[gdal-dev] [PROPOSAL] Add CADRG write support via RPFTOC/NITF drivers
Brad Hards
bradh at frogmouth.net
Thu Jun 12 03:39:42 PDT 2025
On Thursday, 12 June 2025 6:47:43 PM AEST Patrik Sylve via gdal-dev wrote:
> Thank you all for the inputs!
> Our use cases are also within defense applications, so the format is indeed
> still in active use.
I also have some interest in the format. Its important in aviation since it
has fast, and bounded, decode time. That makes it suitable for high assurance
systems like moving maps on primary flight displays.
> Frank, good point. I'll definitely keep that in mind. Some scripting
> approach sounds like a reasonable approach if it turns out to be too
> tedious to get full support working directly in the driver.
I think the TOC part will probably need to be in GDAL. Note that there are
really two formats in one - the RPF part and the NITF part, and they
physically overlap in the file. Its a messy format in that respect. Against
that, the tiles are always the same size, and the offsets are therefore
essentially constant.
> As for VQ-compression, yes, this will be an interesting challenge. I came
> across this paper referenced in the CADRG standard, looks like promising
> reading https://apps.dtic.mil/sti/tr/pdf/ADA283396.pdf.
I assume you already have MIL-STD-2411, -2411-1 etc, plus the VQ standard.
See https://nsgreg.nga.mil/ntb.jsp if not.
I would also suggest the Southard paper - https://apps.dtic.mil/sti/tr/pdf/
ADA250707.pdf - which is referenced in Nancy Markuson's analysis paper.
There is a recommendation for unsharp filtering. I think that is a good idea,
but should only be applied to CADRG, not to CIB...
It would also be worth considering options like constant colour palette. There
is a limited (216) number of colours in the LUT, but if you do each image tile
separately, the seamlines can be really obvious.
The VQ part would probably be best in a separate library.
Brad
More information about the gdal-dev
mailing list