Improved Hessian Curvature for Fingerprint Analysis

Abstract

Curvature analysis is important to the fingerprint image which is constructed by a group of regular ridgelines. Five kinds of curvatures constructed by the eigenvalues are tested, including the Gaussian curvature, Mean curvature, Laplacian curvature, Hessian curvature and our improved Hessian curvature. Researching on these five kinds of curvature, we find the relative advantages of our improved Hessian curvature for fingerprint curves. A measure for fingerprint reference point determination based on the eigenvalues of the covariance matrix of the ridges. Fingerprint reference points are determined by combining the global and local proposed curvature properties of ridges by using multi-resolution levels. The performance of the proposed one is verified through simulations and its analysis.