Texture Image And Point Cloud With Three-dimensional Laser Scanning Technology Accurately

3D reconstruction of real scenes by 3D laser scanning technology has been used widely in many fields, such as Cultural Relic Protection, Archaeological Research, and Virtual Reality. Recently, with the reduced cost of data acquisition equipment, there has been an increasing attention to this field. However, most of the researches are done for 3D surface reconstruction, neglecting texture features of the reconstruction model, so the need of photo-realistic 3D reconstruction hasn't been satisfied. Texture mapping is a necessary step to show photo-realistic results of the 3D reconstruction, which can be realized only after registration of texture and 3D surface model (or pointclouds) is done.The above-mentioned problem is studied in this paper, and the main contributions are as follows:(1) The theory and technology of 3D reconstruction is analyzed, taking the 3D laser scanner in our laboratory as an example. The basic theory of registration of textures and pointclouds is also summarized.(2) For texture mapping of indoor environmental data, an algorithm is presented based on reflectors as tiepoints, which extracts centroid coordinates in texture images and pointclouds respectively and takes them as registration elements to compute the transformation between the two coordinate systems, so as to achieve the registration of textures and pointclouds.(3) For texture mapping of outdoor environmental data, especially those with a great number of lines and planes, a registration method based on linear and planar features is designed. Firstly, the lines in the texture and pointclouds are extracted respectively, then the 3D features of real scene are computed by 2D image according to the properties of vanishing point, and finally the rotation and translation transformation between the two coordinate systems are both computed to achieve the registration.The above algorithms are all realized on the platform VC++ and OpenGL, tested with the data acquired by our laser scanner, and also proved valid by experimental results.