Feature Extraction in Fingerprint Images

Abstract

A fingerprint image is comprised of a spatial map of the friction ridges of the skin and the valleys between them. An automated fingerprint indentification system (AFIS) compares two fingerprints by examining the “landmarks” (or features) of the ridges and valleys in order to decide whether they are a matching pair. Two fingerprint images are called a “matching pair” if it can be determined that they both are produced by the same finger of the same individual regardless of the time and method by which each image is collected. In most of the current fingerprint matching systems, the features used in the matching process are the fingerprint minutiae, mainly ridge bifurcation and ridge ending. However, the features do not necessarily have to be minutiae, as there are several characteristics that can also be utilized in the matching process. Minutiae-based fingerprint matching requires each fingerprint image be transformed into a minutiae map. This transformation generally includes the following steps: preprocessing, ridge direction and ridge width, enhancement, and minutiae detection. This chapter discusses several approaches for extracting the fingerprint feature maps from the spatial-domain images.