IEEE Geoscience and Remote Sensing Magazine - June 2020 - 24

TABLE 1. A SUMMARY OF THE MAIN DIFFERENCES IN TomoSAR TECHNIQUES, WITH REFERENCES TO SOME ILLUSTRATIVE
WORKS IN THE LITERATURE.
REFERENCE
FEATURES

[22] [29] [23] [35] [72] [71] [83] [84] [66] [96] [93] [25] [98] [43] [41] [90] [86] [91] [88] [89] [99] [70]

What is in input
Covariance matrix
Vector of SLC values at a
given pixel

ü
ü

ü
ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

Stack of SLC images

ü

Polarimetric information

ü

ü
ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

What is reconstructed
Spatial distribution of
the reflectivity

ü

ü

Point cloud (zero to three
scatterers per resolution
cell)

ü

Power distribution

ü

ü

ü

ü
ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

Surface

ü

ü

ü

ü

Phase model
Dependency with height

ü

ü

ü

ü

Dependency with thermal
dilation

ü

ü

ü

ü

ü
ü

ü

ü

ü

ü

ü

ü

ü

ü

Dependency with
subsidence motion

ü

ü

ü

ü

ü

ü

ü

ü

ü

Regularization
Postprocessing

ü

Indirect spatial smoothing
when estimating covariance

ü

Sparsity: limited number
of scatterers

ü

ü

Sparsity: L1 norm minimum

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

Spatial regularity

ü

ü

ü

Spatial prior (surface
localization)

ü

Scatterer selection

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

Pros
Height superresolution
Dense reconstruction

ü
ü

ü

ü
ü

ü

ü

ü

ü

ü

ü

Complete reconstruction

ü

Outlier rejection based
on PFA

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü

ü
ü

ü

ü

ü

ü

ü

Cons
Requires parameter tuning

ü

Requires a priori
information

ü

ü

ü

ü

ü

ü
ü

smooth surfaces and vegetation in urban areas are often
difficult to retrieve. Richer models that combine or select different models depending on the scattering mechanisms could be considered to better address this difficulty and provide a better understanding of urban scenes.
The use of polarimetric information in tomographic
inversion is probably a way to improve reconstructions
in these configurations.
Extensions and combination of SAR tomography with
other acquisition modes, such as staring spotlight [110],
24

ü

ü

ü

ü

ü

ü

ü
ü

circular SAR [111]-[113], and multiple-input/multipleoutput SAR [114], is also an active research front. In particular, due to limited temporal decorrelation and cancellation of the atmospheric path delays during interferometric
processing, bistatic (when one satellite transmits and both
receive) and multistatic (e.g., Tandem-X [106] and TandemL 1 [107]) systems or MirrorSAR [108] will facilitate tomographic processing, allowing bistatic interferograms with
better phase quality compared to conventional repeat-pass
methods. In this way, temporal decorrelation, atmosphere
IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE

JUNE 2020
IEEE Geoscience and Remote Sensing Magazine - June 2020

Table of Contents for the Digital Edition of IEEE Geoscience and Remote Sensing Magazine - June 2020
