DORI Calculation Walkthrough (2026): Step-by-Step EN 62676-4 Compliance for Real Projects

What you'll learn

Reading about DORI is fine. Calculating DORI on a real project is what actually wins bids and survives commissioning. This article is the hands-on counterpart to the conceptual DORI explainer. We pick three project archetypes that come up in roughly 80% of CCTV designs and we run the EN 62676-4 numbers through each of them, end-to-end, with the math visible. By the end you should be able to look at a camera datasheet, a lens choice and a target distance and know — within thirty seconds — which DORI level the design satisfies.

Why three examples and not one? Because each archetype stresses a different part of the calculation. The warehouse forces you to think about how DORI degrades with distance. The ANPR gate forces you to think about pixel headroom for plate readability. The bank teller forces you to think about face-recognition pixel density at close range with a constrained lens choice. Together they cover the calculation muscles you need for almost every design you will run into.

DORI math refresher — the formula in one paragraph

EN 62676-4 defines DORI in terms of pixels per metre at the target distance. The arithmetic is high-school geometry: the angular field of view of the lens projects onto a horizontal slice at the scene, and the camera's horizontal pixel count is distributed across that slice. The compact formula is:

Compare the result against the EN 62676-4 thresholds: 25 ppm Detection, 62 ppm Observation, 125 ppm Recognition, 250 ppm Identification. The 2025 amendment introduces OODPCVS modes that extend this vocabulary, but the underlying pixel arithmetic is identical.

Two practical notes before we hit the walkthroughs. First, sensor size is usually given as a fraction of an inch (1/2.8", 1/1.8", 1/1.2") — these are not literal inches but a legacy nomenclature from vidicon-tube cameras. Use the conversion table below or trust your calculator to handle the lookup. Second, the formula assumes a standard rectilinear lens. Wide-angle and fisheye lenses introduce barrel distortion that subtly changes the effective pixel density across the frame; for those, use the manufacturer's effective-focal-length figure rather than the nominal one.

Walkthrough 1 — Warehouse loading dock

The setup. A logistics customer is fitting CCTV to a 25m-wide loading dock with one camera mounted on the building wall directly above the dock door, looking out into the yard. The selected camera is 4MP with a 1/2.8" sensor (2560 × 1440, 5.376mm horizontal) on a 6mm fixed lens. The question: what DORI level does this camera achieve at 10m, 20m and 35m from the wall?

The interpretation. A single 4MP-on-6mm camera covers the dock with three different DORI levels at three different ranges: Identification right at the door, Recognition mid-yard, Observation deeper into the loading bay. For a 25m-wide dock, every pixel along the dock face satisfies Recognition or better, which is the right threshold for "we know who picked up which container".

The design conclusion. If the customer needs Identification at 20m (for example, to read forklift driver IDs at full yard range), this camera is not enough — they need either a longer focal length (which sacrifices the wide near-door view) or a higher-resolution sensor (8MP on the same 6mm lens reaches Identification past 20m). The walkthrough makes the trade-off visible in numbers, which is exactly the conversation you want to have with the customer before the install rather than after.

Walkthrough 2 — Parking lot ANPR gate

The setup. A commercial customer wants automatic number-plate recognition (ANPR) at the entrance gate of an underground car park. The barrier is 5m from the camera mount. The selected camera is 8MP with a 1/1.8" sensor (3840 × 2160, 7.20mm horizontal) on a 12mm fixed lens. The question: does this combination meet Identification (250 ppm) at the gate, and how much margin is there?

The interpretation. At 5m the camera delivers 1280 ppm — over five times the EN 62676-4 Identification threshold. That headroom is not wasted. ANPR works robustly only when the plate is well-lit, well-angled and free of motion blur — having pixel headroom means the system tolerates the inevitable real-world degradation (rain on the lens, low sun, a slight tilt of the vehicle) without dropping below the readability threshold. A design that just clears 250 ppm in lab conditions usually fails in production.

The design conclusion. The 8MP-on-12mm choice is well-matched to a 5m gate with significant margin. If the same camera were mounted further back — up to roughly 25m — it would still satisfy Identification but with much less environmental tolerance. For a single-purpose ANPR camera, generous pixel headroom is the right design call rather than a wasteful one.

Walkthrough 3 — Bank teller station

The setup. A retail bank is upgrading the in-branch CCTV at the teller counters. Each teller window has a dedicated camera mounted on the back wall, 4m from the customer side of the counter. The selected camera is 6MP with a 1/1.8" sensor (3072 × 2048, 7.20mm horizontal) on an 8mm fixed lens. The question: does this satisfy 250 ppm Identification at the customer side of the counter for face-recognition use?

The interpretation. The 6MP-on-8mm camera satisfies 250 ppm Identification at the teller counter with substantial pixel headroom. EN 62676-4 designates 250 ppm as the level at which an unfamiliar individual can be identified from the image — exactly the use case for face-recognition triggered by a flagged transaction. The same camera continues to satisfy Identification out to roughly 12m, which is more than the depth of any normal teller hall.

The design conclusion. At this geometry, mounting height and the angle to the customer's face matter more than additional pixels. A face viewed from a steep overhead angle is harder to identify than the same pixel count on a face viewed near-frontally — pixel density is a necessary condition, not a sufficient one. The walkthrough surfaces the right design conversation: focus the next iteration on mounting position rather than on camera spec.

EN 62676-4 thresholds reminder. The 250 ppm Identification threshold is the standardised boundary above which an unfamiliar individual can be reliably identified from the image. Local face-recognition algorithms commonly need substantially less to work, but the 250 ppm number is what survives evidentiary scrutiny in court — which is the bar a bank teller deployment is ultimately designed against.

Common mistakes

Three mistakes account for almost every "the math says yes but the install fails" conversation we have. Saving you the round-trip is a large part of why this article exists.

How to verify automatically

Doing the math by hand is a great way to learn it. Doing the math by hand on every camera in a 40-camera project is a great way to introduce errors. CCTVplanner ships two free, browser-based tools that wrap exactly the formula above with the EN 62676-4 thresholds and the 65,000+ camera catalogue.

For a multi-camera project, the designer is the right entry point. Place every camera on the floor plan, set the lens, and the canvas colour-codes which area satisfies which DORI level. Compliance flags surface in red anywhere a camera is asked to do more pixels per metre than its lens and resolution allow. The exported multi-page PDF carries the per-camera DORI level into the equipment table — the procurement officer reads the number you computed without having to recompute it.

For a deeper read on the standards-side update behind these tools, the EN 62676-4:2025 OODPCVS update article walks through the seven new mode labels that sit on top of classic DORI in 2026 EU procurement.

Frequently Asked Questions