1

I have no idea how the sensors inside a camera work. I know I can take jpeg and separate the R, G and B channels but I have no idea whether these channels are the real red, blue and green in visible spectrum of light.

Why do I ask this:

Working with remote sensing data (i.e. Landsat 8 data) I can composite bands 2,3 and 4 (blue, green and red respectively) into a visible image. I can operate bands 4,5 into a NDVI index, I can do lots of combinations, image classifications, etc. I know there are many corrections to perform (DN to radiance, reflectance, atmospheric, etc.) but this is another topic.

So imagine I have one regular camera and one camera that captures only NIR (or some IR wavelength), could I separate the RGB values of the first camera and calculate the NDVI index (or others) with the NIR camera? Does this make any sense?

4

In short, yes. You can do that.

The sensors in most cameras are sensitive to light from UV to IR. To change the information into the standard RGB, most cameras use a Bayer Filter (see the Bayer Filter wikipedia for more info on how this is done) approach to filter the visible light into red, green and blue, while throwing away UV and IR information. As such, the red channel from your jpeg can be considered comparable to something you'd be used to seeing from Landsat-8 band 4.

You can also look into Full-spectrum photography as this can point you in the right direction for modifying your camera (or purchasing one that is designed for it) to allow you to acquire both IR and red at the same time, bypassing the issues with having to correlate the two images from different cameras.

  • Thank you for your reply Mikkel. I found this information very useful. I see you can take UV and IR pictures with some modifications of the camera but these are not minor, so you cannot change filters easily and take one Visible and one IR shot after another . However, I'm still not sure about how this is stored. I mean, if you take a picture with IR the IR data is kept in the R channel? Or it is all mixed up? I have the feeling this technique is very beautiful and artistic but I'm still not sure about the potentiality in vegetation studies. Nevertheless, thank you very much for your reply! – Albert Jun 12 '15 at 13:29
  • @AlbertC one option is to remove the Bayer Filter, and use lens filters that only allow a single band of data to pass through. That way you can first take a picture with a filter allowing only IR information, simulating band 5, and then change the filter to one optimized for red and simulate band 4. Then combining these two into a NDVI-ish result. – Mikkel Lydholm Rasmussen Jun 12 '15 at 13:38
1

You, in theory, can do this but I would question the reliability and replication of this approach. Most digital cameras are not calibrated so, it would be quite difficult to standardize the imagery to make them directly comparable through time and space. If you only intend to acquire a single image or are not planning on comparing data, there would be no issues with what you are proposing.

There are some specialized NGB cameras (eg., NGB Converted Canon SX260 HS) that block red light and record near infrared at 700-800nm. Otherwise you cannot derive NDVI from a standard digital camera.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.