Standards

DIN EN IEC 62676-4:2026-10: Seven Quality Levels from Overview to Scrutinise

19 October 20263 min readcctvfinder

DIN EN IEC 62676-4:2026-10 has applied since 18 September 2026. It replaces the six quality levels of the 2016-07 edition with seven pixel densities from 20 to 1,500 px/m and limits the camera tilt angle to 13°. cctvfinder now calculates against this edition. Here is what changed and what it means for your planning.

What has changed

The 2016-07 edition worked with six quality levels. The new edition has seven, described as pixel densities in pixels per metre: 20, 40, 80, 125, 250, 500 and 1,500 px/m. Each level has a name and a typical application example.

Level	Pixel density	Group	Typical application example
Overview	≥ 20 px/m	LPDO	Recognise whether objects or people are moving in an area; general overview of a site
Outline	≥ 40 px/m	LPDO	Recognise the outline of an object and determine its direction of movement, for example people or vehicles on an access road
Discern	≥ 80 px/m	LPDO	Recognise objects and their movements and tell people, vehicles and animals apart
Perceive	≥ 125 px/m	HPDO	Perceive people and movements in more detail, without being able to reliably tell, for example, gender or distinguishing features
Characterise	≥ 250 px/m	HPDO	Assess type of person, gait and behaviour, or determine vehicle type and category
Validate	≥ 500 px/m	HPDO	Verify known persons, follow actions or capture a vehicle licence plate
Scrutinise	≥ 1,500 px/m	HPDO	Assess a person’s identity with high certainty, recognise vehicle model or year, or read a licence plate

The first letters of the seven level names give the abbreviation OODPCVS. Older labels such as “Recognise” and “Identify” no longer exist in this form.

LPDO and HPDO: two groups with different purposes

The seven levels are split into two groups. LPDO stands for Low Pixel Density Objects and covers Overview (20), Outline (40) and Discern (80 px/m). They suit large-area and perimeter surveillance, where the question is movement, direction and kind of object, not detail.

HPDO stands for High Pixel Density Objects and covers Perceive (125), Characterise (250), Validate (500) and Scrutinise (1,500 px/m). Here the image needs considerably more detail, for example to assess people or vehicles more closely.

Up to 80 px/m you see that something is there and where it is going. From 125 px/m the assessment of detail begins.

The meaning of the thresholds has shifted

If you know the old edition, do not read the table from memory. At 250 px/m it now says “Characterise”: assess type of person, gait and behaviour. Assessing a person’s identity with high certainty only begins at 1,500 px/m (“Scrutinise”). Verifying known persons or capturing a licence plate belongs to 500 px/m (“Validate”).

For running projects this means: a specification that demands 250 px/m for identification should be checked against the new edition. Which edition applies to a project is a matter for the contract and the client. This article gives no legal advice.

How pixel density follows from resolution, focal length and distance is covered in the article explaining the standard’s levels.

Tilt angle of at most 13°

The new edition specifies that the tilt angle of the surveillance camera must not exceed 13°. In the 2016-07 edition the limit was 22.5°. A camera looking steeply down at a nearby point quickly exceeds 13°, especially with a high mounting position and a short distance.

Whether the angle is met can be calculated before installation. It depends on mounting height, target height and distance.

How cctvfinder reflects the new standard

Under “Surveillance goal” the seven levels are listed one below the other, from Overview (20 px/m) to Scrutinise (1,500 px/m).
The three LPDO levels are tinted in a soft, rising orange, the four HPDO levels in a soft, rising green. The groups are also labelled, so colour is never the only cue.
If the tilt angle exceeds 13°, a warning appears. The pixel density result is not affected by it, so you see both separately.
The camera selection rates every model against the chosen level, including the new values up to 1,500 px/m.

Try it with your own scene: choose a level, place the camera, read the result.