KC.1 Spatio-temporal Facial Feature Extraction
Automatic facial expression analysis and recognition is one of the most important tasks in affective computing, which provides an underpinning technique for the interaction between human and computer. For real time facial expression analysis and recognition, the extraction of various facial features from a facial image sequence is essential and critically determines the performance of a facial experssion recognition system.
Facial expression is produced by the activation of facial muscles triggered by the nerve impluses, and facial muscle actions cause the movement and deformations of facial skin and other features. Thus, salient features of a facial expression can be modeled via some critical facial points and/or regions. In this project, a number of existing algorithms for extracting various facial features are investigated to establish a robust facial feature extraction system. Theses algorithms include facial feature point tracking, dense flow tracking, and high gradient component analysis in the spatio-temporal domain. The deliverable of this project will be a robust facial feature extraction system to produce a spatio-temporal facial feature representation from facial video clips.
Prerequisites: Good programming skills are essential. It would be an advantage to have an
interest in computer vision and pattern recognition as well as good mathematics background .