Visual saliency as an aid to updating digital maps

摘要

Visual attention is considered in the context of a professional computer-based task, using aerial photography for updating topographic mapping data (photogrammetry). There is potential for using visual attention models to help develop various semi-automated ‘attention-aware’ support systems for this task, and these are discussed. An experimental study is described which examined the potential influence of expertise, image type and exposure duration on the role of visual saliency or salience (as calculated by Itti, L., & Koch, C. (2000). A saliency-based search mechanism for overt and covert shifts of visual attention. Vision Research, 40, 1389–1506 saliency maps) in the distribution of visual attention with such imagery. Using a non-intrusive, low-resolution and low-cost method to determine the approximate distribution of visual attention, effects of expertise and landscape type were found. Unexpectedly, saliency appeared to be more relevant to visual attention among expert users than novices, and potential reasons for this are explored. Implications and further research plans are discussed.