Fingerprint Orientation Field Modeling
and Its Applications

Jinwei Gu, Jie Zhou and Chunyu Yang

Abstract

As a global feature of fingerprints, the orientation field is important for automatic fingerprint recognition. Many algorithms have been proposed for orientation field estimation, but their results are unsatisfactory for poor quality fingerprint images. In this project, we proposed a analytic model for the 2D orientation field of fingerprint texture patterns. The proposed model includes both a bivariate polynomial component for fitting the global orientation field, and "point-charge" models for each of the singular points to further improve the accuracy of the modeling.

This analytic model then can be used to: compute high-quality orientation field for poor quality images, store the orientation field with a few parameters, and together with the minutiae points form a complete representation of fingerprints which can then be used for reconstruction and synthesis of fingerprint images.

Based on this analytic model, we can also use the orientation field as as a discriminant global feature for recognition. We combined it with the widely-used local feature---minutiae points in recognition and showed it can largely improve the recognition performance.

Finally, we investigated the orientation field patterns around the singular points (i.e., core and delta points), derived topological principles for the singular points, and applied these principles for detecting singular points, especially for poor-quality images.

Publication


1. Modeling of Fingerprint Orientation Field

orientation field modeling

2. Orientation Field As a Global Feature for Fingerprint Recognition

recognition

3. Topological Characterisitc of Fingerprint Orientation Field and Its Application for Singular Point Detection

topology