Three-dimensional Surface Reconstruction From Single Image

As one important research direction of computer vision, 3D reconstruction plays more and more important role that can't be ignored in the various fields of production and life in society. Due to its special feature, the three dimensional shape can be reconsturcted just needing a single gray image. Recently, shape from shading (SFS) gets great improvements in methodology and application, and the method becomes a new research focus of 3D reconstruction.At the beginning of this paper, a comprehensive survey about SFS research work is given. The recent research focuses in this field are summarized. On the base of understanding for SFS problem in nature, a systematic summarry is made for all current SFS algorithms at present, which are minimization approaches, propagation approaches, local approaches, and linear approaches. Based on local approaches, an improved SFS approache is presented for the reconstruction of biomedical engineering images, such as human face, body models, or local endoscopy images. Not as most traditional SFS algorithms, which adopt rigid assumptions, the approach obtains better reconstruction results in terms of robustness, convergence and reconstruction speed. Some reconstruction applications in other fields are given as well, in which this approach also performed well.The method of 3D reconstruction researched in the paper is an effective method and has a wide applicability. Along with developments of computer image process technology, the reconstruction precision will be improved and the method will have a wide application foreground.