Fast fingerprint verification using subregions of fingerprint images

We propose a fast fingerprint matching methodology based on localizing the matching regions in captured fingerprint images. The determination of the locations of such regions relies on the accurate detection of reference points in the images together with a priori knowledge of the complete fingerprint obtained during the fingerprint enrollment procedure. The relationship between authentication reliability and region size is studied experimentally. Results show that sufficiently accurate fingerprint matching can be achieved using very small bitmaps, making it possible to implement very fast fingerprint authentication systems using relatively slow embedding processors.