An investigation into automated processes for generating focus maps

The use of geographic information for mobile applications such as wayfinding has increased rapidly, enabling users to view information on their current position in relation to the neighbouring environment. This is due to the ubiquity of small devices like mobile phones, coupled with location finding devices utilising global positioning system. However, such applications are still not attractive to users because of the difficulties in viewing and identifying the details of the immediate surroundings that help users to follow directions along a route. This results from a lack of presentation techniques to highlight the salient features (such as landmarks) among other unique features. Another problem is that since such applications do not provide any eye-catching distinction between information about the region of interest along the route and the background information, users are not tempted to focus and engage with wayfinding applications. Although several approaches have previously been attempted to solve these deficiencies by developing focus maps, such applications still need to be improved in order to provide users with a visually appealing presentation of information to assist them in wayfinding. The primary goal of this research is to investigate the processes involved in generating a visual representation that allows key features in an area of interest to stand out from the background in focus maps for wayfinding users. In order to achieve this, the automated processes in four key areas - spatial data structuring, spatial data enrichment, automatic map generalization and spatial data mining - have been thoroughly investigated by testing existing algorithms and tools. Having identified the gaps that need to be filled in these processes, the research has developed new algorithms and tools in each area through thorough testing and validation. Thus, a new triangulation data structure is developed to retrieve the adjacency relationship between polygon features required for data enrichment and automatic map generalization. Further, a new hierarchical clustering algorithm is developed to group polygon features under data enrichment required in the automatic generalization process. In addition, two generalization algorithms for polygon merging are developed for generating a generalized background for focus maps, and finally a decision tree algorithm - C4.5 - is customised for deriving salient features,
including the development of a new framework to validate derived landmark saliency in order to improve the representation of focus maps.