Image analysis – Image transformation or preprocessing – Changing the image coordinates
Reexamination Certificate
2002-04-24
2004-10-26
Couso, Yon J. (Department: 2625)
Image analysis
Image transformation or preprocessing
Changing the image coordinates
C382S300000, C382S100000
Reexamination Certificate
active
06810153
ABSTRACT:
BACKGROUND OF THE INVENTION
a) Field of the Invention
This invention relates to an image processing method for correcting distortions in a ground image acquired from a satellite, and especially to a method for orthocorrecting distortions caused by the relief of a ground surface.
b) Description of the Related Art
Firstly, a description will be made about an image acquired by a satellite (satellite image) and the principle of its correction.
FIG. 6
is a simplified pictorial illustration of acquisition of an image by a satellite, and shows the satellite at numeral
1
and also depicts a ground surface
2
and a ground-surface scanning line
3
.
In
FIG. 6
, the satellite
1
acquires an image of the ground surface
2
by detecting electromagnetic waves from the ground surface
2
with a sensor (not illustrated) mounted on the satellite
1
such that the sensor is directed downwardly. When the sensor is a line sensor, such as CCD, commonly employed as a satellite sensor, the sensor is arranged such that cells (sensor elements) which make up the sensor are arrayed in a row extending at a right angle relative to a flying direction L of the satellite
1
. A range on the ground surface
2
, which is observed at the same time by the sensor, is the ground-surface scanning line
3
. The ground-surface scanning line
3
moves together with the satellite
1
in its flying direction L. As the satellite
1
moves, detection data by the individual cells are inputted in predetermined sampling cycles. The input timing of such detection data is the same for all the cells. The data over the entire ground-surface scanning line
3
are equivalent to a single line of satellite image. Accordingly, data equivalent to a single line of satellite image are acquired by a single input of data (single sampling). As the satellite
1
flies on an orbit, the ground surface
2
is continuously observed at constant swaths along scanning lines
3
as described above.
FIG. 7
is a view showing the principle of a correction of a satellite image obtained as described above.
FIG. 7
depicts an observed image
4
and a corrected image
5
.
In
FIG. 7
, the image acquired by the satellite as described above is transmitted as an observed image
4
to the ground. In the observed image
4
, configurational distortions have been developed by causes such as fluctuations in the satellite orbit and attitude, distortions in an optical system of the sensor, and terrain on the earth's surface. To make good use of the image acquired by the satellite, it is hence necessary to correct these distortions in the observed image
4
and to convert it into a corrected image in accordance with a predetermined cartographic projection method. This correction is to perform mapping from a pixel A(p,l) in the observed image
4
into a pixel A′(x,y) in the corresponding corrected image
5
. This mapping is called “mapping F”.
The observed image
4
is represented by a p-l coordinate system, in which the arrayed direction of the cells in the sensor on the satellite
1
(in other words, the longitudinal direction of the ground-surface scanning line
3
) is set as the p-coordinate axis while the traveling direction L of the satellite
1
is set as the l-coordinate axis. The corrected image
5
, on the other hand, is represented by an x-y coordinate system, in which a direction corresponding to the p-coordinate axis of the observed image
4
is set as the x-coordinate axis while a direction corresponding to the l-coordinate axis of the observed image
4
is set as the y-coordinate axis.
FIG. 8
is a flow chart illustrating processing for the correction of the observed image
4
.
In the corrected image
5
depicted in
FIG. 7
, pixels are arrayed and set at equal intervals. In the observed image
4
, on the other hand, the positions of pixels corresponding to the individual pixels in the corrected image
5
are not arrayed at equal intervals for such distortions as described above. The correction illustrated in
FIG. 8
is to find out, with respect to each pixel position in the corrected image
5
, a pixel at a corresponding pixel position in the observed image
4
and to place the pixel at the relevant pixel position in the corrected image
5
.
In
FIG. 8
, the correction consists of a distortion modeling step
100
and a resampling step
101
. In the distortion modeling step
100
, with respect to each pixel position in the corrected image
5
, a corresponding pixel position is determined in the observed image
4
. Referring back to
FIG. 7
, the position of the pixel A(p,l) in the observed image
4
, said position corresponding to the position of the pixel A′(x,y) in the corrected image
5
, is determined using a distortion model
6
prepared in advance. For this purpose, it is necessary to determine an inverse map F
−1
of the map F.
Here, the distortion model
6
is estimated, for example, from fluctuations in the orbit and attitude of the satellite, distortions in the optical system of the sensor and the like at the time of the capture of the observed image
4
, or is obtained from the results of an observation or the like by the satellite.
When the position of the pixel A(p,l) in the observed image
4
, said position corresponding to the position of the pixel A′(x,y) in the corrected image
5
, is determined in the distortion modeling step
100
, the routine then advances to the resampling step
101
. In this step, the pixel intensity at the position of the pixel A(p,l) is determined and is used as a pixel intensity at the pixel A′(x,y) in the corrected image
5
.
FIG. 9
is an illustration of the resampling step
101
in
FIG. 8
, and describes the resampling step
101
on the basis of the observed image
4
and corrected image
5
depicted in FIG.
7
. Incidentally, numeral
7
designates the center positions of pixels (hereinafter called “pixel positions”) in the observed image
4
.
In the distortion modeling step
100
described with reference to
FIG. 8
, the position of the pixel A(p,l) in the observed image
4
, said position corresponding to the position of the pixel A′(x,y) in the corrected image
5
, was determined. As shown in
FIG. 9
, however, the thus-determined position of the pixel A(p,l) generally does not register with a pixel position
7
in data observed by a sensor. It is, therefore, necessary to determine the intensity of the data observed at the position of the pixel A(p,l) by interpolation from the pixel intensity or intensities at one or more pixel positions
7
actually obtained around the position of the pixel A(p,l). As an interpolating method for this purpose, an interpolation formula such as nearest neighbor, bilinear or cubic convolution is commonly employed.
As a first conventional correction method of satellite-acquired image data, said method being based on the above-described principle, specifically as an orthocorrection method, there is known, for example, the method described in Demystification of IKONOS by Thierry Toutin and Philip Cheng, Earth Observation Magazine, 9(7), 17-21, 2000. According to this method, the distortion modeling step
100
in
FIG. 8
is represented by the following formula (1):
l
=
b
(
Nth
-
order
polynominal
in
x
,
y
and
z
)
a
(
Nth
-
order
polynominal
in
x
,
y
and
z
)
p
=
d
(
Nth
-
order
polynominal
in
x
,
y
and
z
)
c
(
Nth
-
order
polynominal
in
x
,
y
and
z
)
(
1
)
This formula is called “the rational polynominal method”. Whenever an image is acquired, the coefficients in the polynominals in x, y and z are calculated and determined by using ground control points GCP as will be described hereinaft
Hino Takashi
Kodaira Takatoshi
Komura Fuminobu
Ueda Kouji
Couso Yon J.
Crowell & Moring LLP
Hitachi Software Global Technology, Ltd.
LandOfFree
Method for orthocorrecting satellite-acquired image does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for orthocorrecting satellite-acquired image, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for orthocorrecting satellite-acquired image will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3277985