Question: What is the correct interpretation for the NDVI infragram color scale?

DuvTorres is asking a question about infragram
Follow this topic

by DuvTorres | May 27, 2015 17:01 | #11845


Hi, Now I’m working in some tests with aerial crop images for NDVI calculating, however I have had some problems for interpretation of infragram results. I don´t understand the color scale used by the software, because theorically the higher NDVI have to show red color (good chlorophylla levels), and near zero have to be Green (soil), but I got different results in each image I have processed, I need your help to can understand this values or to stablish a correct scale that let an adjusted interpretation for determine health of crops.

Thanks!

resume.png



4 Comments

Questions:

What is your input and can you show the input data as well (mainly the red and near infrared images)?

What software did you use to convert the images to NDVI? (cite code snippets if not the photo monitoring plugin in imagej / fiji)

Without this information it is hard to help you figure out what is going on.

Cheers, K

Is this a question? Click here to post it to the Questions page.

Reply to this comment...


That looks like a photo taken with a red filtered infrared camera. It is very hard to get meaningful NDVI results directly from this photo because it is too gray. Grayish photos have similar values in all three color channels, so NDVI will always compute to approximately (x-x)/(x+x) = zero. That seems to be exactly what you got from infragram.org.
.
ResumeHist.JPG
Histogram for an area of vegetation in the photo.
.
Good photos from a red filtered infrared camera will have a turquoise hue. The blue channel (NIR) is much brighter than the red channel.
.
resumeCBHist.jpg
Photoshop was used to make the photo more turquoise. The histogram is for the area in the marquee.
.
Although it is possible to manipulate the color of this photo so that it produces NDVI values in the proper range, dynamic range suffers greatly and there is little variation in NDVI values.
.
NDVICBBs_B.JPG
NDVI image produced in Fiji from the adjusted photo. In Fiji, the histogram of the NIR channel was stretched because otherwise the result was poor.
.
Although it is possible to manipulate the processing so that reasonable looking NDVI images are produced, such results are rather subjective and generally can be used to infer only the most basic patterns of plant health. The processing I did to this photo is probably not possible at infragram.org.

Chris

Reply to this comment...


Thanks @cfastie.

I have processed some RGB images with fiji using the NDVI variogram you used for this one, getting good results. However I never could process filtered images, for this reason I want to know the method you have used for make the corrections on this image, and the tool you used for get NDVI map on fiji (I have inastalled the lut and the plugin mentioned in another publicLab forums)

Thanks for your help,

Reply to this comment...


For a photos from a red-filtered camera like the bluish photo above, in the Photo Monitoring plugin select the red channel to represent red and the blue channel to represent NIR. try selecting "stretch histogram" to see if that improves the result.

To get more turquoise photos, the white balance of the camera might have to be altered.

Reply to this comment...


Log in to comment