Researches On Key Technology Of3D Reconstruction Based On Objects

With the development of computer vision and computer graphics, and3d models have a wide range of application in3D games, digital library, virtual reality, product appearance design, and so on,3D reconstruction technology is gradually becoming a hot research field. Based on a amount of literatures’ research, this thesis has a study on3D reconstruction with two kinds of methods respectively from image which is based on image and3D laser scanner based on structured light. The thesis has completed the following work.(1) The3D reconstruction technology of image only needs a common video camera and gets an image or several images of the object or scene for needing to rebuild. By making use of the information of2D images contain, we can reconstruct the3D model of the object or scene through related algorithm. On the research of image to3D reconstruction, this thesis is based on the theory of computer vision, and uses two kinds of methods to analyze and complete. The two ways are respectively with several images and a single image to3D reconstruction.1)3D reconstruction based on two or several images. It mainly involves to camera calibration, feature extraction and matching and other relative3D reconstruction algorithm. Before the camera calibration, it needs to preprocess the images to remove noises, because it is good for the extraction of feature points correctly. In this paper, firstly, it uses Zhang’s method to calibrate the camera by the homemade calibration board, then detects and extracts and matches the feature points by SIFT method which does not influenced by translation, rotation, scale and illumination. After completing matching and calibration steps, matching points-coordinates and camera matrix would be gained. We could calculate the3D coordinates by least squares method. And finally we can obtain the spatial3d objects through the triangularization and texture technology.2)3D reconstruction based on a single image. Because it is unable to obtain the depth value of reconstruction object, this article uses the method which is firstly to do image segmentation process to separate the target area and the non-target area. The color values of the image in the target area is regarded as height field to add to the two-dimensional grid image, and then render a3D mesh to get a3d model. Finally, give an illumination rendering to the3d model. This method is mainly to3D reconstructions of relief or carved images and sculptured objects on ancient objects such as on the stone and potters. (2)3D laser scanner based on structured light to3D reconstruction. The traditional3D laser scanning has advantages in three-dimensional scan and high precision, however, the high price greatly limits its application and popularization. Based on this problem, a low cost3D laser scanning system is presented which is based on structured light principle. It consists of common camera, linear laser emitter, and calibration boar. This system can realize3d scanning to small objects and complete3D reconstruction.In the implementation of3d reconstruction method, the method which is based on image uses matlab and Visual C++software as development tools to realize3D reconstruction, and applies OpenGL technology to display3D object. The3d scanning system which is based on the structure light combined with DAVID software; The result of the experiment shows that the approaches presented in this research is reliable and has good effects.