3D human reconstruction from multi-image

In this paper, a method for 3D human model and skeleton reconstruction from multi-image was proposed. Firstly , the silhouettes of different views in the image space were projected back to a 3D plane and we could calculate the intersectional silhouettes from all those back-projections, then the corresponding points between two neighboring parallel planes were matched ,so the mesh of 3D human model were reconstructed . Finally, the feature points on each image were matched with the joints in the generic skeleton model, then we got the rays starting from the camera´s position and through the feature points in the image space. The problem of the joint reconstruction could be solved by finding a point whose sum of distance to the rays was the minimum. The experimental results show that this method is accurate and easy to implement.