BIOMETRIC TEMPLATE SIMILARITY BASED ON FEATURE LOCATIONS

摘要

The present invention relates to a method and a system of determining correspondence between location sets. A basic idea of the present invention is to provide a scheme in which correspondence between location sets is determined. A feature location set (X) comprising a number (n+1) of components is transformed into a feature vector that can be used in an HDS. Therefore, a feature density function (ƒjs(x)) is created. A feature vector (XF) for the HDS is chosen to be a sampled version of the feature density function (ƒX,g(x) ), which results in feature vectors of equal and finite dimensions regardless of the number (n+1) of features that are present in the biometric template XT from which the location sets is derived. Thereafter, a distance (d) between two feature location sets (X, Y) is determined. The distance (d) between the sets is chosen to be the Euclidian distance between the corresponding feature density functions.