Facade detection in oblique aerial image using object based image anaylysis

With the development of 3D city model and 3D Geographic Information Systems (GIS), the models of buildings provide many applications like urban planning, disaster assessment, virtual visualization, and others. For city models production, aerial images and photogrammetry are definitely important data source and technology for building reconstruction, since rich geometric and semantic information from images offer a rapid method on reconstructing numerous building models. Comparing with vertical aerial images (VAI), there are more advantages in oblique aerial images (OAI). Oblique images offer 3D information of ground objects from different views, and it can provide not only the top surface information but also the side, such as the building façade. The façade information is useful for building change detection, reconstruction and texture mapping. Thus, in this study we propose to utilize oblique aerial images for façade detection through object-based image analysis (OBIA) technique. At first, we segment the image into objects using multi-scale multi-resolution image segmentation by color information. Some auxiliary information might be useful for image segmentation also analyzed in this study, such as edge detection and image enhancement. In order to detect the façade area, several feature objects were be utilized. 3D information generated by two oblique images from similar view direction is the main index for finding façade in our research. Finally, performance assessment will be evaluated through visual interpretation of ground objects to find the optimal classification scheme.