Logo GISdevelopment.net

GISdevelopment > Proceedings > ACRS > 1999


1989 | 1990 | 1991 | 1992 | 1994 | 1995 | 1996 | 1997 | 1998 | 1999 | 2000 | 2002
Sessions

Agriculture/Soil

Water Resources

Disasters

Measurement and Modeling

Land Use

Forest Resources

Mapping from Space

Oceanography/Coastal Zone

Topics Including Education

Hyper Spectral Image Processing

Image Processing

Geology

Environment

GIS

Global Change

Airborne Remote Sensing

Poster Sessions
  • Session 1
  • Session 2
  • Session 3
  • Session 4
  • Session 5
  • Session 6



  • ACRS 1999


    Poster Session 2

    Printer Friendly Format

    Page 1 of 3
    | Next |

    Correction of OCTS Sensor Alignment

    Toshiaki Hashimoto
    Center for Environmental Remote Sensing (CEReS), Chiba University
    1-33, Yayoi, Inage-Ku Chiba 263, JAPAN
    TEL:(81)-43-290-3945 FAX:(81)-43-290-3857
    hashi@ceres.cr.chiba-u.ac.jp

    Keywords: Sensor alignment, Collinearity condition, ADEOS/OCTS

    Abstract:
    The alignments of ADEOS/OCTS were measured before the launch. But it was found out that the values of the alignments were changed after the launch, which resulted in the serious amount of geometric distortion. The alignments were corrected on the basis of photogrammetry using OCTS imagery and GCPs.

    1. Introduction
    The ADEOS(Advanced Earth Observing Satellite) was launched in August 1996 by NASDA (National Space Development Agency of Japan). The OCTS(Ocean Color and Temperature Scanner) on board the ADEOS spacecraft is an optical mechanical scanner with 12 channels and the spatial resolution of 680 m at nadir. It observed the Earth surface only in daytime. The geometric accuracy of the OCTS at initial check out was about 10 km on the ground. It was too terrible for some applications which needed high geometric accuracy like mosaicking, image composite, etc. The NASDA and some organization examined the reasons for such a terrible accuracy and ascertained three kinds of factors; 1) the bug of software for calculating satellite position, 2) the inadequacy of the algorithm for satellite attitude, 3) the change of alignments after the launch. In those factors, the bug for satellite position was fitted easily. The algorithm for satellite attitude was installed on the onboard computer. The raw attitude data were processed by the algorithm and transmitted to the ground. So the attitude data could be no longer corrected. But the only raw attitude data for first orbit each day in GMT were downlinked for housekeeping. Such raw data were processed with the new algorithm adopted for ADEOS-2/GLI (Global Imager). The subject of this paper is the correction of the alignments utilizing OCTS imagery on the basis of photogrammetry.

    2. Scan geometry of OCTS
    Figure 1 shows the optical pass from the focal plane to the ground. The orbital coordinates are introduced which is right handed where the flight and zenith direction correspond to x and z axis respectively. The OCTS has the tilt function by inclining the scanning mirror around y-axis (NASDA,1996). The vector a from the center of the focal plane to the scanning mirror is expressed as (1,0,0). If the tilt angle (t) is equal to zero, the normal vector of mirror 0will be ( cos(45° +e ), 0, sin(45° +e ) ), where e is the miss-inclination of mirror and equal to 14 milli-radians. When the scanning mirror is tilted by t and the scan angle is q , the normal vector of the mirror 1 is expressed as follows.


    If the alignments of the scanning mirror ( a1, b1, l1) are considered, the normal vector of the mirror is expressed as =Q11.

    The view vector reflected from the mirror is expressed as 1=-2( ,)•. If the alignments of the sensor itself to the platform (a2, b2, g2) are considered, the view vector will be re-written as =Q21. In these equations, both Q1 and Q2 are the rotation matrices around three axes. The view vector is expressed by the components as follows.





    Figure 1 Scan geometry of OCTS (q=0)

    3. Alignments to be determined
    The effect of each alignment on the view vector is examined. The view vector without any alignment is expressed as 1=(1-2hx2, -2hxhy,-2hxhz. If each alignment exist individually, the view vector is expressed as follows.


    The differences of view vectors are expressed in Figure 2 by the error vectors on the image where each alignment has 0.001 radian independently. The calculation was performed on the equatorial region. Both the formula and the errors on the image proved that the a1.and a2 have the same effect on the image. Eventually the parameters (b1, g1) and (a2, b2, g2) are selected as the alignments to be determined and the value of a1 is fixed to the re-launch value in this work.



    Figure 2 Errors on image raised by alignment error

    Page 1 of 3
    | Next |

    Applications | Technology | Policy | History | News | Tenders | Events | Interviews | Career | Companies | Country Pages | Books | Publications | Education | Glossary | Tutorials | Downloads | Site Map | Subscribe | GIS@development Magazine | Updates | Guest Book