Αποτελεσματικές Τεχνικές Εξαγωγής Χαρακτηριστικών Και Ανάλυσης Εκφράσεων Στην Επικοινωνία Ανθρώπου Μηχανής

Abstract

The main research area of this Ph.D. thesis is facial expression recognition through facial feature analysis. A novel multicue feature extraction technique is proposed which is able to perform well under a large variation of facial image quality, color and resolution. MPEG-4 features are estimated and used for fuzzy rule-based facial expression and emotion recognition. Real data and results are presented, involving both extreme and intermediate expression/emotional states, obtained within the sensitive artificial listener HCI environment that was generated in the framework of related European projects. Two approaches are presented for facial expression recognition, a possibilistic fuzzy rule evaluation approach which is able to handle uncertaininty provided by the feature extraction stage, and a neurofuzzy network, able to adapt its knowledge and estimations to specific users. Gesture analysis is also being considered as an additional cue, and towards this goal a multiple-cue segmentation technique is proposed, which can be used by higher-level gesture analysis systems. The proposed system investigates fuzzy data fusion techniques which are capable of integrating the results of multiple cue segmentations, and provide time consistent spatiotemporal image partitions corresponding to moving objects.