Eyes and Eyebrows Detection for Performance Driven Animation

Real-time localization of eyes and eyebrows in video sequences can find several applications in nowadays. However, this task is still a challenge, mainly if we consider images of low quality as those achieved by web cams. This paper describes a methodology to accurately detect meaningful feature points in eyes and eyebrows of people in frontal pose, in images captured by web cams. Also it discusses the mapping of the movement of the eyes of a real person to an avatar, in order to provide Performance Driven Animation (PDA). Furthermore, this paper presents obtained results and the limitations of the proposed methodology.