Fingerprint Image-Quality Estimation to Improve the Identification Performance using DWT

Computer Engineering

Elastic distortion of fingerprints is the chief causes for fake mismatch. As this reason disturbs all fingerprint detection applications, it is particularly risk in negative identification applications, such as watch list and deduplication applications. In such applications, cruel persons may deliberately distort their fingerprints to hide recognition. In this paper, we recommended novel algorithms to identify and correct skin distortion based on a single fingerprint image. Distortion detection is demonstrated as a two-class classification problem, for which the registered ridge orientation map and period map of a fingerprint are helpful as the feature vector and a SVM classifier is educated to act the classification task. Distortion rectification (or equivalently distortion field estimation) is analyzed as a regression difficulty, where the input is a distorted fingerprint and the output is the distortion field. To simplify this problem, a database (called reference database) of various distorted reference fingerprints and matching distortion fields is built in the offline stage, and then in the online stage, the closest neighbor of the input fingerprint is planned in the reference database and the equivalent distortion field is used to change (Convert) the input fingerprint into a normal fingerprints. Capable results have been achieved on three databases having many distorted fingerprints, namely NIST SD27 latent fingerprint database, Tsinghua Distorted Fingerprint database, FVC2004 DB1.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.