A High Resolution Stereo Imaging System

The authors present a version of the C3D stereo-photogrammetry based 3D imaging system that has been extended to incorporate commercial high-resolution digital cameras. Previously, it had been necessary to project synthetic texture onto human subjects to achieve reliable stereo-pair image matching and hence the recovery of dense 3D surfaces. Taking advantage of the natural texture present on human skin, high-resolution imaging enables us to capture the shape of the face and body without texture projection. Although this approach simplifies 3D imaging hardware systems, it can result in increased memory, storage and computational requirements. In this context, we present new algorithms to provide robust calibration and integration of multiple range maps (when the subject is captured by multiple stereo-pairs). Finally, our results demonstrate the viability of 3D facial shape capture using C3D in combination with high-resolution cameras and standard studio flash illumination.