Not sure if this is the best stack exchange but I'll give it a shot.

I am performing background subtraction to collect Electroluminescent emission (i.e image w/ emission - image w/o emission = emission). Anyway, I am using a DSLR and collecting these images in .cr2 (canon raw format). Since the emission is weak, we wan't to make sure that we have the most data available to us and that it is as unprocessed as possible so we are converting the RAW images to .pgm (pre-bayer image data).

There is an issue, in some of the background that does not emit EL, for example a cabinet with drawers, when we perform background subtraction, we can detect all of the edges. Additionally, even if I try to do subtraction on 2 background images (both have no emission), the edges can still be clearly detected where we would expect simply black areas in the image.

Would anyone have insight on whether this is due to using .pgm, etc. Any suggestions at all?

Feel free to ask more questions about the project etc. Thanks

Below is what I get subtracting 2 images which should be essentially the same. I capture the raw image .cr2, convert it to .pgm using dcraw then subtract them using python numpy arrays and output a viewable png (what you see below.). Although, we would expect a black image(i.e subtracting two same/similar images), here I can detect all of the edges which makes no sense to me. Any suggestions?

Subtracting two background images

  • $\begingroup$ Can you put up some examples of your code and images? $\endgroup$ – geometrikal Jul 12 '15 at 5:52
  • $\begingroup$ @geometrikal Just added the image I am outputting after image subtraction of two background images. $\endgroup$ – bazzoli92 Jul 13 '15 at 22:50
  • $\begingroup$ Is this indoors? The flickering of the light may be causing it. Try an experiment outdoors. $\endgroup$ – geometrikal Jul 15 '15 at 0:05
  • $\begingroup$ @geometrikal This is indoors! I did suspect that, but I am using a 13 second exposure time so I didn't expect the 60Hz flickering of the lights to have much of an impact. I also tried this with the lights off and subtracted two images and I can still detect the edges and certain details on the solar panel. Might be useful to know, I removed the IR mirror on the DSLR and am using a 720nm lens filter to block out most of the visible light. I want to capture only IR light (specifically Electroluminescent emission). $\endgroup$ – bazzoli92 Jul 15 '15 at 14:33
  • $\begingroup$ You might have to have everything the same. ISO etc but I assume you've already done that. Is it always the second image that is brighter? $\endgroup$ – geometrikal Jul 16 '15 at 12:09

Your Answer

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

Browse other questions tagged or ask your own question.