3D Geometric Modeling Of Buildings With A Single High Resolution Remotely Sensed Image

With the rapid development of space technology and remote sensing technique, the launch of many high spatial resolution satellites in succession, an urgent task is how to get the information from massive remote sensing data quickly. On building3D city model, city buildings are the fist and foremost, which is the representative of the city scenery. In order to gain more visualized information from remote sensing images, the3D building modeling based on a single image has caught more and more attention.This thesis aims to how to use a single high-spatial-resolution remotely sensed image to obtain buildings’height in batch by means of the shadow information. MATLAB,VisualC++are used as data processing tools. As the geometry data of buildings’roof are used instead of the spatial position data,3D models of city buildings can be fast obtained. The specific process is as follows:Firstly, the building shadows should be extracted automatically. Based on analyzing the gray value and geometrical characteristic of building shadow, which is existed in pan-image, ostu and area threshold method are used to extract the shadow areas of building. In order to avoid the complexity of various operators in bias test, mathematical morphology is also used to extract the shadow boundaries, with fast computation velocity, good effect and high versatility.Secondly, the method of automatic evaluate for the shadows is focused on to calculate the height of building. Based on the shape feature of shadow, the idealized shadow length is calculated by using area method. Considering the shadow edge is uneven, the method of maximum eigenvalue is used to detect the corner of shadow boundary, and the mean of the distances between the nearest corners is taken as optimum length value. Through the comparison of the tow computed values above and the measured one practically, the best method could be determined. The mathematical model, which aims at calculating the length of shadow, is given on account of two aspects:when the building shadow is taken shaped, making a full analysis of the position relationship among satellite sun and the building; considering the head angle between building and shadow further. Through picking up different measuring areas, the accuracy of different methods can be then verified.Finally, the geometry data of building surfaces are extracted. In this study, as it’s difficult to extract the surface geometry data from remote sensing image directly, the top building geometry data is taken the place of the bottom. In a general way, the buildings are not orthographic in remote sensing image. As a result, there are some offsets between the top and bottom of building. In this thesis, based on the relationship between building height and satellite altitude angle, a method is used to calculate the offset values, and then adjust the offsets of the spatial location of buildings in the remote sensing image.In this thesis, a QuickBird image is studied to verify the method of3D modeling through testing. Then3D model building has been finished by3D analysis tools, as a result, a better fruit is shown in this process. At last, the limit conditions of applying the method are analyzed systematically.