Use this URL to cite or link to this record in EThOS:
Title: Wavelet based approaches for detection and recognition in ear biometrics
Author: Ibrahim, Mina Ibrahim Samaan
ISNI:       0000 0004 2727 6429
Awarding Body: University of Southampton
Current Institution: University of Southampton
Date of Award: 2012
Availability of Full Text:
Access from EThOS:
Access from Institution:
One of the most recent trends in biometrics is recognition by ear appearance in head profile images. Ear localization to determine the region of interest containing ears is an important step in an ear biometric system. To this end, we propose a robust, simple and effective method for ear detection from profile images by employing a bank of curved and stretched Gabor wavelets, known as banana wavelets. Our analysis shows that the banana wavelets demonstrate better performance than Gabor wavelets technique for ear localization. This indicates that the curved wavelets are advantageous for the detection of curved structures such as ears. This ear detection technique is fully automated, has encouraging performance and appears to be robust to degradation by noise. Addition of a preprocessing stage, based on skin detection using colour and texture, can improve the detection results even further. For recognition, we convolve the banana wavelets with an ear image and then apply local binary pattern (LBP) for texture analysis to the convolved image. The LBP histograms of the produced image are then used as features to describe an ear. A histogram intersection technique is then applied on the LBP histograms of two ears to measure their similarity for recognition. Analysis of variance is also exploited here to select features to identify the best banana filters for the recognition process. We show that the new banana wavelets, in combination with other analysis, can be used to achieve recognition by the ear, with practical advantages. The analyses focus particularly in simulating addition of noise and occlusion to a standard database, and their evaluation on a newer and much more demanding ear database. We also present an experimental study to investigate the effect of time difference between image acquisition for gallery and probe on the performance of ear recognition. This experimental research is the first study on the effect of time on ear biometrics and show that the recognition rate remains unchanged over time, confirming another advantage of deploying the human ear as a biometric.
Supervisor: Nixon, Mark ; Mahmoodi, Sasan Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available
Keywords: QA75 Electronic computers. Computer science ; QP Physiology