Mittwoch, 30. Juni 2010

ProMeRe an der GEOBIA

An der dritten international Konferenz zur "Geographic Object-based Image Analysis" vom 29. Juni - 2. Juli 2010 in Ghent, Belgien, war Natalie Lack mit dem Beitrag "Object-based change detection for a cultural-historical survey of the landscape - from cow trails to walking paths" vertreten. Die Tagung bringt Wissenschaftler und Anwender aus der ganzen Welt zusammen, um neue Entwicklungen im Bereich der geographischen objektbasierten Bildanalyse zu präsentieren und diskutieren.

Historical maps, aerial images and other historic documents are finding their way into the digital world. This offers new possibilities but also requires adapted technologies and interdisciplinary approaches. The ProMeRe project connects historic, architectural and planning methods with remote sensing and geoinformation technologies. It aims to analyse the development of tourism and its effects in an alpine environment over the last 150 years.
We present an object-based change detection approach with IMAGINE Objective for the monitoring of country roads and walking paths networks in the project area. Changes in these networks contribute essential information for historic research questions concerning the development of tourism and its impact on the landscape. IMAGINE Objective employs feature models which work on objects produced by image segmentation and various other pixel-based algorithms which, after being vectorised, can be processed by geometric and textural parameters. This process is used and optimised for the extraction of the walking path network from high resolution colour ortho imagery of 2000 and 2004.
The results show that the similarity of scree and rocks with walking paths is a major challenge. In comparison of different feature extraction models “centerline detection" yields the best results for walking path network detection. Preliminary change detection tests using a semi automated post-classification approach show reliable results. For this project, the applied methods prove to be useful to detect changes in the walking path network in the example imagery even without producing a complete extraction of all walking paths and country roads. The approach will be extended to the whole project area and additional mostly older black/white sets of ortho imagery.

Keine Kommentare:

Kommentar veröffentlichen