Digital Enhancement of Degraded Fingerprints

Abstract

Once latent fingerprints have been obtained, there are two major operations to be performed; they are enhancement and classification. In this thesis we strictly focus on digital image enhancement, and for this purpose, several techniques for fingerprint enhancement were studied and developed. These techniques may be implemented in either the spatial or spatial frequency domains. Processing techniques in the spatial frequency domain are based on modifying the two-dimensional Fourier transform of the image. The approaches in the spatial domain are based on direct manipulation of the pixels. Since the contrast of latent fingerprints is space-variant, a spatially adaptive technique, in the spatial domain, was studied and developed further. Such a technique was adaptive binarization, which makes use of moving windows with spatially-varying parameters. A double pass using this algorithm improved the fingerprint appearance even further. Other spatial main methods studied were contrast stretching/sliding and the image complement, which provided a better quality of print for the purpose of ridge/valley discrimination. Spatial-frequency domain methods developed included, linear ideal/Butterworth filters, which were individually tested, and a homomorphic filter process, which made use of generalized linear filters. This later method proved effective in removing multiplicative degradations that had been introduced into the latent fingerprint. A one-dimensional fingerprint diffusion model approach led to the development and application of a Laplacian operator, which more accurately describes such a diffusion process, resulting in a fingerprint image where the edges were sharpened.