noproj - decouple from requiring a target body
The changes should allow noproj to work when there is no intersection with a target body, and allow other instruments supported by ISIS
Remove the need to have a target body in the image. Problems include full sky images, and partial/full disk images where reseaus are not on the target.
#2 Updated by Janet Barrett over 4 years ago
- Category set to Applications
I ran some tests on noproj. It gives close but not exact results for Viking. The warping is slightly different between Isis2 and Isis3. I also verified that anyone can copy the
noprojInstruments003.pvl file to their directory and add an instrument (such as Voyager). Even though the noproj program then runs successfully on a Voyager image, it does not give
valid results. Since Viking and Voyager both use the ReseauDistortionMap class, I would have assumed that the noproj program would work equally well for both missions. This does not
appear to be the case. I am continuing to investigate the noproj program to find out why it is not working for Voyager.
#3 Updated by Janet Barrett over 4 years ago
I have done more testing with noproj on Voyager data. It appears that noproj works on full frame images taken by Voyager. It was failing on an image containing the disk of Europa
surrounded by sky. I suspect that the reseau points that don't fall on the target are not being used to warp the data. This would explain why full frame Voyager images work successfully
in noproj. I will be adding Voyager support to noproj through the noprojInstruments004.pvl file in $ISIS3DATA/base/applications.
#4 Updated by Janet Barrett over 4 years ago
After looking through the code that applies the transformation for the noproj program, I have discovered that the reseaus are only being used by noproj to create the
undistorted match cube that is then used by cam2cam to do the transformation. The cam2cam transforms the input file to match the geometry of the match cube, but it
only uses data that falls on the target to do the transformation. This is done by taking a line,sample in the output file and obtaining the lat,lon for that point.
The lat,lon is then projected into the input file to get the line,sample to transform to. If the lat,lon doesn't intersect the target in the input file, then the
point isn't used in the transformation. This results in the transformation only being done on those points that intersect the target in the input file. This is a
problem with limb images because the SPICE usually isn't accurate enough to fully intersect the target because it thinks the target is offset from where it actually lies.
ISIS2 uses all of the reseau positions when transforming Viking and Voyager data, regardless of whether they fall on the target or not. ISIS2 noproj does not do any
mapping to ground coordinates when generating an undistorted image. As a result, ISIS2 noproj works on partial frame images.
The noproj program needs to be decoupled from the cam2cam program so that it can do transformations using only the distorted/undistorted focal plane coordinate values and
not using any ground coordinates.
#5 Updated by Tammy Becker about 4 years ago
- Subject changed from noproj - generalize to support multiple instruments to noproj - generalize to support multiple instruments and decouple from requiring a target body
- Priority changed from Normal to High
Modified the source from Internal to External based on the recent ISIS-Support HiRISE ticket reported by Sarah Mattson:
Also modified the subject line to reflect the necessary change to not require a target body in order to correct camera distortion.
#28 Updated by Stuart Sides almost 3 years ago
- Subject changed from noproj - generalize to support multiple instruments and decouple from requiring a target body to noproj - decouple from requiring a target body
- Description updated (diff)
This issue contained two unrelated problems. Subject and Description updated to have only one. The relationship to #2090 was left in place because some notes are for both problems.