Abstract:
Currently, many factors like environment, physiological defects of an individual, illumination etc. often influence the reduction of recognition accuracy of a single factor biometric verification system. Face biometric template used as a single factor authentication is highly vulnerable to impersonation attack. To address the vulnerability of a single modal biometric authentication, we proposed to combine both physiological and behavioural traits of a human face which refers to face and facial expression respectively. After preprocessing raw face images, kernel principal component analysis (KPCA) is used for extracting the essential features of the face and facial expression followed by the radial basis function (RBF) to train the face images. We have conducted two case studies by using minimum distance classifier (MDC) to classify face and facial expression of legitimate users. In our preliminary results, it was shown that the proposed work is capable to accurately recognise the identity of a legitimate user with its distinct facial expression.
