Study Of Algorithm Of 3D Reconstruction From Single Image Based On Variable Albedo

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 and has wide application value. Recently, duing to its special feature (it can restore three dimension shape information from one image) shape from shading(SFS) gets great improvement and rapid development in methodology and application . SFS becomes a new research hotspot of 3D passive reconstruction.At the beginning, a full survey about SFS research work is given. The recent hotspots of this field are summarized. Based on understanding SFS problem in nature, a systematic summarization is given for all current SFS algorithms which are minimization approaches, propagation approaches, local approaches, and linear approaches in the world. And this paper gave the analysis and comparison of these reconstruction algorithms. Most traditional algorithms assumed that the albedo of object surface is a single value in an image, but this assumption made application of traditional SFS algorithms restrictive. This paper presents a local albedo estimation method and computes the albedo for each pixel of an image and works out a SFS algorithm with variable albedo. Then this paper presents two sets of experiments, The first group which was done on several synthetic images for 3D reconstruction shows that SFS linear algorithm converges to the correct answer. The other group which was performed on several real images shows that the new method is modified and effective and extends the application range of SFS comparing to the traditional methods.The SFS method introduced in this paper can be used in different fields, such as simulation of natural scenery, recovering of cultural relics, arts, human face recognition and some fields which people can't reach. Along with development of computer image process technology and improvement of study on SFS problem, the SFS method will have a wide application foreground.