Every DRI figure on a thermal datasheet is a calculation, not a measurement. It comes from pixel pitch, focal length and a set of assumptions about target size and atmospheric transmission. Once you can run that calculation yourself, you can tell which vendor claims are conservative and which are arithmetic optimism.
Key takeaways
- Detection range follows from three numbers: detector pitch, focal length, and the criterion you accept for detection.
- Change the criterion and the same hardware yields ranges that differ by more than an order of magnitude — this is why range claims are incomparable without their standard.
- NETD enters as a limit on thermal contrast, not on geometry; it decides whether the target is distinguishable, not whether it is resolvable.
- Model the mission first, then read datasheets against it — the arithmetic turns a marketing figure into a procurement decision.
On this page
The three numbers that set the answer
Instantaneous field of view (IFOV) is the angle one pixel subtends: pixel pitch divided by focal length. A 12 µm pixel behind a 25 mm lens gives 0.48 mrad. At 1,000 m that pixel covers 0.48 m of ground. This is the geometric limit and nothing downstream can beat it.
Johnson criteria convert a required number of line pairs across the target into a range. The classic values are 1.0 line pair for detection, 4.0 for recognition and 8.0 for identification, measured across the target’s critical dimension. Because one line pair needs two pixels, detection needs roughly 2 pixels across the target, recognition 8 and identification 16.
Critical dimension is the target’s minimum meaningful size. A standing person is conventionally 1.8 × 0.5 m with a critical dimension of 0.75 m; a light vehicle 4.2 × 1.8 m with 2.3 m. Change that assumption and every published range changes with it — which is precisely how two vendors quote different numbers for the same sensor.
Worked example
Take a 640 × 512 core with 12 µm pixels behind a 25 mm lens, looking for a person with a 0.75 m critical dimension.
| Step | Calculation | Result |
|---|---|---|
| IFOV | 12 µm ÷ 25 mm | 0.48 mrad |
| Detection (1 line pair ≈ 2 px) | 0.75 m ÷ (2 × 0.48 mrad) | ≈ 780 m |
| Recognition (4 lp ≈ 8 px) | 0.75 m ÷ (8 × 0.48 mrad) | ≈ 195 m |
| Identification (8 lp ≈ 16 px) | 0.75 m ÷ (16 × 0.48 mrad) | ≈ 98 m |
| Same core, 50 mm lens | IFOV halves to 0.24 mrad | Ranges double |
| Same lens, 17 µm pixels | IFOV rises to 0.68 mrad | Ranges fall ~29 % |
Where NETD enters
The geometry above assumes the target is visible at all. NETD — noise-equivalent temperature difference — sets whether there is enough thermal contrast to resolve it. A 50 mK detector needs roughly 50 mK of apparent target-to-background difference before the target emerges from noise.
In practice this matters most at the margins. A person on cool ground presents several kelvin of contrast and NETD is irrelevant. A person in warm water, a buried pipe signature, or a delaminated concrete patch presents a few hundred millikelvin, and there NETD is the whole ballgame. This is why we specify 40 mK or better for measurement applications and treat 50–60 mK as adequate for search.
Atmospheric transmission is the third eraser. Humidity, aerosol and path length all attenuate LWIR. A clear-day 780 m detection can become 300 m in haze without anything changing inside the payload.
Reading a datasheet critically
- Ask which criteria set was used. Johnson and EN 62676-4 produce very different numbers for the same sensor — often a factor of five to ten apart. See DRI ranges explained.
- Ask for the assumed target dimensions. A vendor quoting a 1.8 m critical dimension instead of 0.75 m more than doubles every range.
- Check whether the figure is for the thermal channel or the EO zoom channel. Mixing them is the most common datasheet sleight of hand.
- Confirm the atmospheric assumption. “Clear atmosphere” is a laboratory condition, not a Tuesday in November.
How this shows up in our payloads
We publish both Johnson and EN 62676-4 figures side by side on every product page precisely because the gap between them is large and meaningful. On OP-90D the Johnson detection figure for a person is 8,103 m and the EN 62676-4 figure is 709 m — same sensor, same optics, different question being asked.
For long-range work, LX-9B trades size and weight for focal length, which is the only variable in the IFOV equation you can move once the detector is chosen.
Related reading
Technology: how uncooled VOx microbolometers work and thermal infrared optics.
Field data: real DRI data from six payloads, thermal sensitivity (NETD) explained and why 12 µm replaced 17 µm.
Turning the maths into a procurement decision
The reason to run this calculation yourself is that it converts a vague requirement into a specification. “We need to see people at 500 m” is not procurable. “We need recognition of a 0.75 m critical dimension at 500 m, which requires an IFOV no worse than 0.19 mrad” is — and it immediately tells you that a 12 µm core needs at least 64 mm of focal length to get there.
Work backwards from the operational requirement in three steps. Decide the task level first: detection is enough for a search sweep, recognition is needed before you dispatch a crew, identification is needed before you make a legal or targeting decision. Then fix the target critical dimension from the actual thing you are looking for, not from a convenient standard. Finally solve for IFOV, and from IFOV for the focal length and pixel pitch combination that delivers it.
That process usually produces an uncomfortable answer, which is the point. Most disappointment with thermal payloads comes from buying to a headline detection figure and then discovering the mission actually required recognition — a four-fold range penalty that no amount of image processing recovers.
- Specify the task level (detection, recognition or identification) explicitly in the requirement.
- State the target critical dimension you are designing for, in metres.
- Derive the required IFOV, then check candidate payloads against it rather than against their marketing ranges.
- Add margin for atmosphere: assume real-world ranges of 50–70 % of the geometric figure in typical conditions.
- Verify with a field trial against a known target at a measured distance before committing to a fleet.
FAQ
Why do two vendors quote such different detection ranges for the same sensor?
Because the range depends on the criteria set and the assumed target size, not only on the hardware. Johnson criteria with a generous target dimension produce figures several times larger than EN 62676-4 with a conservative one. Always ask which standard and which target dimensions were used.
Does a lower NETD always mean longer detection range?
Only when contrast is the limiting factor. Against high-contrast targets such as a person on cool ground, geometry limits range and NETD makes no difference. Against low-contrast targets — warm water, buried pipes, delamination — NETD dominates.
Can digital zoom extend DRI range?
No. Digital zoom enlarges pixels without adding information; IFOV is fixed by pixel pitch and focal length. Only optical focal length or a finer pixel pitch changes the geometric limit.