Fingerprint recognition using fuzzy artmap neural network architecture

In this paper, a neural network architecture, Fuzzy Artmap (FAM), was used to classify fingerprints. A fingerprint database of the University of Balamand (UOB) was created. The algorithm uses the filterbank representation which captures the local and global details in a fingerprint as a 320-byte fixed length FingerCode which is suitable for storage. In the first stage, the fingerprint is classified using FAM to one of the five categories (whorl, left loop, right loop, arch, and tented arch). Then, the algorithm matches the fingerprint by searching in the specific category in the database. Matching is done using the two corresponding FingerCodes and thus, is very fast.