Use this URL to cite or link to this record in EThOS:
Title: Automated 3D modelling of buildings from aerial and space imagery using image understanding techniques
Author: Kim, Taejung
ISNI:       0000 0001 3495 525X
Awarding Body: University of London
Current Institution: University College London (University of London)
Date of Award: 1996
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
The development of a fully automated mapping system is one of the fundamental goals in photogrammetry and remote sensing. As an approach towards this goal, this thesis describes the work carried out in the automated 3D modelling of buildings in urban scenes. The whole work is divided into three parts: the development of an automated height extraction system, the development of an automated building detection system, and the combination of these two systems. After an analysis of the key problems of urban-area imagery for stereo matching, buildings were found to create isolated regions and blunders. From these findings, an automated building height extraction system was developed. This stereoscopic system is based on a pyramidal (area-based) matching algorithm with automatic seed points and a tile-based control strategy. To remove possible blunders and extract buildings from other background objects, a series of "smart" operations using linear elements from buildings were also applied. A new monoscopic building detection system was developed based on a graph constructed from extracted lines and their relations. After extracting lines from a single image using low-level image processing techniques, line relations are searched for and a graph constructed. By finding closed loops in the graph, building hypotheses are generated. These are then merged and verified using shadow analysis and perspective geometry. After verification, each building hypothesis indicates either a building or a part of a building. By combining results from these two systems, 3D building roofs can be modelled automatically. The modelling is performed using height information obtained from the height extraction system and interpolation boundaries obtained from the building detection system. Other fusion techniques and the potential improvements due to these are also discussed. Quantitative analysis was performed for each algorithm presented in this thesis and the results support the newly developed algorithms' effectiveness and robustness.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available
Keywords: Photogrammetry; Urban scenes