Reconstruction of 3D Landscape from Images with Large Scale for Urban Region

Reconstruction of 3D landscape for urban region is an important research in digital photogrammetry and visualization. By means of object extracting technique, corners of the house have been located automatically according to approximate points firstly in this paper. Then, the window, including much area of house proof instead of centering at the corner of house, is created to perform image matching for house corner. The reliability of image matching is improved. After the 3D data of houses are measured semi-automatically, texture restoration of visual surface on the buildings is achieved based on the 2.5D triangulated irregular network of ground and building surface and original digital image. The real 3D landscape of urban is reconstructed.