The Euler Character: A New Type of Visual Landscape Metric?

Landscape-preference theories such as prospect-refuge theory and Kaplan and Kaplan's landscape-preference matrix and theories of visual perception propose that the physical structure of the landscape has a direct psychological effect on people due to evolved sensitivity to particular defining characteristics. Efforts to identify consistent quantitative relationships between metrics of these characteristics and human preference have had some success. However, the field has also faced some criticisms due to low explanatory power in the results reducing confidence that relationships found can be applied to other contexts. In this paper we argue that dependence on generalised planar maps for the derivation of the metrics but on viewpoint-specific perspective photographs for the preference data is a potential cause of low explanatory power. How viewpoint change may affect scene characteristics needs to be better understood if representative viewpoints are to be chosen to allow results which are general to an area, in particular the difference between discreet (topological) changes and continuous changes. This paper presents the results of an experiment to test whether the topological complexity of a view, as measured by the Euler character of the horizon graph, has perceptual significance. We investigate if images with higher horizon-graph complexity were considered more interesting than those with lower graph complexity via a forced-choice Internet survey.