Fingerprint image analysis for automatic identification |
| |
Authors: | B M Mehtre |
| |
Affiliation: | (1) R and D Division, CMC Limited, 115 Sarojini Devi Road, 500 003 Secunderabad, India |
| |
Abstract: | Most of the papers on fingerprints deal with classification of fingerprint images. Fingerprint databases being large (in the
range of millions), the effort in matching of fingerprints within a class or when the class is unknown, is very significant.
This requires fingerprint image analysis and extraction of the “minutiae” features, which are used for matching FPs. In this
paper a scheme of preprocessing and feature extraction of fingerprint images for automatic identification is presented, which
works even if the pattern class is unknown. The identification of fingerprints is based on matching the minutiae features
of a given finger-print against those stored in the database. The core and delta information is used for classification and
for registration while matching. These algorithms have been tested for more than 10,000 fingerprint images of different qualities.
The results are manually verified and found to be very good for practical application. A few sample results are presented. |
| |
Keywords: | fingerprint preprocessing restoration contextual filters segmentation foreground background directional image thinning minutiae extraction postprocessing |
本文献已被 SpringerLink 等数据库收录! |
|