PS-DInSAR Deformation Velocity Estimation by the
Compressive Sensing
Jingting Li
Sino-French Engineering School
Beihang University
Beijing, China
lijingting@buaa.edu.cn
Huaping Xu
School of Electronic and Information Engineering
Beihang University
Beijing, China
xuhuaping@buaa.edu.cn
Abstract—PS-DInSAR has been a tool for detecting surface
micro-deformation. However, the technique is constrained by the
quantity of SAR images which should be more than 30.
Compressive Sensing (CS) is a new method of signal processing
and allows recovering signal stably with fewer measurements.
The paper applied CS to PS-DInSAR after analyzing the sparsity
of data and proposed a novel method to estimate the deformation
velocity with a high accuracy by using fewer SAR images. Our
method will reduce the redundant data. A scene with a cone-
shaped peak is designed to generate SAR images. Simulation
results are presented to validate our method.
Keywords—PS-DInSAR; Compressive Sensing; Deformation
I. INTRODUCTION
Differential interferometric synthetic aperture radar
(DInSAR) is the technique to estimate the surface deformation,
and has a remarkable achievement in the term of the
geodynamic. The accuracy has reached the centimeter to
millimeter level. The Permanent Scatterers (PS) DInSAR can
estimate the surface micro-deformation by eliminating the
influence caused by time decorrelation and the atmospheric
phase. But this technique has been largely restricted by the data
acquisition.
To get the reliable image of fault’s movement, enough
number of SAR images (>30) [1] should be available for the
PS-DInSAR technique. However, spaceborne SAR images are
usually obtained by repeat orbits, which will take a long time
(e.g. five to six years) [1] and will be unable to estimate the
deformation in time. As a consequence, it is necessary to look
for an approach which can estimate the surface deformation
highly accurately with fewer SAR images.
We proposed that every SAR image is a sample for the
estimation of deformation velocity, as a result, the problem of
reducing the number of SAR images has turned to a problem of
reducing the number of sample.
The Nyquist/Shannon sampling theory requires that a signal
must be sampled at a rate twice its highest frequency. In the
recent years, Compressive Sensing (CS) [2], which is a new
signal acquisition method, has been proposed. Signals can be
reconstructed stably from far fewer measurements based on the
sparsity of signal in a severely underdetermined linear system.
This young field has inspired an explosion of research in a
wide range of topics including theory, recovery algorithms, and
its applications, such as Compressive Imaging, Medical
Imaging, Analog-to-Information Conversion, Geophysical Data
Analysis, Compressive Radar, Communication, etc.
Compressive Sensing has been used in 4D SAR [3] to solve the
problem of non-uniform distribution of the acquisition
geometry. The method can estimate the received signal that
includes the information of deformation velocity.
The paper proposes a novel PS-DInSAR method based on
the theory CS after analyzing the sparsity of data obtained by
PS-DInSAR. The new method allows estimating the surface
deformation with fewer SAR images, which means it can
reduce the number of measurement and redundancy of data.
The organization of this paper is as following. Section Τ
presents the approach of PS-DInSAR. Section Υ shows the
method in the combination with the theory CS. The simulation
results and the comparison with PS-DInSAR are presented in
section Φ. Section Χ makes a conclusion and presents our
future work.
II. PS-DINSAR
METHOD
PS-DInSAR [4] chooses the points which have a stable
amplitude behaviors over long time intervals as PS points from
a temporal series SAR images. The phase information can be
obtained from these sparsely distributed PS candidates, and
then reliable deformation velocity measurements can be
estimated.
Firstly, one of the N SAR images is considered as the
reference “master”. And then N-1 interferometric SAR images
are generated by the interferometry of the slave images with
the master image.
The differential interferometric phase can be represented in
the following formula [4]:
߮
ௗ
ሺ
ݔ
ǡ݇
ሻ
ൌ
ܥ
ݒǡ݇
ȉݒ
ሺ
ݔ
݅
ሻ
ܥ
ܳǡ݇
ȉߝ
ሺ
ݔ
݅
ሻ
ݒ
ܰܮ
ሺ
ݔ
݅
ǡ݇
ሻ
ܽ
ሺ
ݔ
݅
ǡ݇
ሻ
ܾሺݔ
݅
ǡ݇ሻ
(2.1)
where
978-1-4799-8633-0/15/$31.00 ©2015 IEEE