Identifying divergent building structures using fuzzy clustering of isovist features

Abstract

Nowadays indoor navigation and the understanding of indoor maps and floor plans are becoming increasingly important fields of research and application. This paper introduces clustering of floor plan areas of buildings according to different characteristics. These characteristics consist of computed human perception of space, namely isovist features. Based on the calculated isovist features of floorplans we can show the possible existence of greatly varying alternative routes inside and around buildings. These routes are archetypes, since they are products of archetypal analysis, a fuzzy clustering method that allows the identification of observations with extreme values. Besides archetypal routes in a building we derive floor plan area archetypes. This has the intention of gaining more knowledge on how parts of selected indoor environments are perceived by humans. Finally, our approach helps to find a connection between subjective human perceptions and defined functional spaces in indoor environments.