A Dense Stereovision System for 3D Body Imaging

Abstract

Three-dimensional (3D) body imaging has become an important mean for applications based on body measurement, such as apparel customization and obesity assessment. In this article, we present a 3D body imaging system built upon stereovision technology. The system utilizes paired, high-resolution single-lens reflex (SLR) cameras to image the front and back body surfaces of a person, and robust and efficient stereo matching algorithms to reconstruct the 3D surface of the body with high-density data clouds. The system’s accuracy and repeatability have been evaluated on mannequins and human subjects in comparison with other measurement methods. It was found that the geometrical measurements from reconstructed 3D body models, including body circumferences and volumes, were highly repeatable and consistent with the manual and other 3D instrumental measurements (<inline-formula> <tex-math notation= LaTeX >$CV < 0.1\\%$ </tex-math></inline-formula>, and <inline-formula> <tex-math notation= LaTeX >$R^{2} > 0.99$ </tex-math></inline-formula>).