#4322 closed defect (fixed)
Tile size in NITF header needs to match tile size in JPEG2000 codestream
| Reported by: | jeepingben | Owned by: | warmerdam |
|---|---|---|---|
| Priority: | normal | Milestone: | 2.2.0 |
| Component: | default | Version: | svn-trunk |
| Severity: | normal | Keywords: | |
| Cc: |
Description
When writing a NTIF using JPEG2000 compression, the tile size stored in the NITF header (NPBBH and NPPBV) need to match the tile size used in the JPEG2000 codestream (passed using :
BLOCKXSIZE and BLOCKYSIZE in JP2KAK driver TILE_WIDTH and TILE_HEIGHT for JP2ECW driver <tilesize> %d %d </tilesize> for JP2MrSID driver BLOCKXSIZE and BLOCKYSIZE for JP2OpenJPEG driver)
The default sizes are large (at least in the JP2KAK driver), X is 20000 and y is the image height. NBPPH only has 4 digits, so it can't exceed 9999 (I saw 8192 as a limit in the NITF documentation), so the codestream tiling should be adjusted rather than adjusting NPPBH and NPPBV.
Change History (5)
comment:1 by , 12 years ago
comment:2 by , 12 years ago
http://isotc.iso.org/livelink/livelink/fetch/2000/2122/327993/327973/654328/6208440/documents/24n3111%20(BPJ2K0110).pdf page 41 says NPPBH and NPPBV must be 0 if the image is larger than 8192 and is not tiled.
http://www.gwg.nga.mil/ntb/baseline/docs/stdi0006/FSTDI-0006-NCDRD-15Sep05.doc page 2-30 This isn't exactly a NITF spec, but describes matching tile size to jpeg block size for NITF commercial products.
It sounds like at a minimum, NPPB* should be set to 0 for large untiled images. Tiling speeds up random access and chipping operations. Large images (like geoeye) are difficult to work with when they are untiled.
comment:5 by , 7 years ago
| Milestone: | → 2.2.0 |
|---|
What is the reference for the fact that the tile size stored in the NITF header (NPBBH and NPPBV) need to match the tile size used in the JPEG2000 codestream ?