Biometrics using Complex eye movements

Abstract

This thesis proposes a solution of how eye movement data can be used for biometric identification. Eye movement data have behavioral characteristics that are useful for biometrics. Different statistical features are extracted from eye movement data. These features are extracted from two main characteristics of eye movement data i.e. saccades and fixations. Different machine learning algorithms are used to train model on different set of features to come up with best possible combination that produces comparatively better accurate results. The thesis focuses on providing mechanism to convert available data into vector form such that machine learning algorithm can be trained over it.