Cross models for twin recognition

Date
2016
Authors
Gu, Datong
Supervisor
Yan, Wei Qi
Item type
Thesis
Degree name
Master of Computer and Information Sciences
Journal Title
Journal ISSN
Volume Title
Publisher
Auckland University of Technology
Abstract

Nowadays, biometrics has become a popular tool in personal identification as it utilizes physiological or behavioral characteristics to identify individuals. Recent advancement in computer science has increased the accuracy of biometrics to a higher level. However, there are still a number of existing problems, such as complex environment, aging and unique problems. Twin identification is notably one of the most challenging issues, because they resemble each other in terms of biometrics. The similarity affects the use of biometrics in general cases and raised the potential risk of biometrics in access control.

This thesis presents and compares four methods for twin recognition, namely, ear recognition, speaker recognition and lip movement recognition. Our results show that speaker recognition has the best performance with 100% accuracy. This is much higher than that of face recognition and ear recognition (with 58% and 53% respectively) while movement recognition that yields 76% accuracy.

The objectives of this thesis are to investigate whether it is possible to identify twins between each other by using biometrics and find out which recognition approach is the best one. Comparing to the-state-of-the-art technologies, our work has taken a further step in twin recognition by using biometrics.

In future, we will take behavioral analysis of twins into consideration, since their growing-up environment may have impact on their behaviors. We assume that other clues for twin recognition could be discovered from the behavioral analysis.

Description
Keywords
Cross models , Twin recognition , Speaker recognition , Face recognition , Ear recognition , Lip movement recognition , Precision , Recall , F-measure
Source
DOI
Publisher's version
Rights statement
Collections