I am using 2 raspberry Pis (one camera on each) and I synchronize the shutters using interrupts. I had some strange behavior (one Pi taking more picture than the other one) but it's negligible: 2 more pictures in a set of 200. My concern is more about image quality (quite poor with the VIS) and the reliability of the results. The Pi Noir seems good but I still have some doubt. Now that Carnaval has ended (everything stop during Carnaval in Brazil ... :( ) I will try to speak with the people of the university of agronomy and make some test with them. I'll continue to post the results of my test here. My next steps will be to check the calibration and make some test with the filter you mentioned. did you know if the Wratten 87 is included with the filter pack sold by public lab?

Do you know how well synchronized the photo pairs are? Do you have to start them both taking photos at the same time, or synchronize their clocks?

Wratten 87 is not included in the filter pack. If you can find somebody who once used color film, they might be able to find a black piece of film at the end of a roll. After development, color slide film is black if it has not been exposed to light, and color print film is black if it has been exposed to light. Either can be mostly opaque to visible light but pass most NIR.

Chris
The KAPtery

Hi!

I was in the filed without internet. I wrote a program that trigger the photo on the second pi when a picture is taken with the first so it is instantaneous. I read here that: "If you put the little blue filter (Rosco) over the lens (or behind it), then you can learn even more about plant health. The Rosco filter blocks most red light but passes the blue end of the spectrum and also near infrared light. The blue channel will now capture visible blue light as usual (plus some infrared) but there is no red light for the red channel to capture so it captures mostly near infrared light."

What is your point of view. I'll go back in the field tomorrow, so, no internet again for a few day.

That's a good description of what the Rosco blue filter does. Both the blue and the red channels will have a mixture of visible and NIR light. The blue channel will capture mostly blue, and the red channel will capture mostly NIR. So it does not produce the most pure NIR photo but it can work. A red filter generally produces a more pure NIR photo in the blue channel. Wratten 87 produces pure NIR photos in all three channels.

Chris

Hi,

Iam back with some update on the project. I didn't tweak the program but I brought the written 87 filter. and put everything on my phantom 2.

I made an othomosaic maps of the the picture obtained. Everything works fine except that Fiji do not accept the picture as they do not have the camera data in exif and when I tried to make the NDVI with photoshop everything is in gray.

Do you have any idea?

The good news is that I talked with a professor at the university and I may be able to calibrate the pinoir. the bad news is that he will only come back in 3 week :/

VIS

Hi Rayishido,

They are beautiful mosaics. I didn't have any problem using these images in Fiji. But the Photo Monitoring plugin does not do a very good job aligning mosaic images like these. Slight differences between the VIS and NIR mosaics don't allow the SIFT algorithm to rubber sheet the images enough to align them well. I did some manual alignment in Photoshop first and then it worked better in Fiji, but the registration is still not consistent across the image. You can see that best in the NRG image:

The same registration is used for the NDVI image, so the values it computes might not be very reliable:

It might be possible to get better registration with the options in the plugin.

I think somebody is working on a Python program that will combine the photo pairs into four channel TIFFs. Then those can be stitched together and the single mosaic can be used to compute NDVI. But I might have than plan wrong.

Chris