BIOMETRIC FEATURE EXTRACTION FOR IRIS SCANS


Abstract
Biometric systems usually use unique features of individuals’ physiological characteristics to
distinguish and identify them. They range from use of fingerprints, facial recognition, retinal
scans and the one presented here, the iris.
This system developed using Matlab is able to accept an already scanned iris image process it
and compare with another. The processing in question involves localizing the circular iris and
pupil region then extracting the unique features of the iris between pupil/iris border and
iris/sclera boundary. Hough Transform was instrumental in obtaining the circles in the images.
Noise sources such as eyelids and eyelashes are occluded. The extracted iris region is then
normalized into a rectangular block of constant dimensions to take care of inconsistencies in the
iris images used. Log Gabor filters are then used to encode the iris unique feature into a bit-wise
biometric template.
Hamming distance between two iris codes was calculated and used in determining if two input
iris images were a match. The system worked well with the 28 out of 30 images tested giving
perfect recognition.

DOWNLOAD PROJECT

DOWNLOAD PRESENTATION

AttachmentSize
BIOMETRIC FEATURE EXTRACTION FOR IRIS SCANS.pdf507.79 KB