imagesave and normalization (bug?)

Share icon

Mon, 06/25/2012 - 02:31 pm

I am building some .tiff files, and am trying to save them without normalizing. After creating my data (and checking it by loading it into a table before saving) I attempt to save as 16 bit using the following:

imagesave/D=16/U/T="tiff"

This results in a normalized image (max value 65534, mean 28307... ) Dropping the d and t flags and using /F instead results in the proper values, albeight single precision float values. (which is fine, but still..)

Is this a bug, or am I making some kind of error with the syntax? My test file was simply a 50x50 matrix filled with poissonNoise(400).
You are not telling us which version of IGOR and what platform/OS you are using.

With the latest version here is my test:
make/n=(100,100)/w/u ddd=poissonNoise(400)
imagesave/D=16/U/T="tiff" ddd
ImageLoad/T=tiff "HardDisk:Users:ag:Desktop:ddd.tif"
Rename 'ddd.tif',iii
MatrixOP/O aa=sum(abs(ddd-iii))
Print aa[0]


This prints 0 as you would expect.

A.G.
WaveMetrics, Inc.

June 25, 2012 at 03:04 pm - Permalink

Sorry, I am using Igor Pro 6.22A on a Windows XP machine (if the OS changes the data IGOR returns, thats a pretty major issue and I would be surprised)

June 25, 2012 at 03:12 pm - Permalink

Its my mistake, I didnt realize I needed these flags in the Make command.

June 25, 2012 at 03:34 pm - Permalink

daggaz wrote:
Sorry, I am using Igor Pro 6.22A on a Windows XP machine (if the OS changes the data IGOR returns, thats a pretty major issue and I would be surprised)

In order to reproduce your results we need to be able to run our tests under the same conditions that you did. As for OS changes: any time you are writing to disk you have machine dependent code. Experience suggests that you have to be an optimist to expect this code to always produce identical results across all platforms and all OS versions.

June 26, 2012 at 10:42 am - Permalink