Image-Based Three-Dimensional Reconstruction Of The Building

With the development of computer graphics and virtual reality, static two-dimensional image has been unable to meet the visual requirements, people want to reproduce the complexity of the real scene. This paper focuses on reconstruction of image-based measurements to start stduy.First it introduced the basic theory and the general process of Image-Based Modeling, and the current algorithms in each stage were summarized and compared. Then offered a proposal of image-based modeling towards buildings. In this scheme, we calibrated the Targeting methods based on plane using Zhang's method, The method is simple, flexible and high precision calibration,Target is also easy to make a lot of the current system of choice for the camera pre-calibration algorithm,matched the homonymous points through the method of SIFT combining of Harris corner detection.Finally,solved projection matrix and key points 3D coordinates. Then we Newman boxes and units were selected for the reconstruction of the object dormitory. Prior to the camera calibration, field sampling a number of sequences of images through the same match, coordinate calculation output three-dimensional coordinates of key points.then reconstructed the 3D two physical models by MultiGen-Creator tools.