Abstract
The presence of locations that possess distinct spatial-cognitive features (salient landmarks) is a fundamental necessity for supporting navigation. Embedding formal or structural variability sufficient to create such landmark locations is therefore an important consideration in the design of large urban and architectural spaces. Despite the availability of diverse theories that seek to identify the characteristics of ‘a salient landmark’, relatively few experimental techniques are available to empirically evaluate saliency in a given architecture plan. This study is therefore motivated by the development of an ability to measure spatial distinctiveness during the architectural design and modelling process. The information from such an analysis can prove useful for evaluating the way in which a design provides support for wayfinding and spatial appeal. Statistical summaries obtained from the three-dimensional (3D) isovists are compared using principal component analysis to differentiate monotonous regions from the more structurally distinct ones. The experiments reported in the paper demonstrate novel utilization of the isovist concept to capture spatial properties and comparison of structural saliency among two well-known architectural designs. Central contributions of the paper include the novel experimentation technique of capturing and utilizing 3D isovists, its interpretation and the quantitative methodology behind saliency computation.
Acknowledgements
This project is supported by ARC DP1092679: ‘Modelling and predicting patterns of pedestrian movement: using robotics and machine learning to improve the design of urban space’. The 3D models used for the test were obtained from the Google Sketchup Warehouse (Villa Savoye by Keka; Dana-Thomas House by Joe Sweeney).