Public Lab Research note


  • 0

Filters and SJ4000

by thechapu |

Hello, my name is Cristofer writing from Chile. I'm interested in learning more about stress in plants.

Buy Red and Blue filter, and the mount in my SJ4000.

Take the first images but not much sunlight :(

What we still do not understand is that bands have my camera now that you place red filter is NGB or NRGB? and capture nm spectrum in that filter?

Stay tuned to comment.

Thank you so much.

índice.jpeg

índice1.jpeg

índice3.jpeg

índice5.jpeg

índice2.jpeg

índice4.jpeg



ndvi filter sj4000

9 Comments

Cristofer,

If you replaced the camera's IR block filter with a red filter (Rosco #19 or Wratten 25) The blue channel should capture mostly near infrared light. Healthy foliage reflects most NIR so the blue channel should be very bright when you photograph leaves. In your photos the blue channel is darker than the red channel. So that is not right. It might be due to the white balance setting of the camera. What was the white balance setting? Does the camera have the option for a custom white balance setting? See this page about setting the white balance: https://publiclab.org/wiki/infrablue-white-balance

histo.JPG

Chris


Hello cfastie,

Thank you very much for your answer.

I bought this pack of filters https://store.publiclab.org/collections/diy-infrared-photography/products/infragram-diy-filter-pack

and I selected the red filter.

When taking the photographs did not have any adjustment, calibration either, that must be the problem.

I must calibrate the camera to take more pictures.

One question, it is important the presence of sunlight?

regards


Yes sunlight is important. The light illuminating the plant should have about the same proportion of visible:NIR as sunlight or the results will be hard to interpret. Check to see what options the camera has for white balance settings.


20160725_153700.jpg

I reviewed my camera has 4 options for white balance, select the option 'sun' and these are the results

2014_0104_005148_009.JPG

2014_0104_005153_010.JPG

2014_0104_005145_008.JPG

2014_0104_005157_011.JPG

2014_0104_005200_012.JPG

2014_0104_005139_007.JPG

2014_0104_005203_013.JPG


Did you remove the camera's IR block filter before you installed the red filter?

If so, the blue channel in your photos should be a good NIR photo. It sort of looks like it is, but I'm not sure.

The red channel in your photos should be mixed NIR and red and could be used to represent visible light in computations of NDVI. But the red channel in your photos is overexposed so it does not have very good information in it.

The standard protocol is to set a custom white balance that exaggerates the brightness of the blue channel in every photo. Your camera apparently does not allow that. So it will be difficult to take photos that can be converted directly to NDVI or other plant health indices.

Chris


Hello, If I removed the fitro for IR camera lens

yesterday was a very sunny day in Santiago de Chile, take the occasion and take photographs changing the white balance. These are the results, I built histograms.

Balance = Auto

2014_0105_204956_006.JPG

2014_0105_204956_006.png

Balance = Sun

2014_0105_205021_007.JPG

2014_0105_205021_007.png

Balance = Clouds

2014_0105_205046_008.JPG

2014_0105_205046_008.png

Balance = Tungsten

2014_0105_205112_009.JPG

2014_0105_205112_009.png

Balance = Fruoro

2014_0105_205134_010.JPG

2014_0105_205134_010.png

Sorry, but my English is not very good


This is a great illustration that all of the standard white balance settings are very similar. The custom white balance setting used to produce photos that can be converted directly to a facsimile of NDVI is very extreme. For a Wratten 25 red filter good photos have a turquoise hue:

PetrW25hist.JPG

The histogram is for a small area of tree foliage (lower right). All of the pixels in that area have:

  1. a blue channel value higher than the red channel value
  2. no values near 0 (far left) or 255 (far right)

Below is an NDVI image for that stitched mosaic:

PetrW25NDVI.JPG

The NDVI values for foliage are mostly around 0.2 to 0.3. Those values are probably well below the appropriate values for NDVI because the original photos did not have a big enough difference between the blue and red channels, or because both channels were too bright. It's a bit of an art to take photos that can be converted directly to NDVI images with meaningful values. A camera which allows custom white balance settings is probably a requirement.

Chris


Thank you very much for your reply.

You use that camera? That software used to calculate NDVI and make histogra


The camera used to make that orthophoto image was a Canon PowerShot A590 from 2008. These are available on ebay for about US$30.00. It is a very good KAP camera but not that easy to convert to NIR. There is more about that flight here: https://publiclab.org/notes/cfastie/06-01-2014/super-redstone

The NDVI image and histogram were made with Fiji and the Photo Monitoring plugin.

Chris


You must be logged in to comment.