A multi-camera network system for markerless 3d human body voxel reconstruction

This paper presents a fully automated system for real-time 3D human visual hull reconstruction and skeleton vox-els extraction. The main contributions include: (1) A novel network based system is presented, which uses AXIS net-work cameras as video capture device, and performs a parallel processing among data capture, 3D voxel recon-struction and display. (2) A new human visual hull re-construction algorithm is given. This approach firstly seg-ments the foreground accurately by an efficient Gaussian Mixture Model (GMM) and a shadow model in HSV color space, then extends the standard Shape-From-Silhouette (SFS) algorithm with online Region-of-Interest (ROI) esti-mation and binary searching, and finally construct skele-ton probability visual hull with distance transform. Exper-iments with real video sequences show that the system can process eleven 640x480 video sequences at a frame rate of 15fps, and construct human body voxels reliably in complex scenarios with cast shadows, various body configurations and multiple persons. 1