Enhancement of low-quality fingerprint images by a three-stage filtering scheme

Fingerprint image quality heavily influences the recognition rate for fingerprint identification/verification systems. A low-quality fingerprint image may consists of broken ridges, scars, smears, falsely conglutinated ridges, poor ridge and valley contrast, etc. In this paper, we propose a novel and effective three-stage scheme to enhance low-quality fingerprint images. The first-stage consists of an oriented linear anisotropic diffusion filter with a local ridge orientation estimation that differs from traditional estimation method. The second and the third stages consist of an oriented local ridge compensation filter and a novel angular filter, respectively. Simulation results show that the proposed method can improve the quality of the fingerprint images with a lower computational cost.