IEEE Geoscience and Remote Sensing Magazine - March 2020 - 27

MARCH 2020

IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE

Perpendicular Baseline (m)

Perpendicular Baseline (m)

established to transform phase unwrapping into an optimiPSI, PSP, IPTA, SqueeSAR, JSInSAR, StaMPS, and STUN.
zation problem.
The multireference mode is mainly aimed at solving the
The 1D (time) and 2D (azimuth and range) phase unproblem of too few interferograms with single-reference
wrapping represent two kinds of phase unwrapping methmode-based interferograms generation. Obviously, usods used in time-series InSAR techniques, each with its
ing the same number of images, more interferograms can
own benefits and drawbacks. The former can correctly unbe generated under the multireference than under the
wrap phases in the discontinuous region but requires more
single-reference mode. Figure 8 shows the baseline coninterferograms, and the phase must satisfy the established
figuration of single-reference and multireference modes.
model; the latter can perform unwrapping using fewer
One of the key factors to consider when generating interinterferograms, and the processing is more flexible after
ferograms using the single-reference mode is the selection
phase unwrapping. However, an unwrapping error occurs
of an appropriate main image. The spatial and temporal
in the jump region, which does not satisfy the spatial conbaselines as well as the Doppler frequency of the main
tinuity assumption.
image should be as close as possible to the center of all
The common shortcoming of these two methods is that
images to minimize total decorrelation. With the multirefthey do not make full use of the 3D characteristics of sigerence mode, the length of the temporal and spatial basenals [56]; therefore, the phase unwrapping methods in 3D
lines, such as SBAS, is generally restricted. The techniques
that use the multireference mode
include SBAS, temporarily coherent
point (TCP) InSAR (TCP-InSAR)
Im
Im
[52], the coherent pixels technique
(CPT) [53], and stable point network
(SPN) [54]. Even the CPT method
provides good deformation estimaRe
Re
tion from a reduced set of only seven
SAR images.
(a)
(b)
(c)
We also can classify time-series
InSAR techniques according to the
Im
Im
manner of phase unwrapping. Phase
unwrapping is a problem related to
restoring phase 2r ambiguity, which
is nondeterministic and requires adRe
Re
ditional assumptions [55]. The standard PSI method establishes the second differential phase model of the
(d)
(e)
(f)
PSs. Phase unwrapping and parameter estimation are completed by solvFIGURE 7. The black triangles represent scatterers, the wavy lines represent the reflected
ing the model; then, the deformation
signal, and their size reflects the strength of the signal. (a) A PS. (b) A PS signal in a complex
map of the whole scene is obtained
plane. (c) The main direction of the PS signal. (d) A DS. (e) A DS signal in a complex plane.
using spatial integration. The stan(f) The main direction of DS signal.
dard PSI method assumes that the
phase satisfies the phase model and
can be regarded as 1D phase unwrap200
200
ping in time. With the SBAS method,
each interferogram is unwrapped
100
100
using a 2D phase-unwrapping algorithm in space (azimuth and range)
0
0
first, such as the minimum cost
f low (MCF) algorithm [27]; then,
-100
-100
the singular value decomposition
(SVD) operation is performed on
-200
-200
the unwrapped phase to obtain the
-400 -200
0
200
400
-400 -200
0
200
400
Temporal Baseline (Days)
Temporal Baseline (Days)
deformation signal. The 2D phase
(a)
(b)
unwrapping method requires that
the phase satisfies the spatial continuity assumption. Then, L0-norm, FIGURE 8. The baseline configuration of (a) a single-reference mode and (b) a multi--
L1-norm, or L2-norm models are reference mode.
27
IEEE Geoscience and Remote Sensing Magazine - March 2020

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