The Research of 3D Reconstruction Based on Single Image

Traditional silhouette extraction algorithm is easy to cause disconnection problem, this article adopts the method of threshold segmentation and morphology repair to extract a more complete outline. The conventional SFS minimization algorithm extracted 3D information with a consistent smooth factor will make the effect is not very well, so this paper will provide a method that uses the inconsistent smooth factor to obtain a better 3D model. This paper combined the optimized silhouette information with the 3D information which obtained by improved SFS minimization method, then filter the background points, retain target points, and finally get the 3D model that only contains the target points. Experimental results show that the combination of this two kinds of improved algorithms can get a more real 3D model.