The Methodology for Facial Features Detection
Jacek Naruniec · InTech eBooks · 2011
Will-be-set-by-IN-TECH•e y e s c e n t e r s ,• multiple face fiducial points,•c o n t o u r s o f t h e f a c e p a r t s ,• a set of rectangles covering individual parts of the face.While from human point of view the area parameters are more convincing, for face recognition system, fiducial points are more important since they allow to perform facial image normalization -the crucial task before facial features extraction and face matching.Facial features localization algorithms are commonly divided into four groups (Ben Jemaa & Khanfir (2009); Celiktutan et al. (2008); Naruniec (2010)):• appearance-based,•g e o m e t r y -b a s e d ,• knowledge-based,• 3D vision-based.This chapter aims at defining more general scheme for facial features localization, since all of the defined groups have common methodology.Moreover most of the efficient schemes doesn't rely on one of the methods, rather combining few different approaches.The goal of this work isn't concerned about giving the detailed information about each algorithm, but rather to develop the intuition of the approach to the described subject.Review of the given steps is presented in the following sections. General facial features detection schemeTypical image analysis and in particular facial features detection usually consists of several steps: How to referenceIn order to correctly reference this scholarly work, feel free to copy and paste the following:Jacek Naruniec (2011).The Methodology for Facial Features Detection, New Approaches to Characterization and Recognition of Faces, Dr. Peter Corcoran (Ed.),