As an Amazon Associate, Akermin earns from qualifying purchases. Product links on this page are affiliate links.
A thermal imaging camera is calibrated at manufacture against blackbody references, and that factory calibration is not something a user redoes. What you can and should do is set the parameters the camera uses to convert radiance into temperature: emissivity, reflected apparent temperature, distance, and relative humidity. On most handhelds these live in a measurement menu; on smartphone units they live in the app. If a reading is wrong after those are set correctly, the camera needs service, not a field adjustment. The instruments named here, such as the MiLESEEY TR20 and the FLIR One Pro, publish accuracy figures but no user calibration procedure, which is normal for the class.
What calibration means on a thermal imager, and what it does not
A thermal camera does not measure temperature directly. It measures infrared radiance arriving at the detector, then converts that radiance into a temperature using a model that assumes something about the surface it is looking at. That model is the calibration. At the factory the detector is characterised against blackbody sources at known temperatures, and the resulting curve is stored in the camera. Every reading you see is the camera solving that curve for the radiance it received, adjusted by the parameters you have entered.
This is why the word calibration gets used for two different things. Factory calibration is the characterisation of the sensor and its optics, and it is not user serviceable. User calibration, more accurately called parameter setting, is the business of telling the camera what it is looking at so the stored curve is applied correctly. The first is done once and periodically at a service centre. The second is done every time the target changes.
The listings here reflect that split. The MiLESEEY TR20 publishes an accuracy of plus or minus 2 percent, and the FOXWELL RT280 and the FLIR units publish no accuracy figure at all in the supplied bullets. None of them publishes a user calibration routine. That is not an omission specific to these products; it is how the category works. If a listing offered a field calibration procedure, that would be unusual and worth reading closely.
The four settings that actually change the number
Before you conclude the camera is out of calibration, check these four. They account for most of the wrong readings people bring to a service desk.
- Emissivity. A surface does not emit infrared perfectly unless it is a blackbody. Shiny metal might emit only a tenth of what a matte black surface does at the same temperature, and the camera will read it low unless you tell it so. The correct value depends on the material and finish, and the camera cannot know it. Setting emissivity is covered in more depth in how to set emissivity on a thermal camera.
- Reflected apparent temperature. Every surface reflects some infrared from its surroundings. In a room at normal temperature the reflection is modest. Looking at a busbar inside a warm enclosure, or at anything near a furnace, the reflection can dominate. You enter the temperature of the surroundings as the reflected temperature, and the camera subtracts it.
- Distance. The atmosphere between camera and target absorbs some infrared, and the camera compensates if it knows how far away the target is. For short-range work the effect is small. For outdoor or industrial work at tens of metres it is not.
- Relative humidity and air temperature. These feed the same atmospheric transmission model. Again, small at close range, meaningful at distance.
On a phone-based unit such as the FLIR One for Android these parameters are set in the app, not on the device. On a handheld like the FOXWELL RT280 they are in the camera menu. The principle is identical. If any of the four is wrong, the reading is wrong, and no factory calibration can rescue it.
The arithmetic of an accuracy specification
An accuracy figure of plus or minus 2 percent is usually expressed as plus or minus 2 percent of reading, or plus or minus 2 percent of reading plus a fixed offset in degrees. The listing for the MiLESEEY TR20 states plus or minus 2 percent without spelling out which form, so treat it as the reading-based form for planning and confirm with the manufacturer if the difference matters to your work.
What that means in practice is that the error grows with the temperature. At a reading of 20 degrees above ambient, 2 percent is a fraction of a degree. At a reading of 200 degrees, 2 percent is 4 degrees. This is the single most useful thing to understand about thermal accuracy: the same camera is more precise on cool targets than on hot ones, in absolute terms.
| Nominal reading | 2 percent of reading | What it means for a fault call |
|---|---|---|
| 30 C | 0.6 C | Fine for finding a warm joint; not enough to grade it |
| 80 C | 1.6 C | Useful for comparing like with like |
| 150 C | 3.0 C | Trend matters more than the absolute number |
| 400 C | 8.0 C | Absolute readings are rough; use differences |
The table is arithmetic, not product data. It shows why a camera that is perfectly in calibration can still give you a number you should not act on alone. If your decision depends on a 2 degree difference at 150 degrees, the instrument class is wrong for the job, not the calibration.
When the honest answer is that you cannot calibrate it yourself
If you have set emissivity, reflected temperature, distance and humidity correctly, and the camera still disagrees with a known reference by more than its stated accuracy, the camera needs service. There is no field adjustment on a sealed handheld imager. The detector is characterised at manufacture, the lens and shutter are part of the optical path, and the internal reference is not exposed to the user.
What you can do is verify. A blackbody calibrator is the proper tool, and it is a laboratory instrument. A cheaper check is a surface of known, stable temperature with known emissivity, measured with a contact probe at the same time, in a still room with no sunlight or nearby hot or cold surfaces. If the camera and the probe disagree by more than the combined uncertainty of the two instruments, you have a case for service. If they agree, the camera is doing its job and the earlier reading was a parameter problem.
This is also where the resolution question bites. A camera with a low detector resolution may be averaging a hot spot with its cooler surroundings inside one pixel, which reads as a lower temperature than the true peak. That is not a calibration error and cannot be corrected by adjustment. It is covered in how to compare thermal camera resolution. The FLIR One units use super-resolution processing to upscale the image, which improves the picture but does not change the underlying detector count.
What the listed cameras do and do not publish
The supplied bullets for these four products describe resolution, frame rate, field of view, display, storage and, in one case, accuracy. They do not describe a calibration procedure, a calibration certificate, or a recalibration interval. That is typical of consumer and prosumer listings. A camera sold for regulated inspection work would normally state a calibration interval and offer a traceable certificate; none of these listings does.
| Product | Stated accuracy | User calibration procedure | Recalibration interval |
|---|---|---|---|
| MiLESEEY TR20 | plus or minus 2 percent | Not stated in the listing bullets | Not stated in the listing bullets |
| FOXWELL RT280 | Not stated in the listing bullets | Not stated in the listing bullets | Not stated in the listing bullets |
| FLIR One for Android | Not stated in the listing bullets | Not stated in the listing bullets | Not stated in the listing bullets |
| FLIR One Pro for Android | Not stated in the listing bullets | Not stated in the listing bullets | Not stated in the listing bullets |
If traceable calibration is a requirement of your work, ask the seller for a certificate before purchase. A listing that does not mention one is unlikely to include one. The absence is information.
The maintenance that keeps a reading honest between calibrations
Between services, the things that degrade a thermal reading are mostly physical. A dirty lens scatters and absorbs infrared, which reads as a cooler target. A scratched lens does the same and cannot be polished out. A camera that has been stored in a hot vehicle may drift until it reaches thermal equilibrium, which is why the internal shutter flags periodically: it is re-referencing the detector against a known internal surface. Let the camera settle after a temperature change before you trust a reading.
The shutter is also your field indicator. When it clicks, the camera is correcting for detector drift. If the image visibly shifts at each click, the camera is working harder than it should, which can mean the detector is not stable at that ambient temperature. That is a service question, not a user adjustment.
Finally, the parameter settings themselves are a maintenance item. A camera left set to an emissivity of 0.95 from a previous job will read a shiny busbar wrongly on the next one. Get into the habit of checking the four settings at the start of each inspection, the way you would check a meter’s range. The explanation of how a thermal camera sees heat is worth reading if the emissivity point is not yet intuitive.
The short version
You cannot recalibrate a sealed thermal imager in the field, and none of the cameras here publishes a procedure for doing so. What you can do is set emissivity, reflected temperature, distance and humidity correctly, verify against a known reference, and send the camera for service if it still misses its stated accuracy. Most wrong readings are parameter errors, not calibration drift.
Frequently Asked Questions
Can I calibrate my thermal camera at home?
Not in the sense of adjusting the sensor. You can and should set the measurement parameters, which is what most people mean when they say calibrate. True recalibration requires a blackbody reference and access to the sensor characterisation, which is a service centre job. No procedure for it appears in the listings for any of the cameras named here.
How often does a thermal camera need recalibration?
The listings do not publish an interval. For regulated inspection work an annual check against a traceable reference is common practice, but that is industry convention, not something stated in these bullets. If your work requires a certificate, confirm the interval with the supplier before you buy.
Why does my camera read a shiny metal surface much colder than it is?
Because the surface has low emissivity and is reflecting its surroundings rather than emitting its own heat strongly. The camera assumes an emissivity you have set, and if that is wrong the conversion is wrong. Either set the correct low emissivity for the material, or apply a high-emissivity tape or paint to the surface and measure that. The second method is often more reliable in the field.
My camera agrees with a contact probe on a matte surface but not on a hot pipe. Is it out of calibration?
Probably not. A hot pipe is often shiny, curved and surrounded by cooler surfaces, so emissivity, reflected temperature and spot size all work against you. Check those three before suspecting the camera. If it agrees on a matte, flat, stable target at a similar temperature, the camera is likely fine.
Does super-resolution change the calibration?
It changes the image, not the radiometric calibration. The FLIR One units use processing to upscale the displayed image, and the underlying detector still determines the spatial detail available. A reading taken from an upscaled image is still subject to the same emissivity and spot-size limits. See how to compare thermal camera resolution for the detail.
Is a thermal camera the right tool if I need a certified temperature measurement?
Usually not. A thermal camera is a survey instrument: it finds where to look. For a certified number, use a contact probe or a calibrated reference thermometer at the point the camera identified. The camera tells you where the problem is; the contact instrument tells you how hot it is, traceably.