Implementation of Signature Facial Mannerism Extraction Model

Abstract

Facial expressions not only show an individual’s intents and their emotional state they can also be used for person identification, caricature generation as well as diagnosis of diseases. Each person can not only be distinguished from another with respect to their gender, age, ethnicity, appearance, facial features, but they can also be identified by their expressions and gestures, also known as facial mannerism, that are uniquely their as everyone responds to any given situations in their own unique distinguishing way. An individual's mannerism usually reflects their subconscious behaviour or conscious and deliberate gestures they acquire or are schooled over time. Individual differences in facial mannerism reveal the idea of recognizing individuals on the basis of their facial mannerism. So far no automated system has been effectively designed to recognize facial mannerism to extract differences. This thesis proposes a model to extract significant facial mannerism that could then be used by various applications: surveillance systems, medical diagnostic systems, graphic modeling systems for caricature generation; these are few applications of signature mannerism. Proposed system is designed by making use of approaches and techniques being currently used in the fields of computer vision, pattern recognition and psychology. Static images are processed to extract expression information and then this extracted information is used to compare multiple images to extract signature facial mannerism from these images. System is designed such that it could be implemented using different techniques not just the approach used in represented thesis work.