Options
2008
Conference Paper
Titel
Feature extraction and change detection for bridges over water in airborne and spaceborne SAR image data
Abstract
Key elements of man-made infrastructure are bridges. In case of natural disasters, it is important to get real-time information about these objects. Such up-to-date information-requirements can be fulfilled by Synthetic Aperture Radar sensors. The main advantage of SAR is the availability of data under nearly all weather conditions and at any day time. Especially in the case of bridges over water, the SAR specific side looking imaging geometry can lead to special characteristics in the image. A bridge can appear as several bright stripes in the SAR image if certain prerequisites are fulfilled. These stripes can be segmented, and some bridge features like width and height can be derived. In this paper, the possibilities to extract features like width and height from the mentioned stripes are discussed. An approach is presented to segment these stripes in SAR or InSAR data and to exploit this special signature for change detection. The investigations are supported by simulations based on a ray tracing approach. Aim of the simulations is an assessment of acceptable imaging constellations for a certain bridge. Here acceptable means constellations, where the special signature can be expected. Real data examples for high-resolution airborne (resolution better than 40 cm) and spaceborne (RADARSAT-1, approx. 9 m resolution) sensors are presented. The results show that a feature like bridge height can successfully be derived from SAR data by using a description of the bridge on an object level. A concept is proposed to use this description for change detection by integration in a GIS-system.