprint image processing and feature extraction is used for both enrollment and verification.
. .
. .
fingerprint image quality control
. .
. .
Image quality control is introduced to maintain the quality of the data collected. This is very
. .
important to a fingerprint security application since data reliability contributes directly to
. .
. .
verification accuracy. Each captured fingerprint image is evaluated and only those that pass the
. .
. . quality criteria are accepted for further use.
. .
. .
Fingerprint image quality control is used for both enrollment and verification.
. .
. .
template generation
. .
. .
In the process of enrollment, fingerprint features extracted, like minutia and ridge counts, are
. .
. .
organized into templates. These templates are called reference templates, or the templates on
. .
record.
. .
. .
During verification, when you scan your fingerprint, a similar template is created. This template is
. .
. .
called a sample template.
. .
. .
template matching
. .
. .
During template matching, the sample template is compared to the set of reference templates
. .
. .
stored on your iPAQ Pocket PC. Because a 100% exact match is unlikely due to various factors, a
. .
probability is established. This is done by using a matching algorithm that tests various
. .
. .
orientations of the extracted image and the degree of correspondence of the minutiae. The
. .
algorithm assigns a numerical score to the match. This score is then compared to a pre-established
. .
. .
setting. If the numerical score is above the threshold, then a match is declared. Otherwise, logon
. .
fails.
. .
. .
Template matching is unique to verification processing.
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
5981-4266EN
5