Fingerprints are the biometric features most widely used for personal identification.

 Abstract

Most of the methods on the market are minutiae-based (Fig. 2), i.e. they extract from the image unique particulars (ridge endings and bifurcations) and use them as fingerprint features to distinguish two users. 

JOANNEUM RESEARCH is developing iFingerSys, a new technology for fingerprint verification and identification. iFingerSys is a completely new approach to the fingerprint recognition problem. It is the result of years of research in the fingerprint field.
iFingerSys transforms the fingerprint image domain to mathematical domains (iFingerSpaces), in which it looks for singular features (iFingerFeatures) used to discriminate between two different fingers. The iFingerSpaces (Fig. 3) are rotation, translation, and scale-invariant, which have a low sensitivity to noise that is invariably present on images generated by commercial sensors.

The research activities of JOANNEUM RESEARCH include both the standard methods and the innovative iFingerSys approach. We can summarize our work in:

• Compression of minutiae neighborhood: Principal Component Analysis (PCA) and wavelet methods for efficient compression and increase in matching accuracy.
• Fusion of several pictures of the same finger: only the minutiae position is needed for the fusion, while the complete image is used for accuracy.
• Fast identification: the use of global features for fast identification of matching fingerprints allows a considerable increase in the number of matches.
• Reconstruction algorithms for sweeping sensors: very complex algorithms for object tracking and optical flow are used to compose the image.
• iFingerSys methods.

We are Members of: