A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Automatic building extraction is a difficult object recognition problem due to a high complexity of the scene content and the object representation. There is a dilemma to select appropriate building models to be reconstructed; the models have to be generic in order to represent a variety of building shape, whereas they also have to be specific to differentiate buildings from other objects in the scene. Therefore, a scientific challenge of building extraction lies in constructing a framework for modelling building objects with appropriate balance between generic and specific models. This thesis investigates a synergy of IKONOS satellite imagery and airborne LIDAR data, which have recently emerged as powerful remote sensing tools, and aims to develop an automatic system, which delineates building outlines with more complex shape, but by less use of geometric constraints. The method described in this thesis is a two step procedure: building detection and building description. A method of automatic building detection that can separate individual buildings from surrounding features is presented. The process is realized in a hierarchical strategy, where terrain, trees, and building objects are sequentially detected. Major research efforts are made on the development of a LIDAR filtering technique, which automatically detects terrain surfaces from a cloud of 3D laser points. The thesis also proposes a method of building description to automatically reconstruct building boundaries. A building object is generally represented as a mosaic of convex polygons. The first stage is to generate polygonal cues by a recursive intersection of both datadriven and model-driven linear features extracted from IKONOS imagery and LIDAR data. The second stage is to collect relevant polygons comprising the building object and to merge them for reconstructing the building outlines. The developed LIDAR filter was tested in a range of different landforms, and showed good results to meet most of the requirements of DTM generation and building ...
Automatic building extraction is a difficult object recognition problem due to a high complexity of the scene content and the object representation. There is a dilemma to select appropriate building models to be reconstructed; the models have to be generic in order to represent a variety of building shape, whereas they also have to be specific to differentiate buildings from other objects in the scene. Therefore, a scientific challenge of building extraction lies in constructing a framework for modelling building objects with appropriate balance between generic and specific models. This thesis investigates a synergy of IKONOS satellite imagery and airborne LIDAR data, which have recently emerged as powerful remote sensing tools, and aims to develop an automatic system, which delineates building outlines with more complex shape, but by less use of geometric constraints. The method described in this thesis is a two step procedure: building detection and building description. A method of automatic building detection that can separate individual buildings from surrounding features is presented. The process is realized in a hierarchical strategy, where terrain, trees, and building objects are sequentially detected. Major research efforts are made on the development of a LIDAR filtering technique, which automatically detects terrain surfaces from a cloud of 3D laser points. The thesis also proposes a method of building description to automatically reconstruct building boundaries. A building object is generally represented as a mosaic of convex polygons. The first stage is to generate polygonal cues by a recursive intersection of both datadriven and model-driven linear features extracted from IKONOS imagery and LIDAR data. The second stage is to collect relevant polygons comprising the building object and to merge them for reconstructing the building outlines. The developed LIDAR filter was tested in a range of different landforms, and showed good results to meet most of the requirements of DTM generation and building ...
Extraction of buildings from high-resolution satellite data and airborne Lidar
Sohn, G-H (author)
2004-01-01
Doctoral thesis, University of London.
Theses
Electronic Resource
English
Data fusion of high-resolution satellite imagery and LiDAR data for automatic building extraction
| Online Contents | 2007
| British Library Conference Proceedings | 2020
Modeling of urban wind ventilation using high resolution airborne LiDAR data
| Online Contents | 2017