Three-dimensional shape analysis and anthropometry: a case study

Abstract

This paper presents leg shape analysis from data collected in the 2012 US Army Anthropometry Survey (ANSUR II) as a case of digital body modelling. We applied principal component analysis (PCA) on 3D leg surfaces (3D PCA) and on relevant anthropometric measurements (1D PCA) of 3,890 male subjects. We compared the first seven significant shape modes (accounting for 94% of shape variation) of 3D PCA with 1D PCA of six anthropometric measurements of the right leg. The analysis shows that each set of PCA results has its own power to describe the shape variation. To explore the relations between anthropometric measurements and 3D shape variation, we also applied a multiple linear regression (MLR) model to 3D PCA scores against anthropometric measurements. The MLR results show a strong correlation between certain measurements and shape modes. Further numerical analysis also revealed a linear relation between the PC weights of length and circumference modes, and the volume and area of leg surfaces.