Person authentication using ASM based lip shape and intensity information

An authentication system solely based on visual lip information is of advantage since the uttering characteristics/manner is unique to the individual and difficult to imitate. This paper presents a study of using lip shape-based and intensity features in person authentication. These features are derived from a 14-point active shape model (ASM) lip model with the use of principal component analysis (PCA). The differential change of the feature parameters reflecting the uttering characteristics are also considered in the study. A database containing the visual utterance of 40 speakers has been generated and each of the utterances is of duration 3 seconds. The visual features are then extracted from this database and a hidden Markov model (HMM) classifier is used to perform the analysis. It is observed that the best authentication result is obtained when the first 8 modes of the intensity profile is used together with the lip shaped-based parameters.