To study eyebrow behavior from video sequences we utilize a new image analysis technique based on an anatomical-mathematical motion model. This technique conceives the eyebrow as a single curved object (arch) that is subject to the deformation due to muscular interactions. The action model defines the simplified 2D (vertical and horizontal) displacements of the arch. Our video analysis algorithm recovers the needed data from the arch representation to deduce the parameters that deformed the proposed model. The present technique is meant to suit real-time requirements and to work under uncontrolled environments: without specific known lighting or special markers on the face, by using only computer vision and image processing methods.
Eyebrow movement analysis over real-time video sequences for synthetic representation
AMDO 2002, 2nd International workshop on articulated motion and defomable objects, November 21-23, 2002, Palma de Mallorca, Spain | LNCS Volume 2492
Palma de Mallorca
© Springer. Personal use of this material is permitted. The definitive version of this paper was published in AMDO 2002, 2nd International workshop on articulated motion and defomable objects, November 21-23, 2002, Palma de Mallorca, Spain | LNCS Volume 2492 and is available at : https://doi.org/10.1007/3-540-36138-3_18
PERMALINK : https://www.eurecom.fr/publication/1014