NDVI analysis

[Christian Rixen]

Dear ImageJ community,

I have a question about NDVI analysis of photos of vegetation plots.
I would like to measure the NDVI as a measure for greenup of alpine  
vegetation during the summer.
I have two digital cameras where the IR filter was removed  
(CanonPowershotG10, IR Filter removed by the company LDP LLC,  
www.maxmax.com). Then I have two filters, one for the visible light  
(XNiteCC1-58) and one for the IR range (850 nm). I should be able to  
get a reliable NDVI measurement by taking 4 pictures (visual plot, IR  
plot, visual sky, IR sky) and calculating the NDVI according the the  
equation NDVI= (NIR-VIS)/(NIR+VIS) (see method description on  
http://www.iac.ethz.ch/staff/stockli/ndvimeasurement/ndvimeasurement.html).
I would of course use a manual exposure, I would not move the camera  
between 2 shots and I would use a frame.

Now, here is my ImageJ question:
I want to get the red channel from my visible-light photo. Can I just  
analyze the red channel of my picture with
Image -> color -> RGB split
and then measure the mean brightness of my red picture? And then I  
could measure the brightness of my IR picture to obtain the brightness  
values that I need for my NDVI equation?
I am a bit uncertain because I am not very familiar with image  
analysis. What is the unit of the brightness value in ImageJ? Is the  
value between 0 and 255 an official brightness value that I can use?  
If yes, I would be set and could do my NDVI calculations.

Thanks to the community for your input.
Christian

[Michael Schmid]

Hi Christian,

a few remarks:

(1) For subtracting, you need a linear transfer function. Most 3*8-
bit color images created by digital cameras, e.g., JPG, are  
nonlinear. You need additional software to convert your camera's RAW  
format (if it has one) to some format (that is readable by ImageJ)  
with a linear transfer function.

(2) Even worse, there is usually some color enhancement, e.g. the  
saturation is increased (Canon cameras have a 'neutral color' mode  
where this is not so much of a problem).

(3) The IR blocking filter will also slightly attenuate visible colors.

(4) ImageJ brightness units are pixel values as they are in the file,  
unless you use Analyze>Calibrate to change this.


Michael
________________________________________________________________

On 2 Jun 2009, at 20:26, Christian Rixen wrote:

> Dear ImageJ community,
>
> I have a question about NDVI analysis of photos of vegetation plots.
> I would like to measure the NDVI as a measure for greenup of alpine  
> vegetation during the summer.
> I have two digital cameras where the IR filter was removed  
> (CanonPowershotG10, IR Filter removed by the company LDP LLC,  
> www.maxmax.com). Then I have two filters, one for the visible light  
> (XNiteCC1-58) and one for the IR range (850 nm). I should be able  
> to get a reliable NDVI measurement by taking 4 pictures (visual  
> plot, IR plot, visual sky, IR sky) and calculating the NDVI  
> according the the equation NDVI= (NIR-VIS)/(NIR+VIS) (see method  
> description on http://www.iac.ethz.ch/staff/stockli/ndvimeasurement/ 
> ndvimeasurement.html).
> I would of course use a manual exposure, I would not move the  
> camera between 2 shots and I would use a frame.
>
> Now, here is my ImageJ question:
> I want to get the red channel from my visible-light photo. Can I  
> just analyze the red channel of my picture with
> Image -> color -> RGB split
> and then measure the mean brightness of my red picture? And then I  
> could measure the brightness of my IR picture to obtain the  
> brightness values that I need for my NDVI equation?
> I am a bit uncertain because I am not very familiar with image  
> analysis. What is the unit of the brightness value in ImageJ? Is  
> the value between 0 and 255 an official brightness value that I can  
> use? If yes, I would be set and could do my NDVI calculations.
>
> Thanks to the community for your input.
> Christian

[Angus Jackson]

Hi Christian,

If you can get access to
Murphy RJ, Underwood AJ, Pinkerton MH (2006) Quantitative imaging to measure photosynthetic biomass on an intertidal rock-platform. Mar Ecol Prog Ser 312:45-55
you may find it interesting/useful.

There is also a MS about doing similar measurements with more conventional digital cameras but this is still in review.  Hopefully available soon.

Much better than using images of the sky to quantify incident radiation is to use a reflectance standard (e.g. spectralon) in the image.  Incident radiation can change rapidly between images so it is better to have incident light quantified in the same image as your sample.  With the reflectance panels available for calibration, there is also no longer any restriction on using manual exposure.

I know it doesn't answer your questions but I hope that helps.

Cheers,

Angus

Christian Rixen wrote

Dear ImageJ community,

I have a question about NDVI analysis of photos of vegetation plots.
I would like to measure the NDVI as a measure for greenup of alpine  
vegetation during the summer.
I have two digital cameras where the IR filter was removed  
(CanonPowershotG10, IR Filter removed by the company LDP LLC,  
www.maxmax.com). Then I have two filters, one for the visible light  
(XNiteCC1-58) and one for the IR range (850 nm). I should be able to  
get a reliable NDVI measurement by taking 4 pictures (visual plot, IR  
plot, visual sky, IR sky) and calculating the NDVI according the the  
equation NDVI= (NIR-VIS)/(NIR+VIS) (see method description on  
http://www.iac.ethz.ch/staff/stockli/ndvimeasurement/ndvimeasurement.html).
I would of course use a manual exposure, I would not move the camera  
between 2 shots and I would use a frame.

Now, here is my ImageJ question:
I want to get the red channel from my visible-light photo. Can I just  
analyze the red channel of my picture with
Image -> color -> RGB split
and then measure the mean brightness of my red picture? And then I  
could measure the brightness of my IR picture to obtain the brightness  
values that I need for my NDVI equation?
I am a bit uncertain because I am not very familiar with image  
analysis. What is the unit of the brightness value in ImageJ? Is the  
value between 0 and 255 an official brightness value that I can use?  
If yes, I would be set and could do my NDVI calculations.

Thanks to the community for your input.
Christian

[dllewellyn]

<quote author="Angus Jackson">
Hi Christian,

If you can get access to
Murphy RJ, Underwood AJ, Pinkerton MH (2006) Quantitative imaging to measure photosynthetic biomass on an intertidal rock-platform. Mar Ecol Prog Ser 312:45-55
you may find it interesting/useful.

There is also a MS about doing similar measurements with more conventional digital cameras but this is still in review.  Hopefully available soon.

I am a bit uncertain because I am not very familiar with image  
analysis. What is the unit of the brightness value in ImageJ? Is the  
value between 0 and 255 an official brightness value that I can use?  
If yes, I would be set and could do my NDVI calculations.

Thanks to the community for your input.
Christian
******************
Christian,

This is Dan from MaxMax.com.  We have a new camera designed just for vegetation studies.  The camera takes a picture in the infrared and visible band with the differnt bands going to different color channels.  We wrote a quick macro for ImageJ to calculate the NDVI.  Please let me know if you are interested.

[dllewellyn]

As a follow up, we are adding more information at

http://www.maxmax.com/vegetation_stress.htm

We have written some macros for ImageJ to calculate NDVI for our vegetation analysis camera.

[larry.lugo]

Hi Dan,
I'm very interested on your macros for ImageJ to calculate the NDVI.
They will be used for teaching undergraduate students regarding NDVI, LAI and their relations and calculations.
I work at Universidad del Zulia, Venezuela, a public university with a very small budget so we need to use Free Software like ImageJ.
I´m not an expert on writing macros or any other script, so your help will be really appreciated.
On the other hand, we are testing this kind og technology and MaxMax cameras could be an option for researching/teaching projects.
Thank you very much in advance.

Larry Lugo, Associate Professor
Cátedras: Ecología / Cultivos perennes
Facultad de Agronomía
Universidad del Zulia (www.luz.edu.ve)
Av. 16, Ciudad Universitaria
Maracaibo 4005, Venezuela
Twitter: @larrylugo, @fagroLUZ
Email: [e-mail]