Abstract
The aim of this study was to examine a potential of three-dimensional human body model (3DHBM) produced from two-dimensional photography to estimate percentage body fat (%BF). Photographs from front and side views were taken from 14 Japanese female college students (aged between 18-22 years old). Pictures taken from the front view was converted to a polygon mesh human body model (PMHBM) and adjusted with measured anthropometric parameters in order to produce 3DHBM. Determined 3DHBM was then utilized to calculate body volume (BV_<3D>) and body density (BD_<3D>) as well as %BF using Siri and Brozek prediction equations. The estimated body composition variables from 3DHBM were compared with a multi-frequency bioelectrical impedance analysis (MFBIA). While BV_<3D> was highly correlated with BV_<MFBIA> (r=0.993, p<0.001), no significant correlation was observed between BD_<3D> and BD_<MFBIA>. Average %BF estimated from 3DHBM was six percent lower than the value obtained from MFBIA. The current study recommends further investigation to minimize differences in body composition predictions from photographs.
