Enhanced quick and dirty estimate for SAR coherence
Andrea Monti Guarnieri(1) and Davide D'Aria(2)
Politecnico di Milano,
piazza Leonardo da Vinci, 32,
(2) Aresys s.r.l. - a POLIMI spin off company, via Garofalo, 39, 20133, Italy
"Quick and dirty" coherence estimators are necessary when the computational speed is the key factor, like for browsing products in processing facilities or data mining.
The algorithm here shown exploits the amplitudes of coregistered SAR images, hence it does not require the computationally expensive step of local frequency estimate, the most demanding step in coherence estimate.
At the same time, a DEM, like an STRM one, is exploited to enhance the quality by means of space-adaptive spectral shift filtering. In the paper we show an implementation of such estimator and the computational cost based on a standard multi-processor Linux machine.
Finally we show some examples and compare, in terms of quality and computing time, the enhanced quick and dirty estimator and a conventional, complex coherence estimate implemented by exploiting the SRTM DEM with no further refinement of the local frequency estimate.