3D model reconstruction refinement from multiple images

3D model reconstructions have attracted various attentions in recent years. In this paper, we present a mechanism to reconstruct 3D model through the refinement of corresponding points using 3D patches. The NCC values between patches in the reference image and that in the visible image are used to estimate the best corresponding points and, hence, the better camera parameters. A more realistic 3D model can be reconstructed using this mechanism.