Fingerprint classification using a homogeneity structure of fingerprint's orientation field and neural net

Abstract

Fingerprint classification is important part of fingerprint identification systems that work on large databases. It provides fingerprint indexing, which results in efficient matching. This work presents usage of a homogeneity structure of fingerprint's orientation field for fingerprint indexing. The homogeneity structure is described through a quad-tree structure. A description of the quad-tree structure is the input vector for neural net that was used as a classification system. The system is tested with fingerprints of three different quality levels to provide real results.