Fingerprint Classification Based on Sparse Representation Using Rotation-Invariant Features

A new fingerprint classification method is proposed under Galton-Henry classification scheme. We first modified FingerCode to generate rotation-invariant distance. Then, the distances between a fingerprint’s FingerCode and templates’ are used to represent the fingerprint. On classification step, we put the rotation-invariant features of the training sets together, and solve a sparse representation problem for a query fingerprint. The experiment results show that the proposed feature is robust and the classification method gives an accurate result.