skyemoor
Posts: 11
Joined: Wed Oct 16, 2013 1:21 pm

IR camera results?

Wed Dec 11, 2013 6:44 pm

Could those who have implemented a project with the IR camera share their results? Anything along the lines of how far away objects were clearly identifiable with a given IR illumination source type and level would be very helpful. Pics that show this would also be quite helpful.

skyemoor
Posts: 11
Joined: Wed Oct 16, 2013 1:21 pm

Re: IR camera results?

Mon Dec 16, 2013 8:42 pm

Any preliminary IR camera results to share?

User avatar
jbeale
Posts: 3898
Joined: Tue Nov 22, 2011 11:51 pm
Contact: Website

Re: IR camera results?

Mon Dec 16, 2013 11:35 pm

I just got a Pi No-IR camera module. I don't have any IR illuminators yet but I played around with the camera a little bit. I was surprised to see that in several indoor and outdoor lighting conditions, the colors actually look somewhat normal, just a little faded and yellowish.

For what it's worth, the cloudy-night-sky images in this post were taken with the NoIR module:
http://www.raspberrypi.org/phpBB3/viewt ... 00#p469701

with 1 second exposure time and ISO 1600 (I assume) and enough ambient city light pollution, you can get some kind of image even without your own lights. Of course a 1-second exposure is pretty useless if you want a security camera; anyone in motion would show up only as a pale ghostly blur, if that. Those photos were on an overcast night. The previous night was less cloudy and I did a timelapse sequence of 1-second ISO 1600 shots and the camera actually picked up a few of the brightest stars, and what appeared to be a small meteor trail, in amidst a lot of speckly noise. I haven't tried those exposure settings with the normal Pi-cam so you might see a few stars with that using the same settings; I'm not sure.

Andreas_Ulven
Posts: 1
Joined: Wed Nov 27, 2013 11:54 pm

Re: IR camera results?

Tue Dec 17, 2013 1:10 pm

I've been making a night-vision google with a NoIR cam and a pi. I'm using about 8 of these: http://www.adafruit.com/products/387 as illumination. They're driven by a 7.2 Ni-MH battery at about 100 mA. Haven't tried them outside, but works great indoors, I get at least 7 meters of illuminated hallway.

Bikeman
Posts: 65
Joined: Sun Nov 04, 2012 9:03 pm
Location: near Hannover, Germany

Re: IR camera results?

Tue Dec 17, 2013 5:54 pm

Here's an example image: a spy shot of my "cat room" in the attic (it used to be my room, but now the two cats have their favorite boxes, bed, and stuff in there :-) ).

The distance to the cat in the armchair to the right is approx 2.5 m. The shot was done in near complete darkness wrt visible light, illuminated by a single 4x1W LED IR bulb which conveniently fits in a GU-10 socket: (this one ==> http://www.synergy21.de/produkte/led/sh ... rity-line/ )

The attached photo is rescaled to satisfy the 64KB size limit here, but is otherwise unprocessed. The EXIF tag of the original image gives the exposure time as 0.6412 s.


Cheers
HB
Attachments
room2_small.jpg
room2_small.jpg (51.8 KiB) Viewed 7358 times

User avatar
jbeale
Posts: 3898
Joined: Tue Nov 22, 2011 11:51 pm
Contact: Website

Re: IR camera results?

Thu Dec 19, 2013 7:19 am

I got some small "bead" type 850nm IR LED (so-called "1 watt" type) from Ebay. It drops 1.7 V at 0.35 A so it uses 0.6 W. I put a 5 degree lens on it. These are the specific items:
1W 850nm LED: http://www.ebay.com/itm/350849956660
5 degree lens: http://www.ebay.com/itm/370526091475

Using one of these LED + lens pair, I tested it out in the backyard around 11 pm. I used an exposure of 1 second at ISO 1600 (you have to use -ex sports to get that, and wait 10 seconds for the AGC to adjust). I tried with the LED off and the LED on. Visible in the backyard are two small solar-powered garden lights, and the suburban skyglow above the trees. The camera and LED are near each other, and the distance from them to the back fence is about 7m or 21 feet. I would call this light barely visible, and I note that a 1 second exposure isn't much practical use if you want to see anything moving. For security-camera use with the Pi camera at that range, I'll take a guess that you probably want 10 to 100x as much power (6 to 60 W), depending on the field of view you need (=how wide you need to spread your light beam).

BY the way I see things on Ebay like a 12 W IR LED illuminator claiming 150 feet of range. http://www.ebay.com/itm/48LED-150feet-L ... 1135503336 I think that distance is *wildly* optimistic unless your target is wearing retro-reflectors, or your camera is vastly more sensitive than the R-Pi NoIR model. For any illuminator, light falls off as 1/r^2 so to double the range you need 4x the power.

The command was raspistill -ex sports -ss 1000000 -t 10000 -o test.jpg which give a 5 MPix image but at that light level it is so blotchy and noisy that the below rescaled 512 pixel wide image shows you pretty much all the detail there is.
LED-off.jpg
LED-off.jpg (15.63 KiB) Viewed 7211 times
LED-on.jpg
LED-on.jpg (16.81 KiB) Viewed 7211 times

skyemoor
Posts: 11
Joined: Wed Oct 16, 2013 1:21 pm

Re: IR camera results?

Thu Jan 02, 2014 5:07 pm

jbeale wrote;

> For security-camera use with the Pi camera at that range, I'll take a guess that you probably want 10 to 100x as much power (6 to 60 W), depending on the field of view you need

With the 1 second exposures you have, and the size and brightness of the area illuminated, I'd roughly estimate that 50 times and above would be needed for sufficient security monitoring, especially if video analytics were used (which is within the range you mention). Thanks for providing these results, the are quite helpful at estimating the performance of the IR camera in concert with various illumination approaches. Have you had a chance to find out the sensitivity of the camera to various IR wavelengths (e.g., 850 vs 940)?

Bikeman, the small 4W array you use at that distance gives a nice image at ~0.6 seconds, though it would need to be at least 4 times faster than that for a basic surveillance system that attempts to classify moving objects via OpenCV (e.g., "is that a cat, dog, or person walking around at 3am?"). Thanks for providing these results!

Andreas, any results to share yet?

skyemoor
Posts: 11
Joined: Wed Oct 16, 2013 1:21 pm

Re: IR camera results?

Tue Feb 25, 2014 6:44 pm

Any new IR trials to share over the last few months?

Peterleroux
Posts: 21
Joined: Fri Mar 29, 2013 8:51 pm
Location: Johannesburg, South Africa

Re: IR camera results?

Sun Mar 16, 2014 3:16 pm

image.jpg
image.jpg (37.74 KiB) Viewed 6312 times
Here is my garden in infrared only, courtesy of the poor man's visible light block/infrared pass filter- two layers of exposed,developed colour negative film. I'm putting together a multi-spectrum compact camera with swappable blue, infrared and natural light filters, so I haven't played with artificial IR illumination yet. Note the bright vegetation (leaves reflect IR very well to prevent overheating) and the dark sky (Raleigh scattering diffuses IR very strongly)

skyemoor
Posts: 11
Joined: Wed Oct 16, 2013 1:21 pm

Re: IR camera results?

Thu Apr 03, 2014 7:02 pm

Anyone know of the spectral response of the IR camera? Specifically, what sensitivity the camera has to various wavelengths, in a view something similar to this graph;
http://www.siliconimaging.com/Images/SI ... sponse.jpg

Image

darth_jan_smit
Posts: 2
Joined: Tue Oct 21, 2014 11:42 am

Re: IR camera results?

Tue Oct 21, 2014 11:54 am

Only 400-700nm but it could be useful http://www.imagesensors.org/Past%20Work ... n_bsi2.pdf

Return to “Camera board”