Where Are Thermal Cameras Used?

Where are thermal cameras used? Perimeter security, dark and foggy environments, early warning scenarios, the radiometric measurement difference and limitations.
Technical summary
Technology
Thermal imaging, radiometric measurement
Topic
Perimeter security
Level
Start
In this article7 sections

A thermal camera is bought and mounted on the perimeter, and then the security team asks: “We see someone in the image, but we can’t tell who it is.” You can’t. A thermal camera looks at heat, not light; it tells you a target is there, not who it is. It’s a powerful detection tool in darkness, in backlight and in many conditions where visual cameras struggle, but if expectations are set wrong, the investment goes to the wrong place. Below is where thermal cameras are useful, where they aren’t and what to look at when choosing one.

How does a thermal camera work?

Every object emits infrared energy depending on its temperature. Security thermal cameras usually use detectors that sense this energy in the 8–14 µm long-wave infrared band and turn temperature differences into an image. Targets at a different temperature from their surroundings, such as people and vehicles, stand out clearly. Since no visible light is needed, it also works in total darkness.

Most security-grade thermal cameras use an uncooled microbolometer detector. These detectors periodically perform an internal calibration (NUC) to balance the image; on some models, the image freezes briefly during this. Knowing this behavior when setting up analytics rules reduces service tickets opened for “the image froze”.

Where are they used?

  • Long perimeters: Intrusion detection around factories, energy facilities and campuses. They don’t need lighting and, together with analytics, can reduce false alarms.
  • Dark open areas: Open stock yards, scrap yards, unlit land.
  • Fog, smoke and backlight: They can perform better than visual cameras; heavy rain and fog still have an effect.
  • Early warning: Detecting a target from a distance and steering a PTZ or visual camera to that point.
  • Radiometric monitoring: Monitoring temperature thresholds in electrical panels, storage areas or process equipment with models that can measure temperature. This is a different use from security and requires correct setting of calibration, emissivity and measurement conditions.

On a perimeter, the real gain from a thermal camera isn’t lighting cost but that analytics sees less “noise”. Rain, insects and shadows that glow under IR on a visual camera at night often don’t stand out from the scene temperature in a thermal image. A tree branch swaying in the wind, however, can also appear as motion in a thermal image if it has warmed in the sun; the virtual fence line should be drawn away from vegetation.

What are they not suitable for?

  • Identifying people: A thermal image has no face or clothing detail.
  • Reading plates: Plate characters can’t be read in a thermal image.
  • Behind glass: Standard glass blocks the thermal image.
  • Medical fever measurement: Security thermal cameras aren’t designed for this; body temperature screening requires special devices, calibration and protocols.

The glass issue surprises people on site. If you mount a thermal camera indoors looking out of a window, the image shows the temperature of the glass itself and reflections from the room instead of the outside area. Thermal cameras are mounted outdoors with a suitable housing.

What determines detection range?

Detection range depends on detector resolution (e.g. 384×288, 640×512), lens focal length, the target’s size and temperature difference from its surroundings, atmospheric conditions and the minimum target size the analytics needs. “Detection / recognition / identification” ranges based on the Johnson criteria are given in manufacturer brochures for ideal conditions. On a summer day, ground warmed by the sun can reduce the temperature difference between a person and the ground and make detection harder. That’s why on critical perimeters the design should be verified with pilot measurements.

Don’t use the detection range in the brochure as the design range. In the Johnson criteria, “detection” means having a few pixels on the target; for analytics to tell a person from a vehicle, more pixels are usually needed. The analytics manufacturer’s minimum target size is the real design limit. Place the camera according to that range, not the brochure value.

Mounting rigidity is also related to range. On a camera looking hundreds of meters away with a narrow lens, a pole swaying a few millimeters in the wind makes the target shift across a large area in the image and triggers motion analytics. For long-range thermal cameras, a sturdy, low-sway pole or wall mount is chosen instead of a thin lighting pole.

Performance also changes with the time of day. Around sunrise and sunset, ground and air temperatures converge and scene contrast drops; this is called thermal crossover. Doing pilot measurements only at night hides this weak window. Do test crossings separately before dawn, in the midday heat and at night; use the worst result as the design input. Heavy rain and snow also shorten detection range because they absorb and scatter infrared energy. A thermal camera isn’t independent of the weather.

Selection criteria

Criterion Explanation
Detector resolution Higher resolution gives a wider area or longer range with the same lens
Lens / field of view A narrow lens for long range, a wide lens for close and wide areas
Thermal sensitivity (NETD) The ability to distinguish small temperature differences
Analytics Virtual fence, direction and size filters; person/vehicle classification
Bi-spectrum Whether a visual channel is needed in the same housing
Radiometry Whether temperature measurement is needed
Medium IP protection, operating temperature, mounting rigidity

Thermal cameras may be subject to export controls in some countries; procurement conditions should be checked for high-resolution, high-frame-rate models.

A radiometric model is an extra cost if it isn’t needed for security. If there is a measurement need such as panel temperature monitoring, it’s planned separately: emissivity is set for the surface, measurement distance and angle are fixed, and the fact that shiny metal surfaces can give false values due to reflection is taken into account. Don’t expect measuring-instrument accuracy from a security camera.

Layered use

The most effective setup is layered: the thermal camera detects the intrusion, the analytics rule raises an alarm, a visual camera or PTZ records the detail, and a notification goes to the responsible team. This approach is the basis of our virtual fence and critical infrastructure perimeter security solutions.

For the layers to work, every link in the chain is tested: does the thermal alarm turn the PTZ to the right preset, does the PTZ have enough lighting at that point at night, is the alarm recorded in the VMS matched with the relevant visual camera, and who does the notification go to after hours? The common gap is the last link. An alarm is raised and nobody looks.

We assess suitability for your site with a pilot as part of our thermal camera service.

Frequently asked questions

Can a thermal camera see through walls or glass?

No. Thermal cameras don’t see behind opaque surfaces; since standard glass largely doesn’t transmit long-wave infrared, a thermal image can’t be taken through glass.

Are thermal camera recordings covered by KVKK?

Although they often don’t allow a person to be identified, a personal data assessment is needed when paired with a visual camera or where the person is identifiable; recordings should be managed with the same care as other cameras.

What is a bi-spectrum camera?

A camera with a thermal and a visual sensor in the same housing. The visual channel provides detail while the thermal channel detects; however, the two channels have different fields of view and levels of detail.

Support

Contact us for support or a discovery session on this topic.

Contact form
[email protected]

Eretek Video Surveillance

IP cameras, recording and central management infrastructure. We plan blind spots, retention periods and the camera network around the risks of each site.

Knowledge Center

Other articles