Algorithm UTWE

Organization: University of Twente, Electical Engineering (Netherlands)

UTWE is a Correlation-Based Fingerprint Verification Algorithm.

Unlike the traditional minutiae-based systems, the correlation-based fingerprint verification system directly operates on the gray-scale information of the fingerprints, since a gray-level fingerprint image contains much richer, more discriminatory, information than only the minutiae locations. The latter only characterize a small part of the local ridge-valley structures.

First, templates in the primary fingerprint are selected. Then, template matching is used to find the corresponding template positions in the secondary print, and finally, both sets of template positions are compared.

The templates to be selected in the primary fingerprint should fit as well as possible at the corresponding location in the secondary fingerprint, but as badly as possible at other locations in that print. Possible template selection methods are: minutiae-based, coherence-based and correlation-based. A combination of these methods provides the most reliable template positions.

The selected templates are localized in the secondary fingerprint by the use of standard template matching methods.

The actual matching algorithm is based on the two sets of template positions. It uses two decision stages. First, elementary decisions are made by classifying the individual template position pairs to be matching or not. Then, the information of all matching template pairs is combined in order to make a final decision whether the primary and secondary fingerprint match or not.

Unlike minutiae-based systems, the correlation-based fingerprint verification system is capable of dealing with bad-quality images from which no minutiae can be extracted reliably and with fingerprints that suffer from non-uniform shape distortions. Although the system cannot deal very well with rotations of the prints, experiments have shown that the performance of this system at the moment is comparable to the performance of many other fingerprint verification systems.

For more information please visit UTWE Fingerprint Identification page.

Detailed results on DB1

Detailed results on DB2

Detailed results on DB3

Detailed results on DB4

Comparison chart