Linking Spatial Video and GIS

Spatial Video is any form of geographically referenced videographic data. The forms in which it is acquired, stored and used vary enormously; as does the standard of accuracy in the spatial data and the quality of the video footage. This research deals with a specific form of Spatial Video where these data have been captured from a moving road-network survey vehicle. The spatial data are GPS sentences while the video orientation is approximately orthogonal and coincident with the direction of travel. GIS that use these data are usually bespoke standalone systems or third party extensions to existing platforms. They specialise in using the video as a visual enhancement with limited spatial functionality and interoperability. While enormous amounts of these data exist, they do not have a generalised, cross-platform spatial data structure that is suitable for use within a GIS. The objectives of this research have been to define, develop and implement a novel Spatial Video data structure and demonstrate how this can achieve a spatial approach to the study of video. This data structure is called a Viewpoint and represents the capture location and geographical extent of each video frame. It is generalised to represent any form or format of Spatial Video. It is shown how a Viewpoint improves on existing data structure methodologies and how it can be theoretically defined in 3D space. A 2D implementation is then developed where Viewpoints are constructed from the spatial and camera parameters of each survey in the study area. A number of problems are defined and solutions provided towards the implementation of a post-processing system to calculate, index and store each video frame Viewpoint in a centralised spatial database. From this spatial database a number of geospatial analysis approaches are demonstrated that represent novel ways of using and studying Spatial Video based on the Viewpoint data structure. Also, a unique application is developed where the Viewpoints are used as a spatial control to dynamically access and play video in a location aware system. While video has been to date largely ignored as a GIS spatial data source; it is shown through this novel Viewpoint implementation and the geospatial analysis demonstrations that this need not be the case anymore