Talks and Poster Presentations (with Proceedings-Entry):
"Urban Data Collection: An Automated Approach in Remote Sensing";
Talk: 24th Urban Data Management Symposium, Information Systems and the Delivery of Societal Benefits,
- 2004-10-29; in: "Proceedings of UDMS'04",
This paper discusses an automated method for creating and updating Digital City Models. The general idea is to use only a pair of panchromatic stereo scenes as input data. This data can either be spaceborne (IKONOS, Quickbird) or airborne imageries. The proposed approach can be divided into three major parts. The first step comprises the finding of potential building areas in the images. This is done by looking for radiometrically homogenous regions in the images that exceed a certain elevation threshold. This way unwanted information like e.g. vegetation and cars is disregarded. Once the location of a potential building is found, the next step is to extract its geometric properties (i.e. building corners). Here an adaptive region-growing algorithm was developed for marking the building outline and afterwards a step-by-step Houghtransformation is employed in order to gather the wanted edge and corner information. To each building one height value is assigned, hence horizontal roofs are assumed. The final step is the comparison of the old state of urban areas (in form of a digital cadastre map or a GIS) with the newly derived Digital City Model representing the current state. In this comparison new buildings are detected, changes of buildings are recognized, as well as areas where old buildings were demolished. The level of detail in all these considerations depends on the resolution of the used input imagery.
Online library catalogue of the TU Vienna:
Electronic version of the publication:
Created from the Publication Database of the Vienna University of Technology.