Table 1 The AVNIR capabilities 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
Multispectral Panchromatic 
1. 0.42 to 0.60 
Observation band 2. 0.52 to 0.60 0.52 to 0.69 
3. 0.61 to 0.69 
4. 0.76 to 0.89 
AVNIR channel IFOV 20 4 rad (16m) 10 rad (8m) 
FOV 5.7 ° (80km) 
S/N above 200 above 90 
MTF above 0.25 above 0.20 
Quantization bit 8bit 7bit 
Observation band 0.61 to 0.69 
Image 
Navigation 
channel S/N above 45 
MTF above 0.12 
FOV direction 0.54 dgrees behind the AVNIR channel 
  
  
PRINCIPLE OF OPERATIONS 
The basic configuration of the IFOV for the image navigation 
channel as well as the main channels installed in AVNIR of 
ADEOS are illustrated in Fig. 1. Main channels are set at the focal 
plane of the AVNIR optics system. The image navigation channel 
is a CCD array with the same parameters as the main channel, but 
the IFOV is slightly separated from the main channel. Due to the 
data transfer restriction, only a part of the data stream of the image 
navigation channel is sent back to the ground station to provide 
two image stripes for this image navigation channel. Since the 
main channel CCD and the image navigation channel CCD are 
mechanically set on the same focal plane unit, motion of the IFOV 
is the same for each of them, but the field of view differs slightly 
on the ground. By using these image data sets, we can find a stream 
of corresponding points of a partial of image an image navigation 
channel on the image of the main channel. Since we have the focal 
plane parameters, we can calculate the ideal corresponding point 
of the image navigation channel data on the image of the main 
channel if there is no attitude change. The difference between the 
actual corresponding point and the ideal corresponding point is 
caused by the attitude change between main channel data 
acquisition and image navigation channel data acquisition. By 
integrating the derivative position knowledge, we can extract the 
precise attitude motion of AVNIR. 
Figure 2 shows the idea of the above mentioned corresponding 
point trace of the image navigation channel on the main images. 
If, as the simplest assumption shown in Fig 2(2), the attitude change 
is due only to the roll angle, the point corresponding to the ideal 
point where no attitude change exists, is always offset in the cross 
track direction. In this case, the time difference between the two 
channels for the same ground target is constant, so that of attitude 
changes can be derived easily. 
In general, motion change consists of cross track and along track 
direction changes. In this case, we must separate the difference 
into three components roll, pitch and yaw, as shown in Fig 2(b). 
For an along track of corresponding point, theses are differences 
in attitude as well as time. Time -attitude domain mapping is used 
to obtain attitude change in the same time interval. Figure 3 shows 
the idea, the difference in corresponding points is mapped along 
the time-motion line. By this mapping method, any attitude 
difference a mean time difference can be calculated by interpolation 
to derive extract attitude, as be mentioned below. The relative 
attitude of IFOV is calculated as follows using roll, pitch and yaw 
attitude difference for the mean time difference of main and image 
navigation channel data acquisition of the same ground point,. 
If the target IFOV motion function is expressed by a function of 
time t as f(z), the attitude change is expressed as, 
d(t)= f(t +Ts)- f(t) e 
where Ts is the mean time difference. 
By taking the Fourier spectrum of the equation (1), 
D(@) = F(@)exp(-j@Tp) - F(@) (2) 
Thus 
i D(@) 
F(œ) = (T=exp(-/@Tp)) (3) 
188 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B1. Vienna 1996