 |
 |
 |
 |
 |
 |
 |
 |
Activiities of ASTER Working Groups |
|
 |
|
 |
 |
|
 |
 |
Overview
Straylight effect is unwanted light entering into the sensor by ghosting or scattering.
Since the ghost problem occurs due to optical and electrical leaks in the observation system
and is inevitable in optical remote sensing images,
a simple method is required for the operational usage of its correction in ground data processing.
Straylight in a satellite sensor is usually investigated through ground testing and sometimes investigated on orbit,
and may be corrected in operational processing.
Any optical sensor will exhibit some straylight effect and many satellite sensors have been examined for straylight effects.
In many case, straylight is reported in the reflective region (<3µm) which relevant to ASTER VNIR and SWIR.
There are same effect in thermal infrared (ASTER TIR) region which caused by emission from an outside source or the instrument itself.
In the case of the TIR bands of ASTER,
straylight testing on the ground was conducted by viewing a heated panel through a hole bored in a water-cooled panel,
and the results showed that straylight effects were at most less than 0.4% of the input radiance.
However, straylight characterization for TIR sensors is generally less accurate than for reflective bands (VNIR, SWIR),
mainly because zero-radiation targets cannot be prepared for laboratory testing unlike the case for reflective sensors,
though ray-tracking methods can be used for TIR sensors as well as for reflective sensors.
Thus, straylight effects on TIR sensors are not always well-known in comparison with reflective sensors.
|
|
Cause of the problem
The SWIR bands of ASTER suffers from spectral crosstalk phenomena that diffuse incident light from band 4 to other bands
by reflection from the metal film at the back of each detector.
The diffused image of band 4 is subtracted from the images of the other bands, and the crosstalk ghost is corrected.
However, the image of band 4 also suffers from stray light that originates from its focal plane.
The radiometric sensitivity is determined during ground calibration using an integrating sphere,
but the on-axis stray light influences the detected incident light;
thus, stray light must be correctly estimated from on-orbit images for precise radiometric calibration.
Lunar observation is one of the most powerful tools for tackling this problem.
This is because conditions of the Earth's atmosphere and sea influence the estimation of
the modulation transfer function performed from bridges and the seashore.
Lunar images do not suffer from these problems; thus, the amount of stray light can be more correctly estimated.
Fig. 1(a) shows an image of the moon acquired by band 3N of VNIR, where 3N designates a nadir viewing band 3 sensor.
The lunar image was obtained by maneuvering the Terra spacecraft with a constant pitch rate,
where the image over sampling rate was approximately 4.5 times that of ground sampling;
this accounts for the elongation in the image.
The image size was reduced by 4.5 times along the track direction to be the same as that of the image obtained at the normal sampling rate,
which was effective in increasing the dynamic range of the signal.
The radiometric coefficient was applied using the radiometric database at the observation time.
The linearly stretched image shows the three components of the stray light pattern, as shown in Fig. 1(b).
One is slightly displaced relative to the original image and is evident at the very top of the lunar image.
Two off-axis ghost images are evident in the bottom lunar image, one of which is diffused and the other less obscure.
Since the VNIR optics includes the dichroic filter as a spectral separation device, the internal reflection is considered
to be the cause of the off-axis stray light.
|
|
 |
Fig. 1. Lunar images of band 3N of VNIR
(a) Original (b) linearly stretched
|
|
|
Fig. 2(a) shows the band 4 of SWIR image of the moon.
The image was processed in the same manner as that of the band 3N image.
The linearly stretched image shows the on-axis stray light pattern, as shown in Fig. 2(b).
The stray light pattern extends around the moon, showing that only an on-axis component exists.
The stray light originates from an aluminum film under the detector,
which is employed to double the sensitivity of the PtSi-CCD detector.
The detectors are arranged in the cross-track direction, and thus the stray light component is larger in the horizontal direction.
|
|
 |
Fig. 2. Lunar images of band 4 of SWIR
(a) Original (b) linearly stretched
|
|
|
Fig. 3 shows the bands 10-14 of TIR which were cropped around the Moon, masked for the Moon, and linearly enhanced.
In all bands, ghosting caused by strong input radiation from the lunar surface appears clearly around the Moon,
while its strength and spread depends on band.
The horizontal stripes around the upper side of the Moon were caused by a detector-dependence on straylight.
|
|
 |
| Fig. 3. Lunar images of TIR |
|
|
Correction method
The straylight correction algorithms have been developed using the lunar images.
The observed image is a linear combination of the original image true radiance to be estimated
and the ghost image resulting from the stray light in the imaging system, considering also additive noise.
The ghost image is expressed by a convolution of the original image with the transfer function.
The amplitudes and widths of the stray light were estimated using the correction algorithm that attempts
to eliminate the ghost pattern in the deep-space region.
For more information, please read the following article.
|
|
Reference
Inflight Straylight Analysis for ASTER Thermal Infrared Bands
Tonooka, H.
Pages: 2752 - 2762
IEEE Volume:43 Issue:12 Date:Dec.2005
(The abstract is free of charge)
Correction of stray light and filter scratch blurring for ASTER imagery
Iwasaki, A.; Oyama, E.
Pages: 2763- 2768
IEEE Volume:43 Issue:12 Date:Dec.2005
(The abstract is free of charge)
|
|
 |
|
