This is the most popular and widely used in commercial applications, due to its good performance and low computation time, especially for good quality images. This method attempts to align the minutiae of the input image (query template) and the stored templates (reference template) and find the number of matching minutiae. After alignment, two minutiae are considered in coincidence if the spatial distance and the steering difference between them are less than a given tolerance. Correct alignment of the fingerprint is very important to maximise the number of coupled minutiae, this requires calculation of translation and rotation information, as well as other geometric transformations such as scale and distortion. Many approaches have been proposed in order to efficiently calculate the alignment of information. In this section we present a method that uses segments (made up of minutiae) instead of isolated minutiae. A segment consists of two pairs of minutiae of the same fingerprint, the way in which the set of segments (eg nearest neighbour, delaunay, etc.) can be constructed. The following figure shows the segments constructed from the set of minutiae.
With identity fraud in our society reaching unprecedented proportions and with a growing emphasis on emerging automatic personal identification applications, bio-metric-based verification, especially fingerprint-based identification, is receiving a lot of attention. There are two major shortcomings of traditional approaches to fingerprinting. For a considerable fraction of the population, representations based on the explicit detection of complete crest structures in the fingerprint are difficult to extract automatically. The heavily used minutiae-based representation does not use a significant component of the rich discriminatory information available on fingerprints. Local crest structures can not be completely characterised by minutiae. In addition, the minutiae-based comparison has difficulty rapidly comparing two fingerprint images containing different numbers of unregistered minutiae points. The proposed filter-based algorithm uses a Gabor filter bank to capture local and global details on a fingerprint such as a fixed fixed length Finger Code. The matching of fingerprints is based on the Euclidean distance between the two corresponding finger codes and, therefore, is extremely fast. We can achieve a verification accuracy that is only marginally lower than the best results of algorithms based on minutiae published in the open literature. Our system works better than a state-of-the-art based minutiae system when the performance requirement of the application system does not require a very low false acceptance rate. Finally, it is demonstrated that the matching performance can be improved by combining the decisions of the matchers based on complementary fingerprint information (based on minutiae and filter-based).