Facial Symmetry in Robust Anthropometrics

Authors


  • Fully supported by “Center of Biomedical Informatics,” Project 1M06014 of the Ministry of Education, Youth and Sports of the Czech Republic.

Additional information and reprint requests:
Jan Kalina, Ph.D.
Center of Biomedical Informatics
Institute of Computer Sciences AS CR
Pod Vodárenskou věží 2
182 07 Praha 8
Czech Republic
E-mail: kalina@euromise.cz

Abstract

Abstract:  Image analysis methods commonly used in forensic anthropology do not have desirable robustness properties, which can be ensured by robust statistical methods. In this paper, the face localization in images is carried out by detecting symmetric areas in the images. Symmetry is measured between two neighboring rectangular areas in the images using a new robust correlation coefficient, which down-weights regions in the face violating the symmetry. Raw images of faces without usual preliminary transformations are considered. The robust correlation coefficient based on the least weighted squares regression yields very promising results also in the localization of such faces, which are not entirely symmetric. Standard methods of statistical machine learning are applied for comparison. The robust correlation analysis can be applicable to other problems of forensic anthropology.

Ancillary