 |
| Titel |
Photometric Lambert Correction for Global Mosaicking of HRSC Data |
| VerfasserIn |
Sebastian Walter, Greg Michael, Stephan van Gasselt, Thomas Kneissl |
| Konferenz |
EGU General Assembly 2015
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 17 (2015) |
| Datensatznummer |
250111622
|
| Publikation (Nr.) |
 EGU/EGU2015-11760.pdf |
|
|
|
|
|
| Zusammenfassung |
| The High Resolution Stereo Camera (HRSC) is a push-broom image sensor onboard Mars
Express recording the Martian surface in 3D and color. Being in orbit since 2004, the camera
has obtained over 3,600 panchromatic image sequences covering about 70% of the
planet’s surface at 10-20 m/pixel. The composition of an homogenous global mosaic
is a major challenge due to the strong elliptical and highly irregular orbit of the
spacecraft, which often results in large variations of illumination and atmospheric
conditions between individual images. For the purpose of a global mosaic in the full
Nadir resolution of 12.5 m per pixel we present a first-order systematic photometric
correction for the individual image sequences based on a Lambertian reflection
model.
During the radiometric calibration of the HRSC data, values for the reflectance scaling
factor and the reflectance offset are added to the individual image labels. These parameters
can be used for a linear transformation from the original DN values into spectral reflectance
values. The spectral reflectance varies with the solar incidence angle, topography (changing
the local incidence angle and therefore adding an exta geometry factor for each
ground pixel), the bi-directional reflectance distribution function (BRDF) of the
surface, and atmospheric effects. Mosaicking the spectral values together as images
sometimes shows large brightness differences. One major contributor to the brightness
differences between two images is the differing solar geometry due to the varying
time of day when the individual images were obtained. This variation causes two
images of the same or adjacent areas to have different image brightnesses. As a
first-order correction for the varying illumination conditions and resulting brightness
variations, the images are corrected for the solar incidence angle by assuming an ideal
diffusely reflecting behaviour of the surface. This correction requires the calculation of
the solar geometry for each image pixel by an image-to-ground function. For the
calculations we are using the VICAR framework and the SPICE library. Under the
Lambertian assumption, the reflectance diminishment resulting from an inclined Sun
angle can be corrected by dividing the measured reflectance by the cosine of the
illumination angle. After rectification of the corrected images, the individual images are
mosaicked together. The overall visual impression shows a much better integration of the
individual image sequences. The correction resolves the direct correlation between the
reflectance and the incidence angles from the data. It does not account for topographic,
atmospheric or BRDF influences to the measurements. Since the main purpose of the
global HRSC image mosaic is the application for geomorphologic studies with a
good visual impression of the albedo variations and the topography, the remaining
distortions at the image seams can be equalized by non-reversible image matching
techniques. |
|
|
|
|
|
|