Building footprint extraction by fusing dual-aspect SAR images

The new spaceborne high resolution synthetic aperture radar (SAR) sensors on board the TerraSAR-X satellites can achieve spatial resolution up to 1m. In high resolution SAR image, signature of urban buildings can be identified and extracted. Due to the side-looking character of SAR, building usually presents as L-shaped bright linear feature caused by the two sides that facing the radar sensor, while echoes from the other two sides´ lost. It´s clear that multi aspect image will be helpful for this problem. This paper deals with building footprint extraction from dual-aspect high-resolution SAR images acquired from ascending and descending orbits. Firstly, the dimension and position of building is estimated from single aspect image separately by several combined methodologies. Then through fusing estimates from each aspect image, more complete footprint of buildings can be obtained. Field measurements proved that building footprint extracted from dual-aspect SAR images is more reliable than that from single aspect SAR image.