Real Estate Measuring And 3D Building Fine Reconstruction Based On Digital Camera

The theory of digital close-range photogrammety is widely used in digital city, ancient architecture and ancient artifact, industry part inspection. Building reconstruction and measuring with digital images is a hot research field in digital close-range photogrammety and computer vision. The research in this area is valuable.The theory of photogrammety, technology in 3D reconstruction and real estate measuring from digital images are deeply studied in this paper. The main contents of this paper include building reconstruction from single image, 3D building reconstruction from 2D vector map and Non-metric camera images, real estate measuring and 3D building fine reconstruction based on digital camera.There is some difference between 3D reconstruction of building from single image and 3D reconstruction of building in tradition close-range photogrammetry. With manual mark and metric camera, the building is usually reconstructed based on binocular stereo photogrammetry in traditionary close-range photogrammetry; 3D reconstruction of building from single image is based on line-photogrammetry, without any control information about building, image information is very important. 3D reconstruction of building from single image includes computing interior and exterior elements, reconstructing building based on CSG (Constructive Solid Geometry). The Key technique is to determine vanishing points of three vertical groups of parallel lines. The original point of the object coordinate system is a corner of a building, three vertical lines of building are X,Y and Z axis of the object coordinate system.Coplanarity based on line feature is the foundation of 3D building reconstruction from single image. The stability of solution to Coplanarity can be improved by reducing the number of unknown parameters. Buildings in single image consist of multiple well-regulated geometry solids; these solids are illustrated by the parameters (rotation, translation in the object coordinate system and the length, width, height of solids).Usually in building reconstructing, not only a building but also multiple buildings are in a single image. But how to define a same coordinate system is difficult. The paper proposes the new method of building reconstructing based on map data and CSG technique. The initial values of image orientation parameters are decided by DLT, the lines of multiple buildings improve the values of image orientation parameters. The purpose of the method is to reconstruct multiple buildings.This paper discusses the method to get the initial value of camera parameters from direct linear transformation (DLT) in 3D reconstruction of building based on 2D vector map and Non-metric camera images. The values of camera parameters and the height of building can be improved by collinearity equation.When the interior parameters (xo, y0) are fixed to the center of image, the least squares adjustment of collinearity equation is stable. Otherwise the values of interior element fluctuate with the relativity of parameters. The edges of image are detected by the least squares template matching. The false edges will be deleted according to the correlation coefficient of the least squares template matching.The camera parameters and the height value of building are solved by coplanarity based on three vertical groups of parallel lines. Texture mapping based on knowledge gives a satisfactory effect of vision.The theory of real estate measuring based on digital camera is discussed, and the process how to get coordinates of building in real estate measuring is described. The images for real estate measuring and 3D building reconstruction are obtained by vertical photograph, so it is easy to connect and automatically give orientation between images. Oblique images in sheltered area can improve the accuracy of plane coordinates of buildings. This paper solves the problem of connection between vertical images and oblique images through vanishing points.Two methods of semiautomatic correspondence line matching based on TIN and different depth face of buildings are proposed. TIN forms from correspondence points. According to TIN, the search range of the correspondence line is forecasted. The correspondence coefficients of the elected line in left image and thematching lines in the right image are computed; the pair of line with the biggest correspondence coefficient is correspondence line. There is no restriction to image and object in the method of line match from TIN, but this method is easy to make a wrong line match in the change of depth face of building. Another method of line match is based on the depth face of building; epipolar line match defines the correspondence line. The test proves the method of line matching based on depth face of building is feasible. This paper discusses how to obtain the initial value of unknown parameters and how to improve accuracy by adding observations in a least squares adjustment. 3D building fine reconstruction based on stereo-pair is disserted.With the development of digital photogrammetry and wide use of digital camera, building reconstruction and building measuring from images will expand application field, and serve to urban planning, urban management and development.