Public Lab Research note


Help about NDVI obtained with Mapir cameras.

by peppepegasus New Contributor |

Read more: publiclab.org/n/12848


Hi guys, it is the first time I write on Publiclab. With my group we are investigatiing on the use of Mapir camera. We have 2 cameras: red and nir. We want to build a NDVI map and we used the raster calculator of QGis. Can someone help me if the results are good or are bad?

Schermata_2016-03-15_alle_15.09.25.jpg

My principal doubt is about the sky: it seems to have a good vegetation index. Any idea? Any help is apreciated.


22 Comments

All of the NDVI values are very low. Could you also post the original photos and maybe we can see what kind of NDVI we get. Could you describe what filters are in the cameras and what settings were used to take the photos?

Thanks,
Chris

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


Hi,

these are the original images.

4.JPG

4A.JPG

Nir camera acquires in the region of near infrared, 808nm of lenghtwave (specifications: http://www.mapir.camera/collections/cameras/products/mapir-camera-near-infrared-light) and the red camera acquired at 650nm (specifications: http://www.mapir.camera/collections/cameras/products/mapir-camera-red).

Hoping these information are helpful. Thanks.


Those photos look good. I used the Photo monitoring plugin to make the NDVI image below. I used the blue channels from both photos because these provided good separation between the NIR and VIS channels. The color scale is from -1.0 (left) to +1.0 (right). NDVIPeg.JPG

A lot of the difference between your NDVI image and mine is the color gradient used.

Chris


Hi Chris, thank you for the support. I tried to install on my Mac Fiji/Image J (I suppose your image is obtained with Fiji) but when I installed (Plugins -> Install --> Photo Monitoring.jar or fiji-nojre-20141125_PM.zip as selected files) Photo monitoring plugin I received these errors:

/Applications/Fiji 2.app/plugins/Photo_Monitoring.jar: java.util.zip.ZipException: error in opening zip file

/Applications/Fiji 2.app/plugins/fiji-nojre-20141125_PM.zip: java.util.zip.ZipException: error in opening zip file

and if I restart Fiji in the plugin list I have not Photo monitoring plugin.

There is something wrong? I found an ImageJ version with Photo Monitoring included. If I select the Photo monitor plugin, I click on the voice Dual Image NDVI Processing, you can see a screen capture of the window,

Schermata_2016-03-16_alle_15.37.05.jpg

I can select only 1 image, it appears the window of the output directory and log file. It appears a small windows as in the figure

Schermata_2016-03-16_alle_15.44.45.jpg

I haven't any NDVI image on the output directory. Why? Thanks.

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


I'm not sure about your installation problems. Maybe @nedhorning can help.

I'm also not sure about the other error. To use two photos to make NDVI, a txt file is needed with paths to the photo pair. I'm not sure what happened there.

Chris


Hi peppepegasus -

I don't have a package for the Mac at this point since I don't have a Mac. You can try to follow these instructions from a previous note: https://publiclab.org/notes/nedhorning/01-13-2016/packaged-photo-monitoring-plugins-available-on-the-github-repositoy If you are able to get it to work and can create a package for Mac can you please let me know so I can post it on my GitHub site. Thanks - Ned


Thanks guys. Tomorrow I move on a windows virtual machine. The content of the log.txt file is:

PARAMETER SETTINGS: Use backup registration method if primary fails? true Select primary registration method: SIFT/Landmark correspondences Select secondary registration method: SIFT/Landmark correspondences using last valid set of points Select transformation type if using SIFT: Rigid Number of tries for SIFT to find correspondence points: 1 Method to improve SIFT point selection: nir=green band vis=green band Output image type: jpeg Output NRG image? true Clip images? true Output clipped visible image? true Output Color NDVI image? true Minimum NDVI value for scaling color NDVI image: -1.0 Maximum NDVI value for scaling color NDVI image: 1.0 Output floating point NDVI image? true Stretch the visible band before creating NDVI? true Stretch the NIR band before creating NDVI? true Saturation value for stretch: 2.0 Channel from visible image to use for Red band to create NDVI: 1 Channel from IR image to use for IR band to create NDVI: 3 Select output color table for color NDVI image: .DS_Store

PHOTO PAIR PROCESSING SETTINGS:

As said @cfastie at the end is required the photo pair but It is not any way to give this information.

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


Hi all, I'm sorry for my late reply. I installed Fiji on a Windows remote virtual machine. From Plugins menu I selected Photo monitoring and Dual Image Index Processing.

Schermata_2016-03-28_alle_18.02.48.jpg

From the window I can select only 1 pictures:

Schermata_2016-03-28_alle_18.03.32.jpg

after I can select only the log file and then apper a small window with an error:

Schermata_2016-03-28_alle_18.03.43.jpg

It is strange on windows I have difficulties as on Mac.


When you get to the step here:

PortraitOblique.jpg

Note that it is asking for a "Photo Pair File." This is made with the plugin called "Create dual image list."

It's not clear whether you have done that or not.

Chris


Hi,

I created the photo pair file from the menu as in figure:

Schermata_2016-04-02_alle_15.03.00.jpg

So I selected the Vis directory and Nir dir

Schermata_2016-04-02_alle_15.59.10.jpg

Schermata_2016-04-02_alle_15.58.43.jpg

Schermata_2016-04-02_alle_15.59.00.jpg

but I have these error and exceptions:

Schermata_2016-04-02_alle_16.18.02.jpg

Schermata_2016-04-02_alle_16.19.45.jpg

Mmh...some suggestion? Thanks.

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


Do you know the file format of your images? I appears that ImageJ can't read them. If you send me one pair of images and I'll see if I can find the problem.

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


Hi,

images are in jpg format. You can find 2 pictures in a previous my comment. These are other 2 pictures:

3A.jpg

3.jpg


Are those photos mosaics of more than one photo? What program did you use to combine the photos? That program might not be making jpegs that Fiji likes. The program might be altering the EXIF header somehow.

Chris

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


Hi Chris,

these photos are a mosaic of different pictures. The mosaic program used Is Microsoft ICE. These are 2 pictures not mosaic.

NIR_2016_0306_152422_065.JPG

RED_2016_0306_151932_028.JPG

Thanks.


Those photos of the cool old ruins work fine in Fiji, and the mosaic images from ICE also work in Fiji.

The Photo Monitoring plugins stopped working for me recently. This was apparently related to a problem referred to in this note: https://publiclab.org/notes/nedhorning/01-13-2016/packaged-photo-monitoring-plugins-available-on-the-github-repositoy. If you install the latest version of Fiji, the plugins do not work. So I followed the instructions and installed an older version of Fiji and the Photo Monitoring plugins work again.

The Mapir red camera produces a photo with three channels which record only red light. So any one of the channels (RGB) can be used for red light to compute NDVI. The Mapir NIR camera produces a photo with three channels which record only NIR light. So any one of the channels (RGB) can be used to compute NDVI. I chose the blue channel for red and the red channel for NIR because it gave the best result. This is rather unscientific, but any computation of NDVI without calibration is rather unscientific.

Capture.JPG


Hi,

I used the posted images to create NDVI map but I have some exceptions as in figure:

errors.jpg

this is the content of the log file:

LOG Geting SIFT correspondence points from 2016_0306_151932_028.JPG and 2016_0306_151932_028.JPG Number of times trying SIFT: 1 Processing SIFT ... took 2482ms. 1273 features extracted. Processing SIFT ... took 1613ms. 1273 features extracted. Identifying correspondence candidates using brute force ... took 370ms. 829 potentially corresponding features identified. Filtering correspondence candidates by geometric consensus ... took 250ms. 829 corresponding features with an average displacement of 0.000px identified. Estimated transformation model: 3,3 0.0 Registering 2016_0306_151932_028.JPG and 2016_0306_151932_028.JPG using Landmark correspondences

It seem the process is stopping on 25% of the elaboration (you can see it on the picture in the figure)..can you help me to understand what is the problem? Thanks!!

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


Good news people. It seems that the problem is represented by the compatibility of the versions of Fiji and Photo Monitoring. Installing the versions suggested by Ned Horning the program proceeds sent up to 100% processing. My doubt however is this: Photo Monitoring in the NDVI calculation, working with a formula in which there is the value of VIS and NIR. With our MAPIR cameras, we acquire pictures in the band of RED and NIR: Photo Monitoring properly perform the NDVI index processing? How to apply the caption of the LUT file? Thanks for collaboration Gianni.

2016_0306_151932_028.JPG

2016_0306_152422_065.JPG

2016_0306_151932_028_NDVI_Color.jpg

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


The Photo Monitoring plugin allows any band (R,G,B) to be used as the visible band (VIS). With the Mapir camera, all three bands from the RED camera capture only red light, so any one can be used. Although all three bands are red light, they are not always equally bright. So you can choose one which is much less bright than the NIR band from the other camera and the NDVI values will be in the right ballpark. This process is obviously very crude unless calibration is done.

The Photo Monitoring plugin does not display the lut with the NDVI image. This has to be done manually.

Chris


Hello Ned, Hello Chris, I'm reading the manual on the use of PhotoMonitoring, and specifically the section on "Calculate image calibration coefficients plugin". The guide talks about how to calculate the calibration coefficient of the image, using a single NIR-modified camera. In our case (with a group of friends including PeppePegasus et alt) We recently purchased 2 MAPIR (RED + NIR) and we would like to understand, if possible, how to proceed with the calibration and use of radiometric target. Would you be kind enough to show us some link or document where to study and investigate the calibration mode, and if it is possible to implement the plugin PhotoMonitoring so that it can calibrate the images from two cameras (RED + NIR)? Thanks so much

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


Hi Gianni, - We are working on a new python-based version of the plugs and that will have the ability to correct dual-photo setups like yours. It will also output the corrected photos in addition to NDVI. Hopefully there will be a version for testing a a few weeks. One way you can do it in ImageJ is to select the bands you want to use and calibrate from the two photos and create a new 3-band image to make an NDVI image. To do this in ImageJ use Image => Color => Split Channels to get the separate bands then use Image => Color => Merge Channels to select and merge the channels you want in the new image. You will loose the EXIF data but you can replace it using exiftool.


Hola, un gusto saludarles, han realizado TEST de calidad de Mapir3? y test con respecto los valores con una camara multiespectral como SEQUIOA o RedEdge? Gracias, Eddison

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


Hello, nice to say hello, have you done quality TEST of Mapir3? and test with respect to the values ​​with a multispectral camera such as SEQUIOA or RedEdge? Thanks, Eddison

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


You must be logged in to comment.

Public Lab is open for anyone and will always be free. By signing up you'll joined a diverse group of community researchers and tapped into a lot of grassroots expertise.

Sign up