Research On3D Model Reconstruction Technology Based On Multiple Images

Multiple images-based3D modeling is one of the most important research fieldsin Computer Vision and Computer Graphics. With the rapid development of VirtualReality, Computer Game, Digit Library and so on, which are more and more popular,causes the steady increase in demand of highly realistic3D models，that is the requirements ofquality and efficiency. Although the constant progress in the traditional surface modelingtechniques, it remains a painful and time-consuming task. On the other hand, despitemodern3D scanning devices can reconstruct the realistic3D models rapidly, it cannotreconstruct the effective models for some objects, which are transparent,strong-reflective surface and villous, and modern3D scanning devices are expensive,these are limitations which cause unable to popularize. So Multiple images-based3Dmodeling receives the widespread concern. Compared with other modeling techniques,its main advantage is that the geometry models, surface texture and reflectance can beobtained from images directly, and reconstruct a highly realistic three-dimensionalmodel based on this information rapidly.In this paper, Multiple images-based3D modeling are studied deeply, especiallycurved surface reconstruction based on two views is analyzed in detail, and get someresearch achievements, The main research contents and achievements are as follows:(1) The researches and developments of the multiple images-based modelingtechniques are summarized at home and abroad, and some main problems and keysteps of the techniques are researched and analyzed.(2) In allusion to shortages of the traditional3D reconstruction technology, somemethods of the multiple images-based modeling techniques are studied mainly, theadvantages and disadvantages of these methods are analyzed.(3) The method of curved surface reconstruction based on two views areresearched and explored, the shortages of its linear constraint mechanism are analyzedand discussed, and then we optimize the linear constraint mechanism and obtain thegood effect by experiments.