Fingerprint Enhancement Using Oriented Diffusion Filter

Fingerprint enhancement is a critical step in a fingerprint identification system. Recently, some anisotropic nonlinear diffusion filter is applied to the fingerprint preprocessed. Impressive results are main reason for using nonlinear diffusion filtering in image processing. Poor efficiency, especially the computational load, is the main reason for not using nonlinear diffusion filtering. In order to improve the efficiency, a novel piecewise nonlinear diffusion for fingerprint enhancement is presented. It shows anisotropic diffusion equation using diffusion tensor which continuously depends on the gradient is not necessary to smooth the fingerprint. We simplify the anisotropic nonlinear-diffusion in order to satisfy a real-time fingerprint recognition system. According to the local character of the fingerprint, the diffusion filter is steered by the orientation of ridge. Experimental results illustrate that our enhancement algorithm can satisfy the requirement of an AFIS.