You are using an outdated browser that does not fully support the intranda viewer.
As a result, some pages may not be displayed correctly.

We recommend you use one of the following browsers:

Full text

Stilla, Uwe

In: Stilla U, Rottensteiner F, Paparoditis N (Eds) CMRT09. IAPRS, Vol. XXXVIII, Part 3/W4 — Paris, France, 3-4 September, 2009
Andreas Schmitt, Birgit Wessel, Achim Roth
German Aerospace Center (DLR)
German Remote Sensing Data Center (DFD), D-82234 Wessling
Andreas.Schmitt@dIr.de, Birgit. Wessel@dlr.de, Achim.Roth@dlr.de
KEY WORDS: SAR, Imagery, Structure, Extraction, Change Detection, Method, Urban
In this paper we present an alternative method for SAR image denoising, structure enhancement, and change detection based on the
curvelet transform. Curvelets can be denoted as a two dimensional further development of the well-known wavelets. The original image
is decomposed into linear ridge-like structures, that appear in different scales (longer or shorter structures), directions (orientation of the
structure) and locations. The influence of these single components on the original image is weighted by the corresponding coefficients.
By means of these coefficients one has direct access to the linear structures present in the image. To suppress noise in a given SAR
image weak structures indicated by low coefficients can be suppressed by setting the corresponding coefficients to zero. To enhance
structures only coefficients in the scale of interest are preserved and all others are set to zero. Two same-sized images assumed even
a change detection can be done in the curvelet coefficient domain. The curvelet coefficients of both images are differentiated and
manipulated in order to enhance strong and to suppress small scale (pixel-wise) changes. After the inverse curvelet transform the
resulting image contains only those structures, that have been chosen via the coefficient manipulation. Our approach is applied to
TerraSAR-X High Resolution Spotlight images of the city of Munich. The curvelet transform turns out to be a powerful tool for image
enhancement in fine-structured areas, whereas it fails in originally homogeneous areas like grassland. In the change detection context
this method is very sensitive towards changes in structures instead of single pixel or large area changes. Therefore, for purely urban
structures or construction sites this method provides excellent and robust results. While this approach runs without any interaction of
an operator, the interpretation of the detected changes requires still much knowledge about the underlying objects.
Nowadays spaceborne SAR data is easily available. Thanks to
the high resolution of up to one meter (TerraSAR-X) it is suitable
for urban applications, e.g. urban growth modeling as well as for
damage mapping in conjunction with (natural) disasters. A main
problem for SAR image interpretation apart from the geometri
cal aspect is the high noise level caused by the combination of
deterministic (speckle effect) and random noise. The reduction
of noise, e.g. by the multi-looking approach, often goes along
with a loss of resolution. While structure preserving filters do
not enhance fine-structured areas, smoothening filters even blur
the structures apparent in SAR data over urban areas. So reso
lution and structure preserving filter algorithms are still a topic
of research. In this context alternative image representations like
wavelets have been applied. While wavelets are used to separate
point singularities (Candes and Donoho, 1999), second genera
tion wavelets, e.g. curvelets, are more suitable for the extraction
of two dimensional features, as they are able to describe image
discontinuities along a smooth line (an edge) with a minimum
number of coefficients (Candes and Donoho, 1999). The ele
mentary components are the so-called ridgelets - due to their
appearance like a ridge - that can have different scales (equiv
alent to their length), directions and positions in the image. This
enables a selection of two dimensional features to be suppressed
(assumed noise) or to be emphasized (structure) by manipulating
the corresponding coefficient of each ridgelet. In the following a
short overview to related work especially to the development of
curvelets is given. Then, the curvelet representation is roughly
explained and three applications are presented: image denoising,
structure enhancement and change detection over the city center
of Munich (imaged by TerraSAR-X in the high resolution spot
light mode and VV polarization). So this paper shows the poten
tial of the curvelet transform for SAR image analysis.
The curvelet transform used in this approach has originally been
developed by (Candes and Donoho, 1999) to describe an object
with edges with a minimal number of coefficients in the contin
uous space. Much research work was done to examine the be
haviour of curvelets (Candes and Donoho, 2002a, Candes and
Demanet, 2002b, Candes and Guo, 2002), to transfer the def
initions from the continuous to the discrete space (Candes and
Donoho, 2003a, Candes and Donoho, 2003b) and to accelerate
the computing time (Candes et al., 2005) so that digital image
processing becomes feasible. Many applications in different sci
entific fields have been published so far, e.g. in geo- and as
trophysics, that are summarized on the curvelet homepage (De
manet, 2007).
Denoising of SAR images to simplify image analysis has also
been a research topic during the last years where many approaches
have been published. (Ali et ah, 2007) proposed a combination of
a wavelet based multi-scale representation and some filters to im
prove the results obtained by the ’’standard” filtering techniques
like the Lee-filter. A bayesian-based method using ”a trous” filter
in the wavelet domain has been proposed by (Moghaddam et ah,
2004). Because of the properties of the wavelet transform, orig
inally developed for one dimensional data, these two methods
are able to smooth regions and to suppress point-like noise, but
they do not take into account the two dimensional nature of im
ages. The advantage of second generation wavelets for despeck-
ling has been examined by (Gleich et ah, 2008) for the bandelet
and the contourlet transform. The application of curvelets on op
tical and ultrasound images respectively in the medical context
has been published by (Ma et ah, 2007). The only publication on
the use of curvelets in the remote sensing context by (Sveinsson
and Benediktsson, 2007) presents a denoising technique with a