Abstract
Synthetic aperture radar (SAR) multi-temporal techniques have been
proposed to improve image resolution, persistent-scatterer detection,
and damage-map generation (Bujor et al., 2004; Hooper, 2008; Yun et al.,
2012). The usage of multiple scenes diminishes the speckle noise
commonly present in a SAR image and increases the overall
signal-to-noise ratio (SNR) of dominant radar scatterers. Following the
same idea, we propose a new SAR processing workflow based on the
back-projection algorithm (Cafforio et al., 1991). We employ the
back-projection method’s flexibility to focus the radar pulses at
multiple depths shifted with respect to the reference digital elevation
model (DEM), yielding a volume of SLCs discretely sampled over a range
of elevations/depths. These scenes are then combined to produce a
single-look complex (SLC) image, which presents increased SNR and
effective resolution. Similar image improvements can be achieved by
spatially averaging a SLC image generated at high spatial sampling.
However, the proposed method requires a fraction of the processing time
of the fine-spatial SLC formation. The proposed approach retains the
same benefits of any back-projection algorithm and enables the formation
of SLC images corresponding to large Earth surface extents.