3D Reconstruction From2D And3D Images

The reconstruction of objects and scene is a necessary fundamental withrespect to computer vision, which has a wide and important application inautomatization in robots, man and machine interface, visual reality and so on. Theprecision and real-time of reconstruction are the main tasks in research. In order toimprove the precision and real-time of the reconstruction, this thesis establishes asystem from RGB2D images and TOF3D images, addresses lots of key problem,and develops a software system of3D reconstruction.The main tasks of this thesis are as follows:1. This thesis introduced several coordinates which used very often in machinevision, did the calibration of the TOF camera and RGB camera using Zhangzhengyou calibration, and got the parameters of the two cameras.2. This thesis proposed an adaptive bilateral filter, which is used to denosie theTOF depth image. It is obvious that this method not only can denoise the TOFdepth images, but also can preserve the edge of the depth images.3. This thesis realized the registration between the2D image and3D images，which is not affected by noise and low-resolution of the images that make itdifficult to extract the feature points of the images. This thesis also proposed anew method to fuse the TOF3D image and RGB2D image based on Delaunaytriangulation and adaptive Gaussian filter.