Mapping heat?

P!-Ro

I was just wondering, how is it done? I know Parallax sells ir heat sensors, however they are not capable of creating maps unless you move it around on a servo. So, how is it done only using a stationary sensor?

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
PG

Phil Pilgrim (PhiPi)

It's done with an IR sensor array and either an IR-transparent lens or reflective optics.

-Phil

P!-Ro

Searching for "IR sensor array", I found this:

http://www.melexis.com/News/Developing_an_IR_Sensor_Array_for_Multizone_Temperature_323.aspx

Unfortunately, it doesn't say anything about how it works, and when I tried going to the page for it I got an error message (404). Do any of you know a place where I can get more info?

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
PG

P!-Ro

Well, I did find this little thing:
http://www.robotshop.us/devantech-8-pixel-thermal-array-sensor.html
But I must say I am somewhat disappointed by the price. If any of you come across one which is cheaper, let me know. I could really use one of these in my project.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
PG

SRLM

You could probably use a digital camera, with the input spectrum translated to IR. Then analyze the images that you get back, and go from there.

Phil Pilgrim (PhiPi)

SRLM said...
You could probably use a digital camera, with the input spectrum translated to IR.

M'kaaay. Sounds neat in principle, but how do you propose doing that? Optical sensors make terrible long-wave IR sensors, and vice-versa. Moreover, long-wave IR will not pass through regular optical glass. Lenses for IR tend to be expensive, being specially fabricated from barium fluoride and other exotic materials. There's a reason such cameras are expensive.

-Phil

SRLM

Okay, here is my thoughts:

Main source: http://www.atsf.co.uk/ilight/tech/thermal.html

Most consumer cameras can see infrared light of one kind or another. To get the heat of an object, we use the emitted infrared light, as opposed to the reflected infrared most IR photos use. Therefore, we have to filter out those wavelengths of IR that are reflected (near-infrared), so you'll have to get a filter for that. You'll have to also filter out visible and UV light too. What's left is far-infrared (after the filters). Since the regular lens of a camera will block far-infrared, you get rid of it. Instead, you use a camera obscura with a pinhole. To increase the number of items that you can reliably photograph, you can chill the camera close to it's minim operating temperature. A special outdoor camera would be good for this, as would probably a security camera that's made to use infrared for night time.

Granted, this system wouldn't work as well as a custom system. In particular, it would only detect hot objects or no objects (hot being some relative term, maybe a soldering iron). I assume a consumer digital camera can see the required wavelength for hot objects based on these two sets of photos:

http://www.flickr.com/photos/jatherton/2772731563/in/set-72157602207040035/
http://www.flickr.com/photos/ciordia/335124406/

Phil Pilgrim (PhiPi)

SRLM said...
Most consumer cameras can see infrared light of one kind or another.

Yes, near IR, corresponding to the silicon response curve, which is pretty much drops to zilch above 1100nm.

The wavelenngths Pi Guy is talking about are much higher and require thermopile detector elements, not photodiodes.

-Phil

SRLM

From the photos, can't I assume that the camera can see the correct wavelength? It can see the flame just fine (and assuming there's nothing special about the camera), it should be able to see other sources provided that they are hot enough. Especially the photo of the bust on the first page: it uses two soldering irons to light up the bust.

Phil Pilgrim (PhiPi)

The image of the bust was made with film, not a digital camera. Flames emit visible light as well as near IR, so of course they will register with a digital camera. But this is not the same as thermal imaging. For example, a candle flame is completely invisible to Parallax's heat sensor, since it's designed for the same longer wavelengths that Pi Guy is interested in.

-Phil

Bean

Yeah, I have trouble explaining this to people who think because a digital camera can detect IR that it works like a thermal camera. It does not.

IR as in a remote control is just outside the visable spectrum. But thermal IR is ALOT longer wavelength. Not even the same animal.

Bean.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
There is a fine line between arrogance and confidence. Make sure you don't cross it...

·

P!-Ro

From your own link (first one):

"Clark explains that infrared photography can see hot bodies at temperatures from about 250 C up to the point where they begin to glow visibly at about 500-600 C, depending on how dark it is and how much your eyes have adjusted to the darkness."

This won't work because my main application for it is detecting people. That's why I wanted the images to·be mapped, so I could ensure it was truly a person, not a·fire or a piece of metal sitting out in the sun.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
PG

LilDi

The human body's thermal IR emission peaks at around 9.5um. The body also emits near IR, but not far IR of any intensity to easily detect. The human eye can see light in the range of 0.4um to 0.7um (visible light). Thermal temperature is a function of thermal IR intensity, not the wavelength of the thermal IR within the range of 3.0um to 14um. Far IR is in the range of 14um to 1000um. Normal cameras see light in the visible light range, but if you remove the IR filter from the lens, most CMOS camera sensors can see in the near IR range of around 0.7um to 1.3um (photographic IR). If you remove the IR filter from the camera lens and replace it with a piece of fully exposed and developed Ektacrome slide film (black in appearance), the camera will see near IR with visible light filtered out by the Ektacrome film. Thermal IR cameras cost thousands of dollars because of the exotic materials used in the sensors and lenses as pointed out by Phil. Hope that clears thing up.

Post Edited (LilDi) : 4/22/2009 4:20:43 PM GMT

P!-Ro

Yes it does, thank you LilDi.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
PG

Leon

I used to work for this company:

www.irisys.co.uk

They make low-cost thermal imagers (under $1,000 IIRC). They are selling like hot cakes, because of the swine flu scare.

I worked on the prototype, which used their 16x16 sensor array with a rotating chopper and a germanium lens.

Leon

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle

Post Edited (Leon) : 5/4/2009 4:16:34 AM GMT

P!-Ro

The price is scaring me Leon....Maybe something in the couple hundred $$$ range?

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
PG

Leon

The Irisys cameras are the cheapest available, AFAIK, at £1,250. Farnell stocks them.

Leon

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle

P!-Ro

Thanks Leon, maybe I'll save up for a couple years and get them (think minimum wage).

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
PG

Phil Pilgrim (PhiPi)

Here is a possible low-cost contender. It was mentioned in the May 2009 Photonics Spectra magazine.

-Phil

P!-Ro

I'm still thinking it might be a little pricey. When watching one of their vids, they didn't comment on the camera's actual price; however the cheapest price they had for their other models was said to be "under $20,000."

Maybe I should just stick with that 100 buck array.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
PG