Fingerprint recognition is one among oldest procedures of identification. An important step in
automatic fingerprint matching is to mechanically and dependably extract features. The quality of the input
fingerprint image has a major impact on the performance of a feature extraction algorithm. The target of
this paper is to present a fingerprint recognition technique that utilizes local features for fingerprint
representation and matching. The adopted local features have determined: (i) the energy of Haar wavelet
subbands, (ii) the normalized of Haar wavelet subbands. Experiments have been made on three completely
different sets of features which are used when partitioning the fingerprint into overlapped blocks.
Experiments are conducted on FVC2004 databases that have a four database; every database is eighty
fingers and eight impressions per finger. The implemented recognition results of the proposed system show
high recognition performance which is 100%.