Authors: Rokaya Dhahri; Samir Belaid
Addresses: MARS Research Unit, Department of Computing Science, FSM – University of Monastir, Tunisia ' MARS Research Unit, Department of Computing Science, FSM – University of Monastir, Tunisia
Abstract: Detection and removal of a beard from three-dimensional human faces model is considered as one of face recognition and age estimation challenges. For this purpose, we propose a new method based on standard deviation measure of angles sum between face normal vectors (SANV). For this purpose, we begin with enhancement steps whose aim to fill the holes present in face model. Once imperfections are eliminated, the face is segmented in order to locate and extract beard area of the face. Then, we estimate the presence or the absence of the beard by testing the value of the SANV measure. The last step consists in a virtual shaving of the beard with removal techniques. The removal step is based on applying Taubin smoothing and the regression of SANV value of bearded area. Experiments carried on FRAV 3D database showed the relevance of our contribution.
Keywords: occlusion; beard detection; beard removal; Taubin smoothing; face normal vectors; angles; 3D modelling; human face models; face recognition; age estimation; image segmentation; virtual shaving; beards.
International Journal of Operational Research, 2016 Vol.27 No.1/2, pp.201 - 211
Available online: 22 Aug 2016 *Full-text access for editors Access for subscribers Purchase this article Comment on this article