3D Shape Reconstruction of Moving Object By Tracking the Sparse Singular Points

In this paper we propose a method to reconstruct the 3D shape of object using its different silhouettes through the rigid movement. The moving object is captured by two cameras during time. A robust curve stereo matching algorithm is employed to extract the precise location of some singular-points for any sequence. The motion of object is estimated by tracking these points and a large number of cameras can be constructed for the moving object during time. Finally, the silhouette cones of all virtual cameras are intersected to extract the fine visual hull. In the proposed method, the quality of reconstruction is improved by fusing the advantages of silhouette, motion and stereo. Because of using the curve matching scheme instead of color matching, our method is less sensitive to color adjustment between cameras and illumination changes of light source. Our method is applicable also to the low-texture object