# Question: Question:

Monoxyde asked on August 09, 2015 08:20
52 | 0 answers | #12127

Goodmorning everyone. I am interested and following your research on NDVI. I have written previously but then, because other problems, I could not deepen. I bought a Point & Shoot Mobius with red filter and the parameters already properly set White Balance, from PublicLAB. I want to get aerial imagery NDVI for agriculture.

I downloaded ImajeJ, followed all instructions and Photo Monitoring plugin installed. Load the test image, and then from the Plugins menu, I select Photo Monitoring / single image processing from index displayed image. At this point it opens the "panel of doubts"

In the "Enter Variables" panel, which settings / values ​​I set? Which should I choose Index Type for calculation suitable to my Mobius? What .Lut I choose? .... So that panel Enter Variables, are driving me crazy! Then help me figure out where to take the relative scale colors appear in the image? I hope not to bore you too much with these problems by beginner and maybe to soon share some evidence serious. Attaching screenshoot and

image proof made Mobius. Thank you!

Monoxide, Try these parameters:

The NDVI_VGRYM.lut comes with the plugin.
For other Mobius photos, try different stretch values (0 to 10), and try stretching one or both channels. The brightness of the Mobius photos (over- or under-exposed) will make a big difference in the NDVI values. Exposure and white balance can be adjusted in the Mobius setup program.
Chris

Thank you so much Chris , I try to take some new photo with MOBIUS and then publish it . Sorry again the same question as before: in this case , which color sccale should I use to evaluate and measure the results ? In your experience , which modified camera, it is more efficient to get NDVI images ? Thanks again for everything that you share! Monoxyde.

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

I think the NDVI_VGRYM.lut color scheme is a good one to start with. What ever you use, include a key to the colors with every NDVI image you share. The camera depends on your needs (wide angle vs. not, light weight vs. not, crisp image vs. not, good in low light vs. not, cheap vs. not, easy to convert to NIR vs. not, etc). Some PowerShots are excellent compromises.

Chris

Thanks again Chris , actually the question as phrased , was not very specific . I noticed that my Mobius against wide-angle lens , for aerial applications, Lost light and contours and show problems in the shadow pixels. I would try to change , removing IR filter and entering blue filter on a Canon PowerShot ELPH 130is , as explained in great tutorial Video by Dennis Baldwin https://www.youtube.com/watch?v=lgBNznwiMJQ.

I believe that the Mobius with RED filter and the right White Balance setup, is best suited for pictures taken from the ground to the vertical vegetation and not for aerial photos. But this is and wants to be just my humble opinion , with little information and few tests to prove it . I'm buying online a Canon PowerShot ELPH 130is because in Italy has never been marketed . Let's see how it will behave once modified and adjusted the White Balance, in flight with gimbal on my exacopter . Some advice? Thank you!

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

The standard lens on the Mobius might be too wide to allow good analytical photos. But other lenses are available: http://publiclab.org/notes/patcoyle/08-01-2015/mobius-non-fish-eye-lens-conversion

A red filter is probably going to offer better results than a Rosco blue filter for the reasons in this note: http://publiclab.org/notes/nedhorning/11-01-2013/why-a-red-filter-should-work-well-for-ndvi

You will probably notice the critical error in that ELPH 130 conversion video when you get to the point where you remove the IR block filter. Pay attention to how you remove the filter so you know how to replace it.

Chris

Hi Chris. Ok , I had read about the best results of the red filter , compared to blue . What worries me most is the difference in value Area between light and dark areas , and I believe , if I'm not mistaken , it is linked more to a factor of the type of lenses. Very interesting conversion lenses Mobius ! Thank you very much for the time that you gave me ! I read a lot about it but the language is still a major handicap.

Good morning, in these days , I modified a Canon PowerShot Elph 130IS . IR filter removed and replaced with Wattern #25 Red . I used the settings for Photo Monitoring Plugin , recommended to me by Cfastie. I post here

some results . Let me know what you think. Thank you!!

Good morning , no comment to My NDVI images ?

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

Hello @monoxyde good pictures there, I can see very well that in the shadow areas no NDVI its shown, good !!!!

Hello @monoxyde,

First of all, let me tell you, those pictures are great. You can actually see the difference between vegetation, soil & sky, which is really great.

I am really new at this, I'm a student doing a BSc thesis about remote sensing and forest fires, I know a lot about vegetation indices, drought and moisture indices, etc. So now I was trying to compare the remote sensing data with some photography and spectroscopy data, this is why I'm here. I already downloaded FIJI for ImageJ and Photo Monitoring plugin. I took some pictures with Mobius Infragram Point & Shoot Plant Cam, and I tried to do what you and @cfastie say about the settings. This is where I start having some problems.. first of all, the images I get doesnt have any green on vegetation (after applying NDVI settings in FIJI, see pictures below) I think it might be a calibration problem? The other thing I noticed is that I didn't get that NDVI scale you got on your pictures (the one that shows NDVI colors from white to pink, from -1 to +1). I know it might be some simple thing that im missing, but Ive really tried most of the options and nothing happened. I just wanted to know if you could tell me how to do it. Id be very grateful.

Thank you very much!

Greetings from Chile!

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

JorgeV,

The process is very sensitive to both the color balance and exposure of each photo. If the grass in your photos was a little less bright and a little more turquoise, the NDVI values would be higher.

It helps us to have an actual photo from your camera (maybe reduced in size) for testing. Just drag it into a research note or comment.

The color key has to be added manually after Fiji makes an NDVI image. You can download that key image here: http://publiclab.org/wiki/photo-monitoring-plugin. A quicker method is to select Image/Color/Show LUT in Fiji when your NDVI image is displayed then resize and drag the window to overlay the color gradient at the edge of the NDVI image. Then capture and crop a screenshot.

Chris

Jorge

Hello Jorge , First of all, Thank you and sorry for the delay in responding ! Unfortunately with the Mobius I only did a few tests . I decided then to switch to a Canon because I believe that the Mobius ( for this type of transformation ) has " large " limits in the lenses and the same number in the sensor . Besides the white balance always it requires a connection to a laptop or PC and this , in the field, can complicate life . For the rest , the answer to Chris , is definitely what you need to know ! If you need more details , please ask . In the coming days I should be able to publish some aerial map . Monoxyde

Hello to all this is a shot made ​​in full shade , in a day of strong sun and with a lot of contrast .

Hello @cfastie and @monoxyde,

I just realize you guys answered me. I really appreciate that you took the time to do it. I did understand how to use the .lut file, but Im still thinking theres something wrong with my pictures. Chris, I'll upload a picture, maybe it is a problem with the white balance and saturation of the lens? The truth is that I havent got enough time to read about it, I know I should do some kind of calibration, isnt it? I say that because when I work with satellite imagery I have to do a lot of stuff after working with spectral data of the image. I know this is someway different, but well.. im just new at this.

I'd just be very grateful if you can give me a quick advice on how to calibrate my camera... and @monoxyde, yes.. I've been thinking that mobius cam is just not good enough, but I just need some kind of reference for my study, If I get to go for a further or in a more specific study, I'd definitely go for a Cannon. I've also been talking to my research professor about getting the chance to use some kind of UAV with multispectral cameras on it. This is really interesting, I've been only using vegetation indices from .tiff data (satellite imagery), so this is quite new for me. But first! I need to understand this! haha...

Thank you, again, very much for your time @cfastie and @monoxyde, Have a great week!

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

Hello Jorge, in my humble opinion, the problem in this approach is in the difficulty to multispectral data to produce and then white balance settings that are reliable and repeatable. Added to this is another variable that is the correct setting of the plugin. If you try to follow the gigantic research Ned,

http://publiclab.org/profile/nedhorningabo

ut how to make correct and repeatable the WB, you will realize this.

These issues are also using a Canon PowerShot filter Red (who definitely prefer a blue one)