Group Actions
Engagement
  • Author Author: fvan
  • Date Created: Date Created
  • Views 7222 views
  • Likes 3 likes
  • Comments 19 comments
Related
Recommended

Testing Pi NoIR with IR LEDs

fvan

When I got my Pi NoIR camera, I didn't have any IR LEDs to play with, so instead I used my TV's remote control for testing (Road Test review).

This was enough to demonstrate the functionality of the Pi NoIR, but it wasn't a permanent solution.

 

In the mean time, I managed to get my hands on a couple of IR LEDs.

 

The voltage drop across the IR LEDs I got is approx. 1,5V for a current rating of 60mA, meaning that with a 5V power supply I could put a resistor and 3 LEDs in series.

The resistor's value would have to be approx. 8 ohms (0,5V / 0,060A), but the closest resistor value I had lying around was 10 ohms.

 

I ended up with following circuit of two times three LEDs, in order to place three LEDs on each side of the camera board:

image

 

I downloaded a camera holder that I found on thingiverse (Raspberry Pi Camera Holder by gryphius - Thingiverse), modified it in Sketchup to be able to hold the LEDs and printed it.

imageimageimageimage

 

The circuit was powered with a separate 5V supply for this test, but ideally I'd like to power this from the pi itself.

Perhaps someone knows if it's ok to use the 5V/GND pins on the GPIOs to do this ? I'll have to look into this ...

 

With the LEDs powered, and the Pi NoIR running, it was time to take pictures and videos in the dark.

 

These are the results (at night, with all lights turned off except for power switches, etc ...):

 

raspistill -o file.jpg

imageimage

 

raspivid -o file.h264 -t 20000

 

 

The raspivid utility saves .h264 format files. Some programs need this in a .mp4 wrapper to use, which you can do in several ways. One is to use "MP4Box" which is in the 'gpac' package.

 

sudo apt-get update
sudo apt-get install gpac
MP4Box -add filename.h264 filename.mp4

(info from: http://elinux.org/Rpi_Camera_Module)

 

pi@raspberrypi ~ $ MP4Box -add testnoir2.h264 testnoir2.mp4

AVC-H264 import - frame size 1920 x 1080 at 25.000 FPS
AVC Import results: 594 samples - Slices: 10 I 584 P 0 B - 0 SEI - 10 IDR
Saving to testnoir2.mp4: 0.500 secs Interleaving

 

Using these IR LEDs, I was able to light the room up to 1.5 to 2m away with a limited field of vision.

 

Perhaps I should try some high power IR LEDs and compare the results, but that will be for next time image

  • in reply to Problemchild

    I'll be posting the images when I have them ready, just a bit more work on the rig to be able to tripod mount it all. some of the IR lighting kits look good on paper but you need to see some images taken using them in the manner you will be applying them as some end up being very poor in 'real life' conditions. cheers

  • in reply to Former Member

    I was wondering about the practicality of using the IR iluminators having 12-120ish LEDs that seem to be all over EBAY.

    They appear to be very bright and designed for the purpose. I would be interested in your experiments using various filters and how

    it effects the picture and or what you can see with the PiNoir

  • Good stuff, it ties in nicely with what I am up to just now. I've built a rig - almost - to hold the Pi and camera on a tripod as one unit and also different coloured filters in front of the lens. I'm trying to determine the best light transmission for night surveillance and also what special effects I can get taking colour images. B & W so far seems the best way to go for this type of work, at night, but I can see the use of colour during the day so may just rely on small colour temperature changes (Filters like 81A, 1B etc) to improve saturation and contrast. I hadn't yet gotten around to the provision of IR lighting and this looks like a very good start which seems easily scalable. Thanks for sharing this.

  • in reply to fvan

    Hi Frederick,

     

    I suggested looking at the LED output wavelengths as they can help.

    The real problem is the silicon sensor.  It is capable of catching near IR, but it is not optimized for it.

    You can improve your signal to noise in the IR band by trying to focus your LED output to the area where you want to detect things.

    Have you considered placing the LED's closer to the subject area?  The more light you can get on your target the better the camera can pick it up.

     

    Just a thought,

    DAB

  • in reply to fvan

    Frederick

    I used 12 (well 11 actually because one was faulty upon receipt ....and of course I only ordered 12), in three strings of 4 leds.

    That could be reconfigured into two strings of 6, as they are 1.5v.

    Send me a PM request and I'll send a link ...

     

    Selsinork

    You are quite right, but in this particular application it was designed to run from a 12v battery, hence keeping the consumption down was a consideration.

    This is about 4-5m away from the source.

    Its possible you don't see them pulsed (these show very slight red glow when on).

    We also need to remember that these are NOT special IR cameras, just a normal without the filter, hence they may be less sensitive at low light.

     

    John

    You're right, and I need to look at the specs for the duty cycle.

    The specs show a quantum leap in power out when you get it to 1A pulsed.

    The camera was on night mode (I used a still subject for this one) which was a bit slow, hence the need to up the output energy for the same overall power.

    It was also behind glass, and I haven't checked the spectral performance to see it passes IR very well.

     

    image

     

    I'm off to write up my challenge ....before I get added to the wrong list.

     

    Mark