Resolution and pixel density (PPM/PPF, DORI)

Pixels on target, DORI per IEC 62676-4:2014, the 2025 edition's new levels, and how to calculate coverage.

⏱ ~40 minModule 4 of 11Quiz ↓Common mistakes ↓

🎯 Learning objectives

Calculate pixel density in px/m and px/ft
Apply the DORI levels from IEC 62676-4:2014
Know that IEC 62676-4:2025 replaced DORI with seven levels
Turn an operational requirement into a camera and lens choice

Pixels on target, not megapixels

The question is never "how many megapixels?" but "how many pixels land on the thing we need to see?" Pixel density is the number of horizontal pixels per metre (PPM, px/m) or per foot (PPF, px/ft) across the scene at the target distance.

Pixel density (px/m) = horizontal resolution (px) ÷ scene width at target (m)
px/ft = px/m × 0.3048

The scene width at the target distance sets pixel density. A narrower field of view at the same distance means more pixels per metre.

DORI levels (IEC 62676-4:2014)

IEC 62676-4:2014 (Video surveillance systems for use in security applications, Part 4: Application guidelines) defines four operational requirements for human operators viewing visible-light video. The pixel densities are based on a 16 cm wide human face:

DORI minimum pixel densities on the target.

Level	Pixel density	Pixels across a face	What the operator can do
Detection	25 px/m (≈8 px/ft)	4	Determine whether or not a person is present
Observation	62.5 px/m (≈19–20 px/ft)	10	See how many people and characteristic details such as distinctive clothing
Recognition	125 px/m (≈38–40 px/ft)	20	Determine whether an individual is the same as someone seen before
Identification	250 px/m (≈76–80 px/ft)	40	Identify an individual beyond reasonable doubt

These values apply to human operators viewing visible-light video. Analytics, LPR and thermal imaging use different requirements, set by the analytics or camera vendor.

IEC 62676-4:2025 changed the scale

Edition 2.0 of IEC 62676-4 (published October 2025) replaced the four DORI levels with seven image-detail levels: Overview 20, Outline 40, Discern 80, Perceive 125, Characterize 250, Validate 500 and Scrutinize 1,500 px/m. The same number no longer means the same thing: 250 px/m is "Identification" under 2014 but "Characterize" under 2025, where confident identification sits at the higher levels. When a specification quotes pixel density, confirm which edition it was written against. This site uses the familiar 2014 DORI terms in examples and flags where the 2025 edition matters.

Worked example

Can a 1080p camera identify people at a 4 m wide entrance?

Requirement: identification (2014 DORI) at a doorway; camera covers a 4 m wide scene at the door.

1920 px ÷ 4 m = 480 px/m. That is well above 250 px/m (identify). It would even meet the 2025 edition's 500 px/m Validate level only if the scene were narrowed slightly (1920 ÷ 3.84 m = 500 px/m).

Same camera covering an 8 m wide lobby: 1920 ÷ 8 = 240 px/m, just under 250. Options: narrow the FOV, use a higher-resolution camera (3840 ÷ 8 = 480 px/m), or add a second camera dedicated to the entrance.

Maximum scene width for identification at 250 px/m: 1920 ÷ 250 = 7.68 m for 1080p; 3840 ÷ 250 = 15.36 m for 4K.

Try it: pixel density calculator

Horizontal resolution (px) Scene width at the target Unit Required density (px/m)

Pixel density

Same in px/ft

2014 DORI level reached

Max scene width for chosen density

Aspect ratio and corridor mode

Most cameras are 16:9. In a long corridor, rotating the image 90° (often called corridor format) puts the larger pixel count along the corridor's height and length, wasting fewer pixels on walls.

Design practice

Write the requirement per view"Identify everyone entering through the main door" or "Detect vehicles in parking row C".
Pick the target zoneThe place where the requirement must be met, e.g. a 3 m wide zone 4 m inside the door.
Calculate densityChoose resolution and focal length that meet the density across that zone.
Check light and angleWDR, low light, shutter speed, mounting height and angle.
Validate on siteUse a test target or a person walking through; adjust zoom before signing off.

Common mistakes

Specifying by megapixels instead of pixel density at the target.
Calculating density at the camera rather than at the target distance.
Assuming a 4K camera identifies people across a huge parking lot.
Mixing 2014 DORI and 2025 IEC 62676-4 levels in the same specification.
Applying DORI to analytics, LPR or thermal, which have their own requirements.

Check your understanding

1.What is the pixel density of a 2560 px wide image covering a 10 m wide scene?

2560 ÷ 10 = 256 px/m, which meets 2014 DORI identification (250 px/m).

2.Which IEC 62676-4:2014 level corresponds to 125 px/m?

Recognition is 125 px/m (20 px across a 16 cm face).

3.What is the widest scene a 1920 px camera can cover while meeting 250 px/m?

1920 ÷ 250 = 7.68 m.

4.A spec written in 2026 asks for "Identification". What should you clarify?

Which edition of IEC 62676-4 it references. Under 2014 DORI, identification is 250 px/m; the 2025 edition uses seven different levels (e.g. Characterize 250, Validate 500, Scrutinize 1,500 px/m).

5.Why does a camera mounted high above a door often fail identification even with enough px/m?

The steep angle shows the top of heads and faces are hidden or foreshortened; pixel density alone is not enough without a shallow viewing angle and good lighting.

Progress is saved in this browser only.