Automated Recognition of Morphological Patterns in Urban Open Spaces

Climate-urban form research tends to identify vernacular (traditional) forms as a successful combination of architecture, urban design and climate. This theory, which draws a parallel between climate and adaptive urban form, is supported by the accepted common belief that harsh climates, such as hot-arid ones, confront those who live in them with unique challenges. These challenges are answered by solutions that are continuously improved over generations by trial and error and therefore evolve into the “proper” response to the climatic conditions. This paper is part of an ongoing doctoral research, which investigates the validity of this theory. Its aim is to study whether a systematic influence of climate on urban form development can be found, focusing on the design of urban open spaces. Towards this goal a quantitative methodology is adopted and a computer-based model of an automated recognition system is under development. This system consists of three main elements: an image processing element for recognizing urban open spaces, a pattern recognition element (developed in a GIS environment) for extracting and clustering morphological characteristics and a statistical element for determining the significance of patterns and evaluating their relation to climatic parameters. The present paper introduces the methodology, which was developed to enable a substantial and statistically anchored evaluation of urban geometry in relation to its climatic context. It presents the first part of this methodology: the object recognition model, focusing on the definition of classes and class descriptors and the unique problems in developing a recognition system for urban open spaces. ∗ Corresponding author.