Gebäudetypisierung basierend auf UltraCamX Daten für die Bestimmung der Attraktivitätsmerkmale der Wohnstandorte

Abstract

This thesis examines the possibilities of an automatized object-based image analysis (OBIA) to classify urban buildings by means of high resolution areal images taken by the UltraCam X system. It was analyzed whether without any flanking data from other sources valid building properties and structures including their heights can be extracted by these studies. For this purpose two target areas in Berlin were chosen, representing a high and typical variability of urban buildings. The discussion of the extracted results includes a detailed analysis of building areas and building heights. The building heights were related to the number of building floors and a classification of nine building classes of different heights was performed. One essential step for this process was the availability of a normalized digital surface model (nDSM). On the basis of this model it is possible to derive information about the object’s height. The high resolution of the areal date enabled to retrieve finer geometric structures, like balconies in a limited way. Special problems that occurred during in the framework of segmentation and classification are discussed and assessed in detail. An accuracy assessment reveals in very satisfactory results indicating the valuable extraction of geometric building parameters. Furthermore, it has been shown that important neighboring characteristics like vegetation and close distances to water bodies could be extracted. They include elevated, ground and Roof greening provinces. In the second part of this thesis an analysis of these parameters in respect to their evaluability for an attractiveness assessment have been investigated. In this case too, no reference data of residential location attractiveness were available and neither should been used for the analysis. Applicable attributes of attractiveness were derived only basing on an extensive literature research of generic relationships between measurable building properties and properties of their close environment. They include relations between building densities (amount of building development), volume of building development, homogeneity of top level building height structure and others. Within this Research area there are not many publications to refer on. Therefore, the automated object-based image analysis of areal images only to extract features of building attractiveness of urban residential locations is an innovative field of research and application. Thus, a particular value of this work is a successful review of general interrelations of the extracted building properties and of their close environment against the background of attractiveness parameters. Properties like building densities, volume and area of buildings, mean building heights within building cells have been investigated in one of the study areas in respect of their spatial variations. For this purpose the study area was divided into sub units of traffic cells enabling to analyze the fine structure of ccurring variations of the extracted properties. By simultaneous use of derived environmental features (vegetation, water) the character of attractiveness features and variations of them have been extracted an analyzed. These results were subjected to a critical review as well. This approach is promising for the notified implementation of the urban building’s attractiveness analysis, for the assessment of the attractiveness of urban residential locations, and for intermodal urban traffic concepts. The outlook of resulting future prospects shows that with inclusion of additional analysis variables further potentials for attractiveness studies can be opened. The value added by this method consists of both, the possibility of a contemporary large area data acquisition at high spatial resolution and the automated supply of the information required. Wherever these issues are the primary driver of application, this approach permits a highly valid basis for the analyses of urban residential locations, their attractiveness and their transport links.