rgdal, multiband CCITT compression
Sorry, the word "expected" was a bad choice. The output is what I get. I'm highlighting two tables that should be the same, but aren't. I appreciate you taking the time to poke around with me. *** Correct Stats: img_1bit_2band band1 band2 0 3086 8353 1 6914 1647 Compression: CCITTRLE from img_1bit_2band_CCITTRLE.tif band1 band2 0 6131 8783 1 3869 1217
On Thu, Nov 4, 2010 at 5:14 PM, Michael Sumner <mdsumner at gmail.com> wrote:
Ok, thanks. I can try later today with Manifold System, which may be. I'm not sure of other non-GDAL tools to double check TIFFs with. (ArcGIS 10 uses GDAL but I don't know to what extent). Is the output above what you expect, or what you get? That's what I get. Sorry to be dim, but I don't see where you've specified what's actually wrong (and I haven't explored enough to see if that's easy to determine from the code). That's a?cunning use of formals() to?change defaults for a function?- a bit radical for my taste, but I like it. Cheers, Mike. On Fri, Nov 5, 2010 at 11:00 AM, Oliver Soong <osoong+r at gmail.com> wrote:
Here's an example:
require(rgdal)
formals(paste)$sep <- ""
SP27GTIF <- readGDAL(
?system.file("pictures/SP27GTIF.TIF", package = "rgdal")[1],
?output.dim=c(100,100)
)
img_1bit_1band <- SP27GTIF
img_1bit_1band at data$band1 <- as.integer(SP27GTIF at data$band1 > 85)
img_1bit_2band <- SP27GTIF
img_1bit_2band at data <- data.frame(
?band1 = img_1bit_1band at data$band1,
?band2 = as.integer(SP27GTIF at data$band1 > 170)
)
for(img in c("img_1bit_1band", "img_1bit_2band")) {
?cat(paste("*** Correct Stats: ", img, "\n"))
?print(sapply(get(img)@data, table))
?for(compress in c("NONE", "DEFLATE", "LZW", "CCITTRLE",
? ? ? ? ? ? ? ? ? ?"CCITTFAX3", "CCITTFAX4")) {
? ?file <- paste(img, "_", compress, ".tif")
? ?writeGDAL(get(img), file, type = "Byte",
? ? ?options = c("NBITS=1", paste("COMPRESS=", compress)))
? ?cat(paste("Compression: ", compress, " from ", file, "\n"))
? ?reread <- readGDAL(file, silent = TRUE)
? ?print(sapply(reread at data, table))
?}
?cat("\n")
}
The expected output is:
*** Correct Stats: img_1bit_1band
?band1
0 ?3086
1 ?6914
Compression: NONE from img_1bit_1band_NONE.tif
?band1
0 ?3086
1 ?6914
Compression: DEFLATE from img_1bit_1band_DEFLATE.tif
?band1
0 ?3086
1 ?6914
Compression: LZW from img_1bit_1band_LZW.tif
?band1
0 ?3086
1 ?6914
Compression: CCITTRLE from img_1bit_1band_CCITTRLE.tif
?band1
0 ?3086
1 ?6914
Compression: CCITTFAX3 from img_1bit_1band_CCITTFAX3.tif
?band1
0 ?3086
1 ?6914
Compression: CCITTFAX4 from img_1bit_1band_CCITTFAX4.tif
?band1
0 ?3086
1 ?6914
*** Correct Stats: img_1bit_2band
?band1 band2
0 ?3086 ?8353
1 ?6914 ?1647
Compression: NONE from img_1bit_2band_NONE.tif
?band1 band2
0 ?3086 ?8353
1 ?6914 ?1647
Compression: DEFLATE from img_1bit_2band_DEFLATE.tif
?band1 band2
0 ?3086 ?8353
1 ?6914 ?1647
Compression: LZW from img_1bit_2band_LZW.tif
?band1 band2
0 ?3086 ?8353
1 ?6914 ?1647
Compression: CCITTRLE from img_1bit_2band_CCITTRLE.tif
?band1 band2
0 ?6131 ?8783
1 ?3869 ?1217
Compression: CCITTFAX3 from img_1bit_2band_CCITTFAX3.tif
?band1 band2
0 ?6131 ?8783
1 ?3869 ?1217
Compression: CCITTFAX4 from img_1bit_2band_CCITTFAX4.tif
?band1 band2
0 ?6131 ?8783
1 ?3869 ?1217
Everything looks fine except CCITT compression with 2 bands.
I wasn't able to dig up Windows GDAL 1.7.1 binaries, and the 1.7.0
binaries from FW_Tools seem to give similar behavior, although it's a
little hard to check because neither ArcGIS nor ENVI seems to like any
of the multi-band files.
-- Michael Sumner Institute for Marine and Antarctic Studies, University of Tasmania Hobart, Australia e-mail: mdsumner at gmail.com
Oliver Soong Donald Bren School of Environmental Science & Management University of California, Santa Barbara Santa Barbara, CA 93106-5131 805-893-7044 (office) 610-291-9706 (cell)