Gorka Treceño Fontecha; Beatriz Mora Barbero. BEE07:02, pp. 78. TEK/avd. för signalbehandling, 2007.
In human fingerprints there are plenty of small details which are the discontinuities
in the ridges, denoted minutiae. Minutiae-based matching is a wellknown
method applied in fingerprint recognition. Minutiae defines a representative
feature vector of a fingerprint. Due to quality variations in fingerprint
images a preprocessing in form of binarization and skeltonization is applied,
before obtaining the feature vector. Once the features of each fingerprint are
obtained, a matching algorithm carries out the comparison task between those
features vectors to determine if they match.
In the new binarization function, quality at the same time that speed are
essential requirements. In order to sort out this subject, areas of an adaptive
size are analyzed instead of a pixel by pixel examination. The binarization
algorithm is implemented in the frequency domain with the concept of ridges
and their orientations.
Skeletonization is needed as a preprocessing step with the aim of obtaining
the minutiae from the fingerprint. It is a matter of fact that the high quality
skeleton is an emphatic factor in the fingerprint recognition. A robust skeletoned
image will ensure a reliable extraction of features. Several strategies
relating to the application of filters will be discussing in order to speed-up the
iteractive thinning algorithms.
The proposed minutiae matching algorithms explore though several variations
of the geometric hashing method. Based on minutiae features, new approaches
are presented to obtain an effectiveness function.