Member Discussion

Beating Thermal Cameras

This may have been discussed before in the past, but it's been in recent news again.

A SITE Intelligence Group summary from back in 2011 said jihadist forums contained several discussions on ways to beat thermal cameras. Some of the strategies included wearing cold wet clothing, or getting on all fours so it was harder to distinguish a human from an animal, or making a suit of mud covered sacks to wear. The summary has since been taken down (or I'm just having a really hard time finding it online).

I don't think any of these would really be great at "beating" thermal cameras, but I'm wondering if they would be effective enough give someone difficulty making out an image. I talked to a couple law enforcement guys with experience with FLIR and got mixed answers. One said maybe. One said absolutely not.

If a human (say, in a desert) has exactly the same body temperature as the surrounding air, can a thermal camera 'see' them?

This was a very common question in my camera training. This would be viable if the surface of the person and the desert had the same thermal emissions which is not necessarily the same temperature due to emissivity differences.

That was a response from long ago. Marty is a lot more familiar with thermal now.

stone cold...

Emissivity is the first word they teach you at FLIR

D'oh. I fell for an old post again!

Didnt the show Mythbusters cover this in one episode? @Marty the issue is that the sand absorbs sunlight and then releases that heat so it can be upwards of 30 deg hotter than the air temp... Which for prolonged time periods would certainly prove harmful to any human

That sounds more like how to beat PIR sensors and not thermal cameras.

Most thermal cameras can "see" sub 1 degree temperature variations. There are certainly things you could do to minimize your thermal contrast, but it would take a lot of pre-planning and would be VERY dependant on the environment (assuming we're talking about outdoors here). A sudden wind gust could ruin your whole plan...

'Beating' is a strong term. Setting that expectation is likely unrealistic. However, you could shift the odds in your favor by studying the environment, picking opportune times and modifying form/dress. So, yes, they could 'be effective enough give someone difficulty making out an image.' Examples - the thermal performance when snowing report and our thermal performance in Hawaii daytime report.

A lot of this depends how close to the edge of maximum performance the thermal cameras are pushed. If they try to monitor a person a 1 pixel wide, even small changes in environment or subject camoflauging could cause an operator to miss (see our discussion on this in the detecton, recognition and identification tutorial).

Can a terrorist ensure a 0% chance of being detected by thermal? Unlikely. Can a clever terrorist drop the percentages from 80% to 40% through careful planning and disciplined execution? Probably.

I have certainly seen dramatic changes in the detection distances of video analytics with thermal cameras between midnight andd midday as temperature contrasts are less. Sometimes leaving big gaps in perimeter detection system.

"I have certainly seen dramatic changes in the detection distances of video analytics with thermal cameras between midnight andd midday as temperature contrasts are less."

This highlights a related point: are these cameras being viewed by people, or processed by analytics? Crouching down so as to present a profile more like an animal may be effective against analytics, but a human viewer would be far more likely to catch subtle cues that would tell them something is amiss.

Matt, but a human operator is far more likely to miss an object if they get tired or distracted. I think both approaches (human and analytics) have important flaws when dealing with adversary countermeasures.

Undisclosed, thanks for sharing that example. It certainly makes sense. I would suspect a similar issue would occur depending on the time of year as well (as overall temperature rise or fall, becoming closer or farther from human body temperature).

True enough, John.

A lot of dubious suggestions to beat thermal when all you need is a sheet of glass.

While AI would have an issue alarming one a sheet of glass hiding a human, seeing a square item move about in the scene would have likely been detected by simple motion detection.