# Question:Should I use two sheets of filter?

by kauemv2 |

Hello, i'm a brazilian student from UFPR university, i am new to the community but very enthusiast of NDVI imaging.

I'm working on a project to map the health of the urban vegetation at my city.

So I have build a platform for a car roof, this one:

In addition to a navigation system (Inertial + GNSS), and a Raspberry Pi 3, it haves a piNoir Camera v2.

So I went to a local store and I bought a sheet (50 x 60 cm) of the reccomended Rosco Supergel #19 (Fire), that is also on the DIY filter pack.

And I've perceived: putting 2 sheets (i have cutted in small pieces) in front of my eyes, the image become redder__, with only one sheet, the image are more orangy like.

So my doubt is: should I use more than 1 layer of rosco filter?

I have another question, (i will state here, but please let me know if is better to make another post):

From the list of presets of white balance for pi camera (PiCamera.AWB_MODES) there one most suitable one for NDVI imaging? (I supose "sunlight", because I will use it only in sunlight) Or there is a predefined tuple of "awb_gains" (red and blue gains) to use in this case?

Congratulations for all of this work, thank you in advance and sorry for my poor english!

The NDVI result will be the same whether you use one or more layers of filter. Using two layers reduces the amount of light reaching the sensor twice as much as using one layer and requires exposures that are longer (or wider aperture). There is no advantage to using two layers.

The preset awb_modes will not produce meaningful NDVI results. The sensor in this and other cameras is much less sensitive to NIR than to visible light. To take photos that capture the proportion of red light to NIR light in the actual scene, you can use a custom white balance setting that exaggerates the amount of NIR light arriving at the sensor. The Picamera does not allow custom white balance, but it does have awb_gains. Awb_gains can be used to exaggerate either the red or blue channel relative to the other. The blue channel will capture NIR (using a red filter), so you want to use awb_gains to amplify the blue channel. Don't forget to disable auto white balance by setting awb_modes to "off."

Maybe someone has found a good setting for awb_gains that works for NDVI. You are looking for a setting that will produce a photo of a healthy green plant leaf in sunlight with about three times more NIR (blue channel) than red (red channel).

Chris

So, can I ask for the question for awb_gains in another thread?

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

That's a good idea. Maybe someone has figured out good values for blue and red.

But, only one more question before I make that post: what if i just apply the proper weights (in post-process, at the time of ndvi calculation) to the blue and red band? (something like what you have suggested: 3 times more to blue...)

My camera is currently being used in a ROS (Robot Operating System) based system, and the current version of the driver does not support awb_gains =/

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