[R] Reading a TIFF file

[Uwe Ligges]

Fine, looking at the code shows that readTiff read the image with using 
libtiff which should be fine. Afterwards, a pixmap obnject is generated 
via  pixmapRGB() which includes the following lines:

     datamax <- max(data)
     datamin <- min(data)
     data <- as.numeric(data)
     if (datamax > 1 || datamin < 0)
         data <- (data - datamin)/(datamax - datamin)

That means the data is perfectly fitted into the [0,1] interval.

That means we have not only rescaled but also another 0 now.

What I did is:

pic <- readTiff("test1_layer1.tif")
pic <- pic at red
ARC <- read.csv2("test1_arcgis.csv", header=TRUE)
ARC <- matrix(ARC[,2], nrow=nrow(pic), byrow=TRUE)

plot(pic)
plot(ARC)# looks *very* similar

plot(as.vector(pic) ~ as.vector(ARC))
# all on one line

summary(lm(as.vector(pic) ~ as.vector(ARC)))
# resuiduals < 10^(-14)

So the formula use to get from the ARC to the pixmap data is
pixmapdata = -0.82278 * 0.01266 ARCdata

if you want to get the original data, you can adapt the readTiff 
function for your own use for greyscales (and save memory that way) as in:


myReadTiff <- function (fn, page = 0)
{
   w <- .C("TiffGetWidth", as.character(fn), w = as.integer(0),
     PACKAGE = "rtiff")$w
   h <- .C("TiffGetHeight", as.character(fn), h = as.integer(0),
     PACKAGE = "rtiff")$h
   nw <- ceiling((1 - reduce) * w)
   nh <- ceiling((1 - reduce) * h)
   if (w > 0 && h > 0) {
     tiff <- .C("TiffReadTIFFRGBA", as.character(fn),
       page = as.integer(page),
       r = integer(w * h), g = integer(w * h), b = integer(w * h),
       PACKAGE = "rtiff")
     tiff <- tiff$r / 255
     tiff <- pixmapGrey(data = tiff, nrow = nh, ncol = nw)
     return(tiff)
   }
   stop("Could not open", fn, ".  File corrupted or missing.\n")
}

pic2 <- myReadTiff("test1_layer1.tif")


Now pic2 at grey will be exactly the same as ARC/255 from above.

Uwe Ligges












On 22.04.2011 15:10, Julio Rojas wrote:
> Dear Uwe, find attached an small portion of the file I'm working with. ArcGIS values for this file are in the CSV file. They are a vector of all rows put together.
>
> Thanks and regards.
>
>
> --- El vie, 4/22/11, Uwe Ligges<ligges at statistik.tu-dortmund.de>  escribió:
>
>> De: Uwe Ligges<ligges at statistik.tu-dortmund.de>
>> Asunto: Re: [R] Reading a TIFF file
>> A: "Julio Rojas"<jcredberry at ymail.com>
>> Cc: r-help at r-project.org
>> Fecha: viernes, 22 de abril de 2011, 12:37 pm
>> Unless you can provide a small
>> reproducible example (say a very small
>> tiff including the values you got from the non-R software)
>> it will be
>> hard to tell what is going on.
>>
>> Uwe Ligges
>>
>>
>> On 22.04.2011 11:23, Julio Rojas wrote:
>>> Dear all, I have been trying to speed up a process we
>> have been done in ArcGIS. We have to read a single layer
>> TIFF (monochrome image) in . For this, I have used the
>> "rtiff" package. After reading the TIFF file, I compared the
>> raw values for each pixel that I have in ArcGIS to the ones
>> obtained in R. In ArcGIS I have discrete values in the range
>> 0..255, while in R I have continuous values between 0..1.
>> This, in itself might not be a problem if the values
>> obtained in R, times 255 would show the values obtained in
>> ArcGIS, but this is not the case. The images are very
>> different. I tried to settle matters using Photoshop, and
>> the values there are completely different from the other two
>> (using RGB, the K value (in CMYK) or the B value (in
>> HSB))!!!
>>>
>>> Can somebody help me with this problem? Can I trust
>> "rtiff"? Should I stick to a very slow process in ArcGIS?
>> Why PS, which should be the perfect measuring stick, is
>> showing another set of values?
>>>
>>> Thanks in advance. Regards.
>>>
>>> Julio
>>>
>>> ______________________________________________
>>> R-help at r-project.org
>> mailing list
>>> https://stat.ethz.ch/mailman/listinfo/r-help
>>> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
>>> and provide commented, minimal, self-contained,
>> reproducible code.
>>