Along track displacement/time line
x
a Tl IA
Lidl bolo Loo TEEN
ja [of 1 4 n1! LL RE | LE SEE pus
hu 44 b 4 I ol es als Fle dd J£ AI, Ld eh
Al track i / / d / / / 1 , ! A = ! / / 4 fr 1 / / ! / / / / / /
Mong Matching point for equal time in image navigation channel data; / / / /
displacement Bib! 1/172 1 L1 173 dy!) NEL TW NU
fee ete is frre) / Pelt fl
fens fron
| 1 sh n lh db ied!
L5 [54 HN 7l
EA
Figure 3 Plot of along track displacement of matching point.
Table 2 The error of estimated attitude to the total data acquisition time
Ratio of the average time offsets
to total data acquisition time
The residual components
(radian)
1.9% 1.5X 10°
3.9% 6.5x 10$
7.8% 2.95X 10?
altitude of 800km. To extract attitude information using image
navigation channel data, the extracting must be about 25 times
the time delay.
EXTRACTION OF THE ATTITUDE DIFFERENCE
AVNIR data is not presently available. To analyze the function of
image navigation, we simulated a AVNIR image using existing
satellite image and tried to extract attitude information from the
simulated image. Figure 5 shows the simulated images of the main
channel and image navigation channel. The elevation influences
in the acquired image of the satellite does not change attitude.
However, parallax due to elevation is introduced in the image if
the satellite attitude changes. The simulated image Field of View
(FOV) changed with the attitude change of the satellite and the
parallax due to elevation.
We will now describe how to generate the simulation image using
DEM. Figure 6, X is the image acquired with a flat surface, Y is
the image with parallax due to elevation, and the satellite is position
R. The satellite observes point y in the image Y instead of point x
in the image X due to the parallax introduced by elevation.
International Archives of Photogrammetry and
Therefore, we simulated the image from a SPOT image and DEM.
Simulation Conditions
1. The satellite altitude H is 800 km.
2. The satellite velocity V is 7 km/s.
3. The attitide change is only in the roll direction.
4. The FOV of the Image Navigation channel is 0.054
degrees behind the FOV of the AVNIR channel.
5. Pixel size is 10 m (same as SPOT)
6. DEM (Geographical Survey Institute publication)
The main channel simulated image is a 700 line x 700 pixel
area around Mt. Fuji in Japan, and the image navigation channel
simulated image is a 700 lines x 100 pixel area.
We have shown that the attitude variance of main channel can
be extracted by means of image navigation. In image navigation,
it is important to extract the precise information about the
difference of attitude change between main and image navigation
channels. We examined the accuracy of attitude change derived
190
Remote Sensing. Vol. XXXI, Part B1. Vienna 1996