Using The Raspberry Pi To See Like A Bee


The Raspberry Pi board camera has a twin brother known as the NoIR camera, a camera without an infrared blocking filter that allows anyone to take some shots of scenes illuminated with ‘invisible’ IR light, investigate the health of plants, and some other cool stuff. The sensor in this camera isn’t just sensitive to IR light – it goes the other way as well, allowing some investigations into the UV spectrum, and showing us what bees and other insects see.

The only problem with examining the UV spectrum with a small camera is that relatively, the camera is much more sensitive to visible and IR than it is to UV. To peer into this strange world, [Oliver] needed a UV pass filter, a filter that only allows UV light through.

By placing the filter between the still life and the camera, [Oliver] was able to shine a deep UV light source and capture the image of a flower in UV. The image above and to the right isn’t what the camera picked up, though – bees cannot see red, so the green channel was shifted to the red, the blue channel to the green, and the UV image was placed where the blue channel once was.


13 thoughts on “Using The Raspberry Pi To See Like A Bee

      1. Maybe it is possible? I do not know yet, but IF the uv sensitivity is high enough, one could possibly use a “filter stack” to only let certain wavelength pass. Such filters are not commercially available, but Dr. Klaus Schmitt did some work on that. Not sure, but I think he sells his XBV filters (bee vision, butterfly vision). Unfortunately I can’t afford to buy anything, so I can’t test it. Anyway, I expect the UV sensitivity to be much lower than for visible light, so best would be to use a filter stack or selfmade filters.
        In my video I only used a Baader U filter, but better would be to filter visible light, too. Because the peak wavelength of what we call green and blue is slightly different from the peak wavelength of a bee (for example). Butterflys are even more complicated, because they are tetrachromates.

    1. Right. Best lenses for UV photography are made of quartz and do not have a focus shift in UV.
      However, there are some cheap Nikkor lenses – made of plastic not glass – which let UV pass. They are the cheapest option to take UV photos with a DSLR. Works best with a modified DSLR. A camera tuner near my home town removes not only the IR cut filter but also the mosaic filter (R,G,B) and they modify the firmware, too. But they take a few hundred euros for that. But unless you don’t want to become an IR/UV professional, there is no need for that.

      If someone would like to get deeper into this interesting topic, here is a database of lenses:
      Here is described, what the camera tuner does (in German only, sorry):
      A forum regarding uv photography:

    1. Definetely true, I apologize for that! There are flowers with more remarkable patterns, but there IS a little difference. However, I just did a first test and a quick edit on a cloudy day with some flowers I found near my flat. I will certainly do more photographs with other flowers and maybe with the sun as the only UV light source. But I will go a step further then – stay tuned! ;)

  1. I wonder if the UV filters you guys are mentioning could be like the ones used for corona cameras and viewers? I know there is a guy somewhere that makes corona viewers for commercial use out of night vision scopes / photo multipiler tubes with UV filters added. You use these to “see” the corona from high voltage leakage /arcing which is in the UV spectrum.

  2. Hi!
    I would like to use a combination of a small quartz prism with a grating (perhaps DVD? Or standard grating slide…) to get a full spectrum.

    Then I’d like to view only one slit of a scene. The spectrum of each point along the e.g. vertical slit would be horizontally splitted in its spectral components, from nIR to a little bit of UV.
    Then I’d do a “video” while turning the apparatus round, by this being able to do 360° “Fotos” of my surrounding with one line of the ccd totally reserved for spectral information.
    For this ideally one would need a cooled down black-white CCD with no coatings as filter, and some UV permitting plastic lenses. have you sen plastic lenses for a pi noIR cam?

    I turn this topic in my mind because I do robotics in a school, with EV3 (Lego mindstorms), and we want to expand a bit, let a raspi “ride” the EV3, doing indoor navigation with “invisible” bar codes at the walls and furniture, so we can triangulate our position by previously known positions of the bar codes (relatively to each other).
    Later we want the bot to be able to paint the wall with stripes of invisible IR or UV absorbing ink, so we can explore an unknown room, travel straight to measure distances to the opposite wall, mark this, and so on.

    Much more later we want to be able to build a “one pixel Inkjet nozzle” to do OID codes or 2D codes on little objects with high resolution, for this we need a “macro cam” to read out the codes. By this we can do “internet of things” stuff without having a transponder on every object we want to handle in a smart way. We can assign even little lego parts some individual ID, if we like, and place helping marks to get a quick orientation info on the object.
    By some Zoom lens we would be able to recognize such patterns in 3D space, on walls, objects, etc., and some distance away.

    The last step would be to use the spectral information to get information on material, paint, etc., that would be sufficient to navigate with good certainty in unknown territory without even to mark the surroundings.

    Good luck!


Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.