E.K. Patterson and A. Gaweda (USA)
face, expression, gesture, identification, biometric, classification
This paper presents a novel technique for the use of facial dynamics, such as demonstrated in typical expressions and gestures, as a biometric for person identification. Much work in psychology and pattern recognition has been directed toward studying the face, expressions, gestures, and their relationships, but in computing this research has largely been applied to affect recognition but not to biometrics for the purpose of human identification. Merging ideas from several areas, this work demonstrates the possibility of using facial gestures and common movements, even blinks, as a biometric for identifying individuals. The method described here applies Active Appearance Models for feature representation and Hidden Markov Models for classification. The results, although acquired from a small data set, suggest that this method and concept may be successful for future applications and merit further research.
Important Links:
Go Back
