Automatic anthropometric personalization of a digital human model from a set of subject’s photographs

Abstract

The objective of the work is to develop a method for automatic dimensioning of a digital human model (DHM) from a set of calibrated photographs of the subject under study. Fifteen subjects (10 males, 5 females, mean age 27) wearing surface retro-reflective markers at major bony landmarks and standing inside a calibrated space, were photographed by means of low cost main stream digital cameras (face, left and right views). The DHM software used is based on a skeletal structure surrounded by contours defined by cross sections along the skeletal links. Specific points corresponding to bony landmarks are defined in body linkage local reference frames. The photograph calibration method (DLT) allows to superimpose a picture of the DHM on the subject’s photographs. A specific algorithm adjusts the length of the manikin skeletal links in function of the measured distance between markers attached to these links (e.g. knee and ankle for the lower leg). Then the contour sections dimensions were adjusted to fit subject’s silhouette extracted from the photographs. The results obtained within a few seconds of calculation consist of a personalized DHM representing the subject with an error less than 2% for stature and less than 4% for weight.