Human posture classification based on 3D body shape recovered using silhouette and infrared cast shadows

We introduce a new mono-camera system with multi-infrared lights for human posture recognition, which is based on the 3D body shape recovered from the body silhouette and cast shadows. We propose a new voxelization method inspired from the Shape From Silhouettes (SFS) approach for Visual Hull (VH) reconstruction. Our setup consists of 4 infrared lights installed in the different upper corners of a room or a corridor, and a camera with an infrared transmitting filter placed in the ceiling. Light sources are turn on and off, cyclically, using a simple electronic system to get one shadow by frame. To illustrate the feasibility of our approach, we used a simple scene, in which shadows are projected directly on the ground. The VHs were reconstructed using the voxelization technique. Features were extracted directly from the VH using an invariant shape descriptor. A Weighted-KNN classifier were used for classification. Promising results were provided with a classification accuracy of 91.6%.