Reconstructing 3D human models with a Kinect (pages 72–85)
Guang Chen, Jituo Li, Bei Wang, Jiping Zeng, Guodong Lu and Dongliang Zhang
Article first published online: 10 FEB 2015 | DOI: 10.1002/cav.1632
In this paper, we provide a method of reconstructing 3D naked human models for human bodies or dressed humans using the depth images captured by a single Kinect. The depth images from multi-views are part-wisely registered together, and implicit-surface-based de-noising method is put forward for the coarse data. Afterwards, we fit a statistical model to the registered depth image iteratively in order to obtain the reconstructed 3D human model and apply the reconstructed models in virtual try-on systems.