Restoration of partially damaged fingerprints using a partial differential equation

초록

We present novel sweeping ordering algorithms for the restoration of partially damaged fingerprint images using a partial differential equation (PDE). To efficiently and numerically solve the PDE in the discretized domain, we use the proposed sweeping ordering algorithm in conjunction with a Gauss-Seidel-type update method and an isotropic Laplacian operator. To achieve accurate and stable restoration, we use the information surrounding the damaged region as Dirichlet boundary conditions and propose a method to match the amplitude and wavelength of the damaged fingerprint image with the numerical solution. The proposed sweeping ordering algorithm starts at an interior point adjacent to a boundary point and spirals inward from this point. In this process, it visits all interior points progressively.

키워드