A NOVEL FUSION METHOD OF SAR AND OPTICAL SENSORS TO RECONSTRUCT 3-D
BUILDINGS
Huan Zhang, Hua-ping XU, Member IEEE, Zhe-feng Wu
School of Electronics & Information Engineering, Beihang University, Beijing, China
ABSTRACT
This paper investigates the method for 3-D buildings
reconstruction using the fusion of SAR and optical sensors.
First, the joint positioning equation is described as the
foundation of this method. Then, the principal steps of this
method are recommended. Image registration is employed to
obtain a mapping function between the two images as the
first step. And the 3-D coordinates are calculated with the
joint positioning equation. Afterwards, the 3-D buildings are
reconstructed using the 3-D coordinates obtained above.
Finally, an experiment is conducted to show the
reconstruction result of this method.
Index Terms— 3-D building, reconstruction, fusion,
synthetic aperture radar, optical image
1. INTRODUCTION
Three-dimensional (3-D) buildings reconstruction is an
important task, which has applications in different areas such
as urban planning, cartographic drawing, environment
protection, emergency response and so on. Many methods
are developed to reconstruct 3-D buildings. One kind is
based on geometric or electromagnetic models [1-2].
Another kind is based on image fusion, such as multi-aspects
synthetic aperture radar (SAR) images[3], interferometric
SAR (InSAR) [4], InSAR and optical image fusion[5] or
SAR and optical image fusion [6].
The methods based on geometric, electromagnetic or InSAR
models always have more requirements for images
compared with the methods based on image fusion, which is
the reason why the latter have a better performance in
applicability. The method proposed in [6] uses the optical
image to detect buildings, but does not employ the
information provided in optical image to calculate the
coordinates of the buildings. Reference [7] indicates that
fusion of SAR and optical images has a better positioning
precision than fusion of SAR images.
In this paper, a new fusion method of SAR and optical
image is proposed to reconstruct 3-D buildings based on the
joint positioning equation presented in [7]. This new method
utilizes the simultaneous equations of the angle equations
from optical image and range equation from SAR image to
calculate the 3-D coordinates of the building. And then the
coordinates are employed to reconstruct the building. In the
experiment, a TerraSAR-X data and a GeoEye-1 data are
used to reconstruct the Chinese National Swimming Center.
2. DESCRIPTION OF THE JOINT POSITIONING
EQUATION
Reference [7] presented the joint positioning equation with
SAR and optical images. For a target with the
position
, the angle equations from the optical
radar image are expressed as
p opt
p opt
tan cos
xx
zz
(1)
Where
is the position of the optical sensor,
and
are respectively the squint angle and side angle of
the target.
From SAR image, the slant range equation of the target is:
2 2 2 2
p SAR p SAR p SAR
( ) ( ) ( )x x y y z z R
(2)
Where
is the position of the SAR
sensor,
is the slant range between the target and SAR
sensor.
With the equations (1) and (2), the 3-D coordinates of a
target can be calculated out. Then the location and height of
the target will be obtained by fusing SAR and optical images.
3. DESCRIPTION OF THE PROPOSED METHOD
Principal steps of the proposed 3-D buildings reconstruction
are showed in Fig.1. First, registrate the SAR and optical
image and obtain a mapping function between the two
images. Then, take advantage of the joint positioning
equation to calculate the building’s 3-D coordinates. Finally,
reconstruct the 3-D building with the coordinates of the
building.
609978-1-4799-7929-5/15/$31.00 ©2015 IEEE IGARSS 2015