Posted by
Charles Davis on
Jun 15, 2016; 1:54pm
URL: http://imagej.273.s1.nabble.com/Help-tp5016640p5016653.html
Hi, again Michael,
I agree. BUT what I'm doing is quantifying how different RAW editors render
JPEG images.
Thus, I HAVE to use JPEGs. :-)
Thanks for your help.
-----Original Message-----
From: ImageJ Interest Group [mailto:
[hidden email]] On Behalf Of
Michael Schmid
Sent: Wednesday, June 15, 2016 3:04 AM
To:
[hidden email]
Subject: Re: Help?
Hi Charles,
if you use JPG images, please don't expect any quantitative results!
(1) JPG is a lossy compression, with lots of compression artifacts.
See
http://imagej.net/Principles#Why_.28lossy.29_JPEGs_should_not_be_used_in_imaging
Especially the color information (chroma) has low resolution and many
artifacts. If you want to compare gray level, make sure you use weighted
conversion, otherwise you have to reduce the resolution by at least a
factor of 2 to get rid of at least the worst compression artifacts.
(2) JPG has a nonlinear relation between pixel value and intensity. If
you calculate differences, the result will NOT be the difference of the
intensity values.
Michael
________________________________________________________________
On 2016-06-14 22:20, Charles Davis wrote:
> I hope I'm doing this reply correctly [this is my 1st reply].
>
> To Melissa: No I'm loading a JPEG image, not a NEF.
On 2016-06-14 22:30, Charles Davis wrote:
> Michael, I had figured out that I needed to convert to grayscale.
Otherwise
> I assumed I would get RGB values. I also dug deeply enough to have found
the
> choice of grey conversion. As I'm not doing perceptual analysis, I kept
the
> non-weighted conversion.
>
> The answer was to simply File>Save As>Text Image. DUH!
>
> I am comparing two images that are very similar. I have been subtracting
> them w/ Photoshop, but have also found that imagej can do that
subtraction.
> :-)
>
> I am wanting to count pixel values and use a custom algorithm to quantify
> the difference. I probably could use imagej, but the learning curve is
very
> steep and I am comfortable w/ Excel.
>
> Thanks...
>
> -----Original Message-----
> From: ImageJ Interest Group [mailto:
[hidden email]] On Behalf Of
> Michael Schmid
> Sent: Tuesday, June 14, 2016 1:52 PM
> To:
[hidden email]
> Subject: Re: Help?
>
> Hi Charles,
>
> it seems you need a grayscale image first. Assuming you are starting
> with RGB, use Image>Type>8 bits.
> Make sure you have the correct choice of "Weighted RGB Conversions" in
> Edit>Options>Conversions
>
>
>
> Then save as text image.
>
> Anyhow, what kind of analysis with all pixels do you want to do in Excel?
> I guess that ImageJ could do essentially everything that can be done in
> Excel, but much faster.
>
> Michael
> ________________________________________________________________
> On 2016-06-14 18:10, Charles Davis wrote:
>> I'm a new imagej user and am having a problem. I know that imagej will
>> generate the data that I want, because I [somehow] got it to do it once.
> But
>> I don't remember the sequence of operations that I went through to get
> that
>> result. I have spent 2 weeks, working several hours each day and have not
>> been able to study and find the answer or try everything that seems
>> reasonable. My pride and ego is at ZERO! I can't do this w/o help.
>>
>>
>>
>> Background:
>>
>>
>>
>> I'm using a Nikon D810 camera which produces large files. I need a .csv
or
>> .txt file w/ the brightness values. I don't want a Histogram, but the
data
>> behind the Histogram.
>>
>>
>>
>> I have been able to produce files w/ all the pixel data [X, Y, R, G, B]
in
>> one row. That is too large to load into Excel [36,200,000 is too many
> rows].
>> Yes, I have successfully attempted to break it up into several smaller
>> files, but that takes hours to do one file! Then I have to extract the
>> summary data from each of these 2-dozen+ files and create a global
summary
>> file.
>>
>>
>>
>> As I said, I was able to [somehow] create a .txt file that was organized
>> differently. I had all the Y's in the rows and all the X's in the
columns.
>> The data in each cell was the gray value [I think]. This had only 4913
> rows
>> x 7359 columns and loaded into Excel [slowly]; it's still big, but
>> manageable.
>>
>>
>>
>> Sorry for the long intro, getting to my question: How do I recreate
> another
>> of those Measurements.txt files, w/ only gray brightness data in cells
and
--
ImageJ mailing list:
http://imagej.nih.gov/ij/list.html--
ImageJ mailing list:
http://imagej.nih.gov/ij/list.html