The automatic classification of urban open space by a pattern-matching method of the viewshed at intersections

Abstract

This research focuses on the automatic classification of small urban fragments through a morphological analysis of cognitivist inspiration. The recognition algorithm is performed on observer-centric forms, constructed through the use of visibility assessment techniques over a series of individual points of view. These tools are (1) the isovist for its capacity to delineate and synthesize the visual properties of the immediate viewshed from a point, and (2) the automatic construction of a typology of intersection patterns. The aim is to assimilate the forms of the theoretical intersection patterns to those extracted from the isovist field generated by a group of strategically placed points. Three different matching methods are proposed, and the significance of the parameters needed for optimal calibration is discussed.