The Study Of Based On The Object's Surface Shape Mode Of Three-dimensional Reconstruction From A Single Image

3-D reconstruction technology as the key areas of computer vision research has been shown the important status and role in various aspects, and has a broad application. The research of this paper is based on the using of computer vision technology, it is a method and technology,which is extracting the objective information from single image to the object of three-dimensional reconstruction.The main research work of thesis are:1) On the basis of comparing the various edge features detection algorithm methods,it use the otsu ways to Segment the image to access objectives. In this paper, based on the RGB color model to split the images,and use the filtering principle to remove the noise of image, at the same time, Extract the edge of the object with LOG detector.2) Analyzing and comparing the various methods of the camera calibration. The paper uses the camera calibration of the based cube, cube on, demarcates the internal and external parameters of the camera by using of the known coordinates of six characteristics points.3) Based on a single image for its surface shape. For the split to the goal region, if it does not considers the brightness distribution, the geometric reconstruction will have two cases, one is a two-dimensional plane, and the other is three-dimensional surface. This paper puts the geometric model assumptions, firstly, uses the geometric model to construct surface structure, reuses objects Lambert diffuse radiation model and the gray-scale information of the objectives image to reconstruct the light's direction and intensity of the objects, and then verifies the correctness of geometric model assumptions. This article to the surface reconstruction methods of the simple rules objects further promoting the reconstruction of an irregular surface, which irregular surface is split into several regions, each region are composed of a unified secondary surface, reconstruct the parameters of the model with the combination of surface Gray-scale distribution, while each region's parameters have been determined, the three-dimensional surface of the entire objects will immediately identified.