Use this URL to cite or link to this record in EThOS:
Title: Constrained modelling of building interiors
Author: Rosser, Julian
ISNI:       0000 0004 6351 3727
Awarding Body: University of Nottingham
Current Institution: University of Nottingham
Date of Award: 2016
Availability of Full Text:
Access from EThOS:
Geometric models of indoor environments are useful for many applications. They provide a crucial foundation for assessing, modifying and communicating the internal layout of buildings. Despite the importance of this information and the increasing availability of sensors and methods for data capture, generating accurate as-built models of residential building interiors remains a challenging task. Users require accurate dimensions of spaces, yet few digital devices are available to consumers capable of delivering precise measurement. These individuals are not professionals with access to high-quality laser tape measures and are not necessarily willing to exhaustively measure internal walls, manually assemble and edit geometry to create an accurate and topologically correct plan. However, nowadays many individuals possess smartphones and tablets which contain a range of different sensors, albeit subject to various imperfections and error. Furthermore, buildings often exhibit predictable regularities in their construction which limits the likely structure of indoor spaces. This thesis considers the use of soft and hard constraints for improving building measurements to form the “best” representation of interior geometry. It investigates using prior knowledge of the building within an optimisation model to make automated adjustments to the geometry. The model exploits high accuracy, external geospatial data in the form of the building’s footprint, such as can be found in large-scale topographic mapping. In addition, knowledge of the topology of the space, assumptions regarding room shape and building construction and information learned from other interior layouts is utilised. All methods presented in this thesis are aimed at providing a semi-automatic, end-to-end system for indoor modelling. Quantitative evaluations of the improvements in positional accuracy arising from using the proposed methods are reported. These methods are evaluated and discussed in the context of CityGML modelling which provides an open and interoperable standard for defining 3D building information
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available