How Do You Photograph Heat? Everything Glows in the Dark

Any object above absolute zero radiates electromagnetic energy, and both the quantity and the wavelength depend on its temperature. Hot objects radiate more and at shorter wavelengths, which is why something heated enough glows visibly red and then white. Objects at ordinary temperatures radiate strongly in the infrared, at wavelengths several times longer than visible light, so they are invisible to the eye and perfectly visible to a sensor built for that range. The image produced therefore requires no illumination at all, since everything in the scene is its own source, which is the property that makes the technique useful in total darkness and through smoke.

Two quite different approaches are used and the difference matters commercially:

• •Cooled photon detectors, which respond to individual infrared photons
• •Those require cooling to cryogenic temperatures and are extremely sensitive
• •Uncooled microbolometers, which are tiny elements that warm when radiation falls on them
• •Their resistance changes with that warming and is measured electrically
• •Uncooled devices need no cooling, cost far less and are less sensitive
• •Nearly all commercial and consumer devices use the uncooled type

The picture shows radiation rather than temperature, and the difference causes constant misinterpretation. How strongly a surface radiates at a given temperature depends on the material, and a polished metal surface radiates very weakly while a matt painted one radiates strongly, so two objects at identical temperature appear completely different. Worse, a poorly radiating surface is a good reflector of infrared from elsewhere, so a shiny pipe frequently shows the reflected image of whatever is nearby rather than its own temperature. Glass is opaque to these wavelengths, so a window shows the glass rather than the room behind. Getting a true temperature therefore requires knowing the surface property and correcting for it, which is why serious measurement uses a patch of tape of known behaviour.

Several expectations about what the technique can do are wrong and they recur constantly. It does not see through walls, since a wall blocks infrared entirely and what appears is the temperature of the wall surface, which may be affected by what is behind it and which is a very indirect signal. It does not see through glass, through water or through most plastics. It cannot identify a person, since the image carries no detail of the kind faces are recognised by, though gait and outline can be distinguished. Resolution is far lower than in ordinary photography, with consumer devices offering a small fraction of the pixels of a phone camera. And it reports surface temperature only, with no information about what lies below.

The applications follow from seeing what the eye cannot rather than from seeing in the dark. Building surveys find missing insulation, cold bridges, damp and leaking pipes as patterns of temperature across a wall. Electrical inspection finds overheating connections before they fail, which is now routine preventive maintenance in industry. Veterinary and medical use includes detecting inflammation and assessing circulation. Firefighters locate people and find hot spots inside walls through smoke. Wildlife surveys count animals at night, including from aircraft. Manufacturing monitors processes continuously. And screening for elevated body temperature was deployed widely at borders during epidemics, with effectiveness that is genuinely disputed since skin temperature tracks core temperature poorly.

Everything above absolute zero radiates according to its temperature, strongly in the infrared at ordinary temperatures, so a sensor built for that range makes an image needing no illumination. Cooled detectors are extremely sensitive and expensive while uncooled ones dominate the market. The picture shows radiation rather than temperature, and polished metal radiates weakly and reflects strongly, which is the commonest source of misreading.