A technician checks fibre-optic connections beside a yellow test instrument.
Fibre-optic transmission is checked during work at the Gemini Observatory. Similar optical measurement principles are used in many scientific instruments. International Gemini Observatory/NOIRLab/NSF/AURA / Wikimedia Commons · CC BY 4.0. Image credit & source ↗

A thermal camera can turn an infrared image into a map of apparent temperature, but its detector does not feel heat or read temperature directly. It records radiation within a band of wavelengths, then software converts that signal into a temperature estimate using calibration and assumptions about the target and its surroundings. That conversion is why a polished metal surface can look misleadingly warm or cool, and why a sharp-looking color image is not automatically a precise measurement. The distinction matters in health settings: an infrared device measures radiation from a surface, while clinical meaning depends on the instrument, the measurement site, and its validated purpose. [3]

From infrared photons to a number

Every object above absolute zero emits electromagnetic radiation. At everyday temperatures, much of the emission is infrared, outside human vision. A thermal camera’s optics collect part of that radiation, and its detector converts the incoming energy into an electrical response. Calibration maps detector response to the radiation expected from reference sources at known temperatures; the displayed value is then calculated from that radiometric signal. [3]

This resembles a light meter: the signal depends both on the source and on how the surface reflects light. An infrared detector likewise receives energy emitted by the target, radiation from surroundings reflected by it, and sometimes radiation absorbed or emitted by air along the path. The camera separates these contributions through a model. [3]

Emissivity changes the mixture

Emissivity describes how strongly a surface emits radiation compared with an ideal blackbody at the same temperature. For an opaque surface, lower emissivity generally means more of the surrounding infrared field is reflected. A matte, high-emissivity surface is therefore easier to interpret than a shiny one: its measured radiation is dominated more by its own emission. A camera’s temperature calculation needs an emissivity setting appropriate to the material and measurement band. [3]

Consider a simplified opaque target with emissivity 0.98. In a first-order energy balance, about 98 percent of the surface-leaving signal is associated with its own emission and about 2 percent with reflected surroundings, before atmospheric effects. That 2 percent is a fraction of radiance, not a two-percent correction to the temperature. Because radiation and temperature are related nonlinearly and the surroundings may differ greatly from the target, even a small reflected contribution can matter in the right conditions. The example illustrates why settings describe the measurement model rather than changing the object itself. [3]

The scene and the camera both matter

A pixel is not necessarily a pinpoint thermometer. If the target is smaller than the instrument’s measurement spot, the detector combines radiation from the target and its background. A small warm feature can then be averaged with cooler surroundings. Image processing may interpolate pixels to make edges look smoother, but interpolation does not add spatial information the detector never resolved. Measurement resolution is set by the optics, detector, distance, and chosen analysis region. [3]

Calibration anchors the conversion

A camera’s calibration is a relationship between detector response and reference radiance. NIST describes radiation-thermometry standards disseminated through calibrated thermometers, lamps, and blackbodies, with traceability to the International System of Units. Variable-temperature blackbodies provide controlled reference radiation; traceability links the calibration chain to recognized standards. This makes the instrument’s radiometric response testable, but it cannot correct every mismatch between the calibration setup and a real scene. [1]

Calibration and correct use answer different questions. Calibration asks how the instrument responds to known radiation. Use in a particular scene also depends on emissivity, reflected apparent temperature, distance, atmosphere, focus, and whether the target fills the measurement area. A device can be calibrated and still give a poor estimate if its assumptions do not fit the measurement. Conversely, adjusting a color palette changes visual contrast, not the underlying calibration or uncertainty. [1]

Surface temperature is not a complete health measurement

Medical non-contact infrared thermometers are designed to estimate temperature at a surface site, not to directly sense temperature deep inside the body. FDA guidance notes that environment, forehead positioning, coverings, and use conditions can affect readings, and directs users to the instructions for the specific device. Thermal cameras can display many spatial samples, while a spot thermometer returns a measurement from a smaller field; both rely on infrared radiation and a conversion model. Their performance claims and intended uses are not automatically interchangeable. [2]

An infrared temperature map can show a pattern worth investigating, but it does not identify the pattern’s cause by itself. FDA specifically says thermal imaging is not an effective diagnostic device for COVID-19, even though a properly designed system can measure surface skin temperature. More broadly, a surface measurement answers only the question for which the device and method have been validated; it is not a personalized assessment or treatment recommendation. [4]

Read the image as a measurement

When interpreting a thermal image, separate three layers: what the detector recorded, how the instrument converted it to radiance or temperature, and what conclusion someone draws from the result. Ask whether the target fills the measurement spot, whether reflections or environmental conditions could matter, and whether the device is being used for its validated purpose. NIST calibration provides a reference chain; device instructions and measurement conditions connect that chain to the observed surface. A colored image is the start of that reasoning, not its endpoint. [1]

FOLLOW THE EVIDENCE

Sources & further reading

  1. Radiation Thermometry National Institute of Standards and Technology (NIST)
  2. Non-contact Infrared Thermometers U.S. Food and Drug Administration (FDA)
  3. Thermographic measurement techniques FLIR Systems, technical publication T810442
  4. FDA Alerts Public about Improper Use of Thermal Imaging Devices U.S. Food and Drug Administration (FDA), 4 March 2021

Sources checked: 15 November 2026. The calculations illustrate physical models; they are not clinical settings.

Prepared for the independent PSI Center reading library. About edition dates and source checking.

This explainer describes scientific principles. It does not assess individual treatment choices.