What Is a Thermal Camera and What Can It Reliably Detect?

A practical explanation of what an industrial thermal camera measures, why apparent temperature is not automatic truth, and how to build a defensible electrical-inspection workflow.

Last updated: 2026-09-18

Quick answer

A thermal camera makes an image from infrared radiation rather than visible light. In industrial inspection, it is used to reveal *apparent temperature patterns*: a warmer connection, a cooler blocked section, an uneven phase, a hot bearing housing, or a localized loss of heat. It is exceptionally useful for screening equipment while it is operating. It is not an X-ray camera, and a displayed temperature is only credible when the target, optics, measurement settings, and operating conditions are understood.

The right question is usually not "which camera has the highest pixel count?" It is "what decision must an operator make from this image?"

What the camera actually sees

Everything above absolute zero emits infrared radiation. A thermal camera converts part of that radiation into a visual image and, in a radiometric camera, preserves data that can be used for temperature measurement at individual image locations. The image is valuable because temperature differences often show a developing fault before it becomes visible, audible, or severe enough to trip a protection device.

For electrical maintenance, a technician commonly compares like with like: phases in the same switchgear, similar breakers under similar load, equivalent cable terminations, or the same asset on a later route. A hot lug may be meaningful because it differs from comparable components under a known load condition, not because a generic temperature threshold says it is hot.

What thermal imaging is genuinely good at

A well-planned inspection can help reveal:

High-resistance electrical connections, imbalance, overloaded components, and unusual load distribution.
Bearing, coupling, belt, gearbox, and motor-housing anomalies that create a recognizable heat pattern.
Heating-system and process losses where the surface condition can be observed.
Insulation discontinuities, moisture-related patterns, and building-envelope defects when the method and environmental conditions are suitable.
A repeatable before-and-after record after maintenance work.

The common thread is comparison and context. The operator needs access to the actual surface, enough target detail at the working distance, a stable operating condition, and a documented route or test method.

What it cannot determine by itself

A thermal camera does not see through opaque metal, cabinet doors, walls, or insulation. It sees radiation from the surface in front of it. It cannot independently identify root cause, prove compliance, or predict remaining life from one image. A warm electrical component may be overloaded, loose, oxidized, poorly cooled, exposed to reflected radiation, or simply operating differently from its neighbor. The image starts an investigation; it does not replace electrical testing, torque verification, load data, or process knowledge.

It is also important not to read a reflected object as the target. Shiny busbars, polished terminals, and low-emissivity metal can reflect radiation from the surroundings. In a poor setup, the camera can report the temperature of a nearby hot object reflected in a bright surface rather than the conductor itself.

Apparent temperature depends on measurement settings

Temperature measurement with infrared imaging is an inference. The camera calculates a result from detected radiation and declared parameters. Emissivity, reflected apparent temperature, object distance, atmospheric temperature, humidity, and, where relevant, infrared-window compensation influence the result. FLIR's measurement documentation treats these as explicit camera parameters for good reason.

For a qualitative inspection route, the priority may be repeatable imaging: same viewpoint, comparable loading, same lens, and a consistent palette or reporting template. For a measured temperature claim, the team must go further. Record the target material and surface state, emissivity assumption, reflected-temperature method, distance, window details, and the load condition. Where a high-reflectivity target matters, use the site's approved procedure or a suitable high-emissivity reference surface; do not invent a correction after the image has been taken.

Resolution and lens choice decide whether the target is usable

Thermal detector resolution does not by itself prove that a small connection will be visible. At a given distance, a wide lens includes more of the cabinet but allocates fewer pixels to each target. A narrower lens may resolve a distant terminal better, but it can make whole-panel framing harder.

Put a real geometry question in the RFQ: What is the smallest target to be distinguished, at the normal and maximum camera-to-target distance, through any safety screen or infrared window? Ask each supplier to show field of view and pixel coverage for the proposed lens at those distances. This simple exercise exposes many unsuitable quotes.

NETD, often described as thermal sensitivity, also needs context. Lower NETD can make weak thermal contrast easier to see, but it does not correct a bad focus, a reflective target, inadequate spatial detail, or changing load. Treat it as one part of the detection chain rather than an accuracy promise.

Focus, access, and the report workflow matter in the field

A camera may look capable on a specification sheet and still fail the route if an operator cannot focus through an infrared window, reach the needed viewpoint, or record enough information to make the image useful later. Confirm minimum focus distance, manual versus autofocus behavior, lens availability, scene temperature range, and the way visible and thermal images are paired.

The reporting workflow should preserve asset ID, date and time, operating load where available, measurement parameters, observed condition, and recommended action. A practical electrical-inspection report lets the maintenance team compare the same connection over time. A folder of unnamed JPEGs does not.

How to compare brands and models responsibly

Models from FLIR, Fluke, HIKMICRO, and other manufacturers can each be appropriate, but they should be compared against the inspection condition, not a brand reputation alone. Request sample radiometric images at your working distance, confirm export and report capability, and ask how calibration, service, lens options, and local support are handled. Include Chinese brands in the comparison when their data format, optics, warranty, and service route meet the stated requirement.

Turn the inspection route into a specification

Use the free AI Blueprint when you can describe the asset, access, operating state, target size, and required reporting outcome. Request an engineer review when a quote includes lenses or infrared windows, when target temperatures must be defensible, or when the inspection will support a formal maintenance decision.