Since Antiquity, studies concerning human face symmetry have tried to explain its attractiveness by means of sociological and psychological investigations. One of the most important human body features to be evaluated is human face symmetry plane. Thanks to the advent of
three - dimensional acquisition techniques the issue of facial symmetry evaluation has been approached for several purposes such as:
Our approach
- face authentication and recognition by middle profile feature extraction;
- quantification of asymmetry in human face and development of computer-aided protocols for half-damaged face reconstruction or cosmetic surgery for aesthetic corrections in MFS (Maxillofacial Surgery);
- correlation between facial asymmetries and symmetry line for the back pathologies in Orthopaedics and Orthodontics;
- correlation between facial asymmetries and cognitive disorders for schizophrenia diagnosis in Neurology.
Generally speaking symmetry is an ideal property of a category of objects which is characterized by a symmetry plane.
Real objects never are ideally symmetric but they are affected by defectiveness of non - ideality inherent in the object itself. Another contribution to non - ideality is given by the way the object is translated into numerical model.
These non - idealities are due to:
In order to overcome the above mentioned limitations we have proposed more robust method for estimation of facial symmetry plane that, in the light of the results we have obtained, can be considered practically insensitive to local asymmetries, whenever they are localised on the face (e.g. local damages, pimples, bumps, etc. and asymmetric facial expressions), reproducible and not affected by the acquisition process.
- generic errors inherent the acquisition process (e.g. background noise, edge effect);
- asymmetries of the acquired area of the object;
- non-uniformity of sampling of acquired object;
- asymmetric sampling of acquired point cloud.
In the following figures a screenshots (click to enlarge) of results obtained by our method.
Details of our method are reported in:
3 August 2013:
The page "Applied Differential Geometry for Tessellated Models" has been updated.
11 February 2013:
The new website is online.