The colour ranges of retinal cones are: Blue-Yellow Red-Green Light-Dark

In photography: The subtractive colour spectrum (start by calibrating White-balance) gives red, green, blue. The additive colour spectrum (start by calibrating Black-balance) gives cyan, yellow, magenta.

This is what it looks like when missing with overlapping filters:

This is how software helps you adjust those balances in RGB & CYM:

I hope this helps you find a spectrum of relevance to you. :)

Greetings, Just post the CSV file for the gradient. It is easy enough to create the LUT file.

Here is a tab delimited file of the new lut:

NDVI_VGYRM-lut.txt

Will Fiji make a lut file from this?

Actually, Fiji needs as CSV file to work on. So, I copied and pasted your text file into EXCEL and deleted the top line. Each column in your text file needs to be in a separate column in the excel file. So, saved as a csv file and then used the LUT IMPORT plugin to bring the csv file up then used IMAGE>COLOR>EDIT LUT to save the LUT file. You can also invert the gradient and save the invert too - which I did. I have tried attaching the files but for some reason they don't want to upload. If you have trouble creating your own LUT files, send me your e-mail to mine as aswicks@gmail.com I will send them to you.

Great colormap! I am using MATLAB to calculate and create images based on the Infragram Point and Shoot (modified Mobius). When I compare my images to those generated using the "Fastie" colormap on the infragram.org page, I noticed that there is a small but subtle difference. Attached are two images comparing the website:

and my MATLAB script:

It looks like there may be some image processing being done on the website. Are you using histogram stretching, or a modified NDVI calculation? Currently, I am just using NDVI = (B-R) / (B+R).

Thanks!

Jeff

Wow, I didn't know there was a Fastie colormap at Infragram.org. I think it will need a new name (How about VGYR for violet green yellow red?). I guess Jeff ported it to Infragram. The translation must have been done by hand so there were probably translation mismatches. But there might also be processing differences.

Your output from MATLAB looks excellent. Nice legend.

The Fastie colormap was already implemented in infragram.org, so I can't take any credit there!

What I am wondering is why there seems to be a shift in the NDVI between the MATLAB and infragram.org outputs. It looks as if the NDVI at infragram.org is a bit higher than the values which I calculated using your colormap. Hence the darker background and more green in the vegetated areas.

When calculating the NDVI, I had to convert uint8 to a double and then reclassify the images. In doing so, I tested both floor and ceiling functions, but the outputs were not shifted enough to change the output to match the infragram.org image. This is why I was wondering if the images were pre-processed with histogram stretching or normalization which may result in a bit wider range...

Jeff

I've checked the javascript from the Infragram.org site and could not see any image preprocessing from what was there, the NDVI index was calculated directly from the IR and R channel intensity values, and the colormap was fitted to a range of +1 to -1.

Still need to figure out where the shift comes from...

In the meantime, I've uploaded an animation showing the results of the Infragram point and shoot taken in a park nearby after a decent rainstorm. I will go back to photograph the same spot once the ground has dried out to see if there are any big differences in the NDVI.

https://www.youtube.com/watch?v=k8T_PpiEgxg&list=UUNp0vouv7d04EqdXvH1phgw

Jeff, It was the other Jeff (Warren) who must have added the new colormap at infragram.org. I assume some differences snuck in during the manual translation. Your video link did not work for me. Chris

Hi, I've updated the YouTube link, should work now. Contacted Jeff regarding the implementation in the Infragram.org site to see what he thinks... Jeff T

with this camera can I see how deep is ice?

I think any camera could allow you to estimate how thick ice is, if you can put the ice between the camera and the light source. The thicker the ice, the more light will be absorbed and the dimmer the light source will appear on the other side. Near infrared light behaves similarly to red light when passing through ice -- both are mostly absorbed by the ice, but NIR is absorbed more strongly.

.
The blue end of the spectrum is not absorbed as strongly, which is why ice can appear blue (only the blue wavelengths pass through). Ice and liquid water both absorb visible and near infrared light about the same amount, although imperfections in the ice can make it more opaque. The variability of those imperfections will make it difficult to use light transmission to estimate the thickness of ice because the light transmission will depend mostly on the imperfections and not the thickness. Estimating ice thickness by using a photo of the surface of the ice on a frozen body of water might be difficult, but on any water body there might be subtle cues related to thickness, especially soon after the first freezing event. Chris

Very interesting post. How can one take this a step further and analyze the % green areas that correspond to greenery? For example, the output would be to get the square footage of area that is cultivated or green.

GIS programs are really good at doing that kind of spatial analysis. If you give them the appropriate information, they can scale and correct the aerial images so they relate to a map (orthorectification). Then they can report the area of each color that is present in a given part of the image and convert that to area on the ground. Other types of image manipulation programs can report about pixel counts of different colors, but might only provide relative numbers without absolute area measurement or correction for non-vertical photos or non-flat terrain. Chris

I have updated this note to include a link to the new colormap, which is also here:

NDVI_VGYRM.lut

This should be copied to the C:\Program Files (x86)\Fiji\Fiji.app\luts directory or something like that on Windows, and I guess to .../Applications/Fiji.app/luts on Mac.

Chris

how to calculate the chlorophyll content in terms of NDVI?

Hi jtuhtan. You can get a MATLAB script? Thanks!

Id like to use the alternating white/black/white/cyan/green/yellow/red/magenta gradient, Im using GIMP 2, and Im wondering how to do it? Ive just joined this forum, and found it really useful.

Ive also got 2 mobius cams, one infrablue, the other RGB. Id like more useful color data, i just dont know how to do it. can you help?

Thank you Patrick

Hi Patrick,

To do this in GIMP you will need to work with indexed images and upload a GIMP palette file (.gpl) which follows the Fastie colormap.

My advice would be up upload Chris' NDVI_VGYRM.lut file and then using a text editor, compare it to the .gpl file to see how you can convert the files from one format to another. Once you have loaded the new .gpl then you can try it on your mobius imagery.

Jeff

Hi Patrick,

To do this in GIMP you will need to work with indexed images and upload a GIMP palette file (.gpl) which follows the Fastie colormap.

My advice would be up upload Chris' NDVI_VGYRM.lut file and then using a text editor, compare it to the .gpl file to see how you can convert the files from one format to another. Once you have loaded the new .gpl then you can try it on your mobius imagery.

Jeff

Thank you for a quick response jtuhtan.

I have the NDVI_VGYRM.lut, but have know idea how to use a .gpl file. Ive seen this: https://www.youtube.com/watch?v=-nNnWEHNO_w Im trying to get this done for a class in the next 2 weeks. This remote sensing just has so much potential.

-Patrick

you can find my posts here: http://publiclab.org/notes/PatrickC/11-13-2014/binocular-mobius-tripod-for-nvdi any help is appreaciated.

I still cant get the .lux file to work, and i need to get this working for class. is there a explanation to use the .gpl type files?

TY, -Patrick

Jeff (Jtuhtan) could you please direct me to the right place so I can create a legend with values on my NDVI photos...

Thank You

David

Hi David,

The code I used was MATLAB, overlaying an image of the Fastie colormap onto the NDVI image. What software are you working with?

I use Fiji with the photomonitoring plugin...I would like to create a legend with the values, what would be AWSOME is to have the values on the image as well, not sure if this is possible.

To be honest I'm new to all this and I'm not real sure what the difference in Fastie, stretched and normal processing do.

I currently have a drone with a NDVI-7, an NDVI lens created by IR pro, camera on it.

Hey jtuhtan ! I'm trying to create a colormap like yours in matlab, but cannot figure it out how to change half of the color range (from -1 to 0) in gray and the rest in RGB. Is it possible to share your matlab script? It will help my visualize better my NDVI image. Thank you in advance! Keep up with the great work!

Hello @cfastie ,

It's a good post, you have good photos. I'm new in photo monitoring. I have a question which camera do you use to take this imagens and in wich format (.jpg, .png, ....)? You only use Wratten 25A filter?

A few different cameras were used to take those photos. Most of the cameras were Canon PowerShots. I think all of the photos were saved by the camera as jpegs. Some of the NDVI images were made from photos taken with a camera modified with a Wratten 25 filter. Some were made with a two camera system (see text).

Chris