Localizing non-ideal irises via Chan-Vese model and variation level set of active contours without re-initializing
Biometrics is the science of recognizing the identity of a person based on the physical or behavioral characteristics of the individual such as signature, face, fingerprint, voice and iris. With a growing emphasis on human identification, iris recognition has recently received increasing attention....
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2013
|
| Subjects: | |
| Online Access: | http://eprints.utm.my/id/eprint/36845/5/QadirKamalMohammedAliMFSKSM2013.pdf http://eprints.utm.my/id/eprint/36845/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Biometrics is the science of recognizing the identity of a person based on the physical or behavioral characteristics of the individual such as signature, face, fingerprint, voice and iris. With a growing emphasis on human identification, iris recognition has recently received increasing attention. Performance of iris recognition scheme depends on the isolation of the iris region from rest of the eye image. In this research, Iris as one of the components of an eye image is chosen due to its uniqueness and stability. Iris recognition scheme involves Acquisition, Localization, Normalization, Feature extraction and Matching. Iris localization is the most significant and crucial stage in iris recognition system, because it determines the inner boundary and outer boundary in an eye image. In conventional localization methods, the inner and outer boundaries are modeled as two circles, but in actual fact, both boundaries are near-circular contour rather than perfect circles. For this research, the non-ideal iris images which are acquired in unconstrained environments are used (i.e. image with bright spots, non uniform intensity, eyelids and eyelashes occlusion). Firstly, Gaussian filter is applied as pre-processing to reduce the iris image noises and then Chan-Vese model to detect the inner boundary and localize pupil region. Next, Gaussian filter is applied again to reduce the effect of eyelids and eyelashes for faster and easier detection of the outer boundary. Finally, Variational Level Set Formulation of Active Contours without Re-initialization is applied to localize the outer boundary. Experimental results of CASIA-Iris-Interval Version 3 database show that the performance of the proposed method is very encouraging with 98.39% accuracy rate. |
|---|