Conventional techniques approach phase unwrapping (PU) as an optimisation problem, where figures of merit like the total branch-cut length, the number of cuts, etc. are to be minimised. They disregard the properties of the field to be unwrapped: the topography, i.e. the DEM, projected in the SAR reference. The purpose of the paper is to fill this gap by providing statistics of the 'fringe maps'. The Woodward theorem is exploited to link the interferogram power spectrum density (PSD) with the probability density function (PDF) of the phase gradient that would result in a likely topography. A parametric model of the expected, unwrapped PG PDF is then derived by exploiting the fractal properties of topography. Its parameters can be accurately estimated given the wrapped PG. This model provides useful statistical information for phase unwrapping. It is then possible, for example, to estimate the number of residuals, to find the best azimuth presuming factor and to find the optimal interferogram range demodulation. Finally, the author exploits the second-order statistics of the PG field (as a fractal) to derive a suitable approximation for the expected length of the branch cuts.
Using topography statistics to help phase unwrapping
MONTI-GUARNIERI, ANDREA VIRGILIO
2003-01-01
Abstract
Conventional techniques approach phase unwrapping (PU) as an optimisation problem, where figures of merit like the total branch-cut length, the number of cuts, etc. are to be minimised. They disregard the properties of the field to be unwrapped: the topography, i.e. the DEM, projected in the SAR reference. The purpose of the paper is to fill this gap by providing statistics of the 'fringe maps'. The Woodward theorem is exploited to link the interferogram power spectrum density (PSD) with the probability density function (PDF) of the phase gradient that would result in a likely topography. A parametric model of the expected, unwrapped PG PDF is then derived by exploiting the fractal properties of topography. Its parameters can be accurately estimated given the wrapped PG. This model provides useful statistical information for phase unwrapping. It is then possible, for example, to estimate the number of residuals, to find the best azimuth presuming factor and to find the optimal interferogram range demodulation. Finally, the author exploits the second-order statistics of the PG field (as a fractal) to derive a suitable approximation for the expected length of the branch cuts.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
