Application of Image-and-Map Registration on Military Land Management
| Autor: | Hsu, Szu-Chi, 徐思齊 |
|---|---|
| Rok vydání: | 2004 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 92 Military Land Information System(MLIS) has been established for a decade, and vector data of MLIS demand prompt revision. The progress of data revision carried out by human operators cannot satisfy the need of MLIS, it is necessary to find a better solution. High-resolution commercial satellite imagery is a significant source to acquire up-to-date data for spatial information system. The integration of satellite imagery and MLIS, providing information of land status rapidly, is an interesting topic. The first issue of such kind of integration is the image-and-map registration. Traditional method of registration is carried out by a human operator. Other work been done for automating the processes focus on low-level image processing techniques. The bottleneck of previous work is how to define the desired target well in image. This research presents a method of high-level image understanding, following a top-down control strategy and utilizing existing knowledge as much as possible to automate image-and-map registration. The truth—walls of military land built on the boundary lines of a cadastral parcel—has been utilized as the relationship between map and image. Registration is controlled by the wall features in the QuickBird satellite image and cadastral parcels of military land in a cadastral map. Analyzing the registration error model, establishing walls model, and weighting factor, and filtering noises by using a geometric-structure-matching technique, are the needed procedures to complete registration automatically. The result shows that, the automatic registration of a geocoded satellite image and a vector map has been achieved by the proposed algorithm. It proves that if existing knowledge is available, besides using road and building features in other work, linear features such as wall could be utilized in computer vision, photogrammetry and remote sensing techniques to achieve image understanding. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|