Predicting seated body shape from standing body shape
Authors: Matthew P. Reed, Byoung-Keon Park, B. D. Corner
A large amount of body surface scan data in standing postures is available from population surveys, but relatively little data in supported seated postures has been gathered. This paper presents a method for predicting seated body shape in a posture typical of automobile driving using data from a standing scan. A principal component analysis (PCA) of template-fitted standing data for 120 women was conducted and 60 PCs were retained. Data for the same women in a seated posture were analyzed using the same technique. A regression analysis was conducted predicting seated PC scores from standing PC scores and the method was evaluated by comparing predicted and measured body shapes for 18 women with a wide range of body size measured in a separate study. The method was found to produce accurate predictions of surface shape, with median errors less than 5 mm after accounting for posture differences. The method shows promise for obtaining predictions of alternative postures for populations for which only a few postures have been scanned.