What is DORI?
DORI is the four-level scale for how much detail a security camera records at a distance. Detection, Observation, Recognition, and Identification each need a minimum pixel density, measured in pixels per metre (px/m). The further the target, the fewer pixels cover it.
DORI comes from the standard EN 62676-4 and answers the installer's question: "How far away can this camera still identify a face?" A camera that detects a person at 96 m may only identify them at 9.6 m — identification needs ten times the pixel density of detection. The four levels are the planning method: pick the level you need, then work out the distance and the lens.
- Pixel density at this distance
- Person (1.8 m) tall in pixels
- Identification — 250 px/m or more
- Recognition — 125 px/m or more
- Observation — 62.5 px/m or more
- Detection — 25 px/m or more
Worked-example camera: 8 MP (3,840 px), 4 mm lens, 1/2-inch sensor (6.4 mm wide). It identifies to 9.6 m, recognises to 19.2 m, observes to 38.4 m and detects to 96.0 m.
Detection: 25 pixels per metre
Detection needs 25 pixels per metre — enough to tell that something is there. A person 1.8 m tall covers 45 pixels at detection range. You know a person is present, but not who they are or what they are doing.
Detection is the cheapest level and the widest coverage: a 4 mm lens on a 1/3-inch sensor watches a 24.0 m wide scene at 20 m. Use it for perimeters, car parks, and open ground, where the job is to spot an intruder and send a guard — not to name the person.
Distance from the camera
- Pixel density
- Person (1.8 m) tall in pixels
- Frame width
Worked-example camera: 8 MP, 4 mm lens, 6.4 mm sensor. At the 96 m detection limit the density is 25 px/m and a person is only 45 pixels tall.
Observation: 62.5 pixels per metre
Observation needs 62.5 pixels per metre — enough to tell what you are looking at. A person covers about 112 pixels, so you can see it is a person, roughly what they are wearing, and which way they are moving. You cannot recognise them.
Observation is the level for general monitoring. The worked example camera (4 mm lens, 8 MP, 6.4 mm sensor) observes out to 38.4 m. Some sources print 62 or 63 px/m instead of 62.5; the difference is rounding, and it moves the distance by about 1% — 38.4 m becomes 38.7 m or 38.1 m. Nothing about the lens choice changes, so this site uses 62.5 px/m throughout.
- DORI distance
- Person (1.8 m) tall in pixels
- DORI distance
- Person (1.8 m) tall in pixels
Same camera, same person: detection reaches 96.0 m with a 45-pixel blob; observation reaches 38.4 m with about 112 pixels — a recognisable human shape.
Recognition: 125 pixels per metre
Recognition needs 125 pixels per metre — enough to recognise a person you know, or a vehicle's make and colour. A person 1.8 m tall covers 225 pixels. You can say "that is the delivery driver", but the image is not strong enough to identify a stranger in court.
Recognition is the level for entrances, reception areas, and loading bays. The worked example camera recognises out to 19.2 m. Doubling the focal length doubles the recognition distance: an 8 mm lens on the same camera recognises out to 38.4 m.
Focal length — same 8 MP camera
- Recognition distance (125 px/m)
- Person tall in pixels at recognition
- Identification distance (250 px/m)
Doubling the focal length doubles the recognition distance: 19.2 m at 4 mm, 28.8 m at 6 mm, 38.4 m at 8 mm, 57.6 m at 12 mm.
Identification: 250 pixels per metre
Identification needs 250 pixels per metre — enough to identify a stranger for evidence. A face about 0.16 m wide covers 40 pixels across. This is the level for entrances, tills, and anywhere an identification may end up in court.
Identification is the hardest level and the shortest distance: the worked example camera identifies only out to 9.6 m. That surprises most buyers, and it is the most common CCTV planning mistake: a wide-angle camera over a car park detects everyone and identifies nobody. If identification is the job, the camera must be close, the lens long, or the resolution high.
Camera and lens
- Identification distance (250 px/m)
- Detection distance (25 px/m)
- Face wide in pixels at the identification limit
Doubling the pixels doubles the distance: 12 mm at 1,920 px identifies to 14.4 m; at 3,840 px it identifies to 28.8 m. A face is 40 pixels across at the identification limit.
How to calculate a DORI distance
DORI distance = (focal length × horizontal pixels) ÷ (required px/m × sensor width). For the worked example — 4 mm lens, 3,840 horizontal pixels, 6.4 mm sensor — the identification distance is (4 × 3,840) ÷ (250 × 6.4) = 15,360 ÷ 1,600 = 9.6 m.
The same formula gives every level — change only the required px/m: Detection 15,360 ÷ (25 × 6.4) = 96.0 m; Observation 15,360 ÷ (62.5 × 6.4) = 38.4 m; Recognition 15,360 ÷ (125 × 6.4) = 19.2 m.
- Pixel density at this distance
- Detection distance
- Observation distance
- Recognition distance
- Identification distance
Worked example: 4 mm lens, 3,840 px, 6.4 mm sensor. Detection 96.0 m, observation 38.4 m, recognition 19.2 m, identification 9.6 m.
- Pixel density at 20 m
- Detection distance
- Observation distance
- Recognition distance
- Identification distance
Same camera, longer lens: at 6 mm the density at 20 m is 225.9 px/m — recognition, just short of identification.
Choosing a CCTV lens from DORI
Start from the level you need and the distance, then solve for the focal length. To identify a face at 36 m with an 8 MP camera (3,840 px) on a 1/2.8-inch sensor (5.1 mm wide), you need a 12 mm lens: (250 × 5.1 × 36) ÷ 3,840 = 12.0 mm.
For a room, a 2.8 mm lens on a 1/2.8-inch sensor gives an 84.6-degree horizontal field and covers 18.2 m wide at 10 m — detection everywhere, identification only close up. For a gate 30 m away, a 12 mm varifocal lens is the starting point. For identification at 50 m, the maths says 16.6 mm on the same camera — so a 12–25 mm varifocal covers it.
DORI level needed (target 30 m away, 8 MP camera, 1/2.8-inch sensor)
- Focal length required
- Horizontal field of that lens
- Focal length required
- Horizontal field of that lens
8 MP camera (3,840 px) on a 1/2.8-inch sensor (5.1 mm). Identify at 30 m needs 10.0 mm — nearest common lens 12 mm. Identify at 36 m needs 12.0 mm; at 50 m it needs 16.6 mm, so a 12–25 mm varifocal covers it.
Blind spot, mounting height, and the 2025 change
A camera mounted high and tilted down cannot see the ground right below it — that is the blind spot. Mounting at 3 m and aiming at face height 1.5 m leaves a blind zone of a few metres at the wall.
The blind spot is the second most common CCTV mistake after the identification-distance error: a camera at 3.5 m with a 15-degree tilt and a 4 mm lens on a 1/3-inch sensor sees a wide scene but misses anyone standing directly under it.
- Distance to the first visible ground point
- Blind spot covers a doorway 3 m out — mount lower or tilt less
4 mm lens on a 1/3-inch sensor: vertical field 48.5°, so the lower frame edge sits 24.2° below the optical axis. First visible ground = height ÷ tan(tilt + 24.2°).
- Distance to the first visible ground point
- Blind spot covers a doorway 3 m out — mount lower or tilt less
Same camera. More tilt moves the first visible ground point closer — but the blind zone under the camera is the price: at 3.5 m and 15° tilt it is 4.29 m.
All calculators
- Focal lengthf = D · d / W
- DistanceD = W · f / d
- Crop factorCF = 43.27 / diagonal
- CompareAOV = 2 · arctan(d / (2f))
- CCTVD = (f · px) / (PPM · sensor width)
- TelescopeTFOV = AFOV / M
- BinocularsExit pupil = D / M
- Gaming FOVH = 2 · arctan(tan(V / 2) × aspect ratio)
- TV distanceDistance = width / (2 · tan(θ / 2))
- From a photoW = D · d / f
- Projector throwThrow distance = throw ratio × screen width
- MicroscopeField diameter = FN / objective magnification
No calculator matches your search.