Help for a new user to analyze percentage of certain colored pixels in digital image?

[Emily Teng]

Hi everyone,
I am completely new to ImageJ.  Just found out about it today and hoping it
is exactly what I need for my research purposes.  I am a PhD student.

I need a method to analyze the percent of coloration of poinsettia plants
over time.  I plan to take digital photos of the top of the canopy over
time and analyze those images.  As I understand it, I could use ImageJ for
this purpose correct?

Taking for example, a red poinsettia - I would need the program to count
all the pixels that are any shade of red (vs green) in the plant.

http://www.ecke.com/poinsettias/bractmeter/

That is what I'm trying to do.  The professor who created that told me he
had someone do programming to count the pixels.  But I'm thinking ImageJ
will be able to do this if I know how to use it?

Can someone tell me
1.  Can ImageJ do this?
2. If yes -  Point me in the direction of how I should go about learning
how to do this?

Thank you in advance for any help and guidance.

Emily

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[Brandon Hurr]

1. Yes
2. Look at color thresholding. Image -> Adjust -> Color Thresholding

Since you are doing color analysis, you need to be very careful with the
settings of your camera and your lighting. Taking the photos under
different white balance settings or different lights will affect your
thresholding.

I would segment for total plant area visible (red + green) and then find
red only and compute the difference in area.

The simplest way to create a macro would be to use the recorder. Do the
steps in pieces and use google to search past questions on this list and
stackoverflow.

Focus on image quality first and foremost, and then the analysis will
follow.

B



On Thu, Jul 21, 2016 at 7:51 PM, Emily Teng <[hidden email]> wrote:

> Hi everyone,
> I am completely new to ImageJ.  Just found out about it today and hoping it
> is exactly what I need for my research purposes.  I am a PhD student.
>
> I need a method to analyze the percent of coloration of poinsettia plants
> over time.  I plan to take digital photos of the top of the canopy over
> time and analyze those images.  As I understand it, I could use ImageJ for
> this purpose correct?
>
> Taking for example, a red poinsettia - I would need the program to count
> all the pixels that are any shade of red (vs green) in the plant.
>
> http://www.ecke.com/poinsettias/bractmeter/
>
> That is what I'm trying to do.  The professor who created that told me he
> had someone do programming to count the pixels.  But I'm thinking ImageJ
> will be able to do this if I know how to use it?
>
> Can someone tell me
> 1.  Can ImageJ do this?
> 2. If yes -  Point me in the direction of how I should go about learning
> how to do this?
>
> Thank you in advance for any help and guidance.
>
> Emily
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[Emily Teng]

Hi Brandon,

Thank you so much for the guidance.  I believe my lab recently purchased a
"real camera" though I'm not sure exactly what kind.  If I always use the
same settings and take the pictures in the same spot indoors - should that
be sufficient?

I've just downloaded the program, so I will play with it a bit and see what
I can learn.

Emily

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[TimFeinstein]

Professional color analysis can get very sophisticated, so it is a
question of how much precision you need (and can afford).  In my view much
of your effort should focus on standardizing the acquisitions.  Acquiring
your images in RAW mode with standard settings will help, but that still
leaves a lot of variables, especially if your plants are in a greenhouse
rather than a fully light controlled grow space.  Your workflow should
include a standardized color correction process, meaning a color card that
you include in each photo with a software that detects the card and
automatically corrects it.  For example here is a workflow that uses the
widely used X-Rite color checker and Rawtherapee, a free and open-source
alternative to Adobe Lightroom.

https://stephenstuff.wordpress.com/2012/07/06/digital-camera-profiling-with
-raw-therapee-and-argyll-cms/

I suggest you color correct the RAW files in Rawtherapee, then export as
TIFFs to analyze in ImageJ/Fiji.  This can almost certainly be automated
to let you process large batches of data at once.

If you are working in a greenhouse, weather conditions will dramatically
change your results.  In that case think about using flash lighting rather
than ambient.  Naked camera flashes cause bright spots that would ruin
your data; for your purposes I'd recommend a diffused ring light (see
example below) at a set distance from the plant.

http://www.bhphotovideo.com/c/product/996516-REG/roundflash_roundflashmb_ma
gnetic_black_rigflash_adapter.html

Best,


Tim



Timothy Feinstein, Ph.D.
Research Scientist
University of Pittsburgh Department of Developmental Biology





On 7/21/16, 10:51 PM, "ImageJ Interest Group on behalf of Emily Teng"
<[hidden email] on behalf of [hidden email]> wrote:

>Hi everyone,
>I am completely new to ImageJ.  Just found out about it today and hoping
>it
>is exactly what I need for my research purposes.  I am a PhD student.
>
>I need a method to analyze the percent of coloration of poinsettia plants
>over time.  I plan to take digital photos of the top of the canopy over
>time and analyze those images.  As I understand it, I could use ImageJ for
>this purpose correct?
>
>Taking for example, a red poinsettia - I would need the program to count
>all the pixels that are any shade of red (vs green) in the plant.
>
>https://na01.safelinks.protection.outlook.com/?url=http%3a%2f%2fwww.ecke.c
>om%2fpoinsettias%2fbractmeter%2f&data=01%7c01%7ctnf8%40pitt.edu%7cfa4db3a9
>12dd46e5129c08d3b1dcaa28%7c9ef9f489e0a04eeb87cc3a526112fd0d%7c1&sdata=69bn
>gVNCr%2bGvhlbT3U7xLAmQ2ASxHI9LPYWP0EYVtxw%3d
>
>That is what I'm trying to do.  The professor who created that told me he
>had someone do programming to count the pixels.  But I'm thinking ImageJ
>will be able to do this if I know how to use it?
>
>Can someone tell me
>1.  Can ImageJ do this?
>2. If yes -  Point me in the direction of how I should go about learning
>how to do this?
>
>Thank you in advance for any help and guidance.
>
>Emily
>
>--
>ImageJ mailing list:
>https://na01.safelinks.protection.outlook.com/?url=http%3a%2f%2fimagej.nih
>.gov%2fij%2flist.html&data=01%7c01%7ctnf8%40pitt.edu%7cfa4db3a912dd46e5129
>c08d3b1dcaa28%7c9ef9f489e0a04eeb87cc3a526112fd0d%7c1&sdata=UHL9cmAP8nfj9ep
>Q9eAiaIliVF53IHbJiKHa76C0aNQ%3d

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[Emily Teng]

Thank you so much Tim for your response.  I actually grew up in
Pittsburgh!  My father was a professor at the business school....

Anyhow, I actually have Adobe Lightroom - it came with my laptop. so that's
helpful I suppose.

I was thinking when I take the photos I would take them indoors  and set up
a black velvet backdrop we use for photos as the outdoors is generally too
sunny for good photos during the day (I'm in Hawaii). Do you think that
light ring would still be necessary? I suppose it couldn't hurt right?
Though keeping costs down is always good.

Another group suggested when processing to convert the image to grayscale
and then threshold it to examine the colors. Is the color correcting card
necessary  Because I see that it is rather expensive.

I'll try to get a sample photo and perhaps that would be helpful.

Thanks again.

Emily

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html