FINGERPRINT RECOGNITION METHOD BASED ON MINUTIAE EXTRACTION AND NEURAL NETWORK Emad Majeed Hameed Supervisor

FINGERPRINT RECOGNITION METHOD BASED ON MINUTIAE
EXTRACTION AND NEURAL NETWORK
Emad Majeed Hameed
Supervisor: Anikin Igor Vyacheslavovich
(Kazan National Research Technical University named after A.N. Tupolev, Kazan)
Abstract: Using fingerprint in biometrics identification system has been the
most widely used. The uniqueness of a fingerprint for each person provides us with
everything we need for a perfect identification. This study uses artificial neural net-
works to provide efficient matching algorithm for fingerprint authentication. The
proposed algorithm is to partition the fingerprint image into equal regions and then
find the number of each type of minutiae for each region. The feature vector, that
contains the number of each type of minutiae, will be trained by neural network. The
database HRF POLYU is used for training and testing the algorithm. This dataset was
divided into two groups; the first is training dataset that consists of 25 users for each
user 9 images. The second group is the testing dataset that consists of 20 users for
each user one image. The algorithm showed acceptable and good results.