Automatic Outline Extraction Methods And System Implementation For Airport Runway And Tarmac In High Resolution Images

In recent years, with the development of remote sensing techniques, spatial resolution becomes more and more high. High-resolution satellite images have become an important data source for information extraction for the artificial surfaces. The use of remote sensing images for artificial feature extraction has become one of the important applications in the field of remote sensing. By extracting the artificial surface features, vector maps can be quickly generated, which can be widely used in different tasks, e.g., the national production and military field applications. As an important man-made object, the airport is closely connected with human life, and is also interested by the research of military target attack. In the airport, runways and aprons are the two of the most important parts, as well as the most frequently changing information during the expansion of airports. Also, in the update task of GIS (Geographic Information System) information, the modification of vector data of these two objects becomes a main issue. Therefore, it is quite necessary to mark them out separately. Currently, most studies focus on the recognition of the airport itself, by the means of finding airports and extracting aircraft bunkers. However, there are still few researches on the automatic accurate outline extraction of airport runways and aprons. Hence, based on existing studies, this paper proposes a hierarchical extraction method for the outlines of airport runways and aprons, which is a coarse-to-fine process. Firstly, by image segmentation, non-airport classes are gradually removed, resulting in the extraction result of airport area. Then, based on the extracted airport area, a region constrained automatic outline extraction of runways and aprons will be conducted by using proper feature extraction methods.The main work of the study is completed as follows:(1) This thesis analyzes the main features of airports in high resolution remote sensing images and features of the basic components of the airport, and design an automatic hierarchical outline extraction method for extracting airport runways and aprons.(2) This thesis analyzes the common methods of remote sensing image segmentation, and design to use the combination of the multi-scale image segmentation method and the spectral difference segmentation method to segment the airport images. Based on segmentation experiments of remote sensing images, the best segmentation parameters are obtained for airport images.(3) Based on the image segmentation result, this thesis designs an airport area extraction strategy by hierarchically removing non-airport areas class by class.(4) Automatic extraction for runway and apron outlines is performed under the constraint of airport areas, by comprehensively applying various image processing algorithms, including Hough transform, mathematical morphology, connected domain calculation, so as to complete the outline extraction of the airport runways and aprons.(5) Redevelopment based on the eCognition software, integrated all kinds of algorithms, the process of complete system implementation.Through this study, the proposed hierarchical process based outline extraction method for airport runways and aprons in high resolution satellite images can, to a certain extent, effectively deal with the tasks of automatically updating GIS data of airports based on high resolution satellite images, which also provides technical ideas for automatically extracting the outline information of other types of surface features.