A model for generating object-based change information from multi-temporal remotely-sensed imagery

As world populations increasingly are clustered in urban areas, so there is a tangible need for accurate mapping of these regions by national mapping agencies. A consequential impact of growing cities is that greater numbers of people across the globe are vulnerable to the effects of natural disasters or anthropogenic catastrophes. Tools such as remote sensing have been widely used by researchers to monitor urban areas for applications such as land use and land cover changes and population distribution to name a few. Air- and space-borne sensors with fine spatio-temporal resolutions have facilitated these analyses, offering an effective and efficient data source for multi-temporal analysis of urban areas.<br/><br/>Alongside the increased data availability from remote sensors is a demand for efficient algorithms for interpretation of these images. This thesis describes the development of a conceptual framework for the iterative processing of fine spatial resolution optical images. It consists of two central components, object detection and object comparison.<br/><br/>In the object detection phase, buildings are identified in the image and extracted as objects stored in a scene model. Object attributes describing the location, geometric, spectral and textural characteristics of each object are stored in a database, allowing the on-demand display as vector or raster entities. The thesis implements the model through exemplars for the detection of circular and cylindrical features on several remote sensing and simulated datasets.<br/><br/>The object comparison phase allows automated change information to be generated describing per-object and intra-object brightness variability over time, hence, allowing change to be quantified for each detected feature. These descriptors facilitate the manual use of qualitative scales for damage assessment. A detailed discussion is presented on the merit of the conceptual model, its limitations and describes how future expansion of the model to full implementation could be achieved.

Name dc:type.qualificationname

Ph.D.

Level dc:type.qualificationlevel

doctoral

Grantor dc:publisher.institution

University of Southampton

Year dc:date.issued

2009

Author dc:creator

• Bevington, John S.

Advisor dc:contributor.advisor

• Lewis, H.G.