Fundamentals

DIN EN 62676-4 explained: what the standard means for your camera plan

24 August 20263 min readcctvfinder

The call usually comes on a Monday. Something happened at the weekend, the recording is there, and on it you can make out a figure. That is all. No jacket, no face, nothing the police can work with. The camera did its job, it recorded, and the recording is still worthless. That is exactly the case DIN EN 62676-4 is meant to prevent. It turns what a system is supposed to achieve into a number you can work out beforehand.

The standard answers one single question

Nobody builds a system in order to get pictures. You build it because you want to know something specific: whether someone walked across the yard, whether that was your own employee, what the registration on the van was. Those are three very different questions, and they are not equally hard to answer. The standard makes that difference measurable.

Its instrument is pixel density: how many pixels land on one metre where the action happens? The unit is pixels per metre, px/m for short. It is the one quantity that pulls resolution, lens and distance together into a single statement.

A quality level belongs to the spot the camera is looking at, not to the camera. The same camera reaches “Identify” at five metres and “Detect” at forty.

The levels and what they demand

The standard knows six levels. Each one roughly doubles the demand of the one below it. The step from Recognise to Identify is the same factor of two as the step from Observe to Recognise.

Level	Pixel density	What it stands for
Inspect	1000 px/m	Read the smallest features: lettering, serial numbers, security marks.
Identify	250 px/m	Put a name to a person beyond doubt. The level for evidence.
Recognise	125 px/m	Recognise a person you already know.
Observe	62.5 px/m	Follow a person’s clothing, posture and actions.
Detect	25 px/m	Establish reliably that there is a person in the picture.
Monitor	12.5 px/m	Keep an overview of movement and numbers of people.

In practice two levels decide everything. “Recognise” at 125 px/m is what you need at gates, tills and entrances. “Identify” at 250 px/m is the level that holds up in court. It costs four times the pixels on the same area.

Why the slant distance counts, and the site plan does not

A common mistake in planning: the calculation uses the distance taken from the floor plan. But a camera rarely hangs at eye level. Put it at six metres, looking at a point ten metres away, and the real line of sight is around eleven metres. The pixel density drops accordingly.

The effect grows with mounting height. Mount high to see more area and you pay for it in pixel density at the target. Weighing those two against each other is the actual work of planning.

Mounting higher makes the picture bigger and the detail smaller. You cannot have both at once.

Why the level alone is not enough

The standard also recommends limiting the vertical viewing angle. The reason is easy to picture: look down steeply at a person and what you mostly see is the top of a head and a pair of shoulders. The pixels are there. The face is not.

Up to and including “Recognise”: no more than around 30° of tilt.
From “Identify” upwards: considerably flatter, around 15°.
Both are recommendations rather than limits. Anyone who exceeds them should be able to say why.

What this means for a tender

A requirement you can hold someone to names three things: the place, the task and the level. “Recognise at the gate, 125 px/m, at 18 metres” can be checked. “High-resolution camera for the entrance” cannot. Eight megapixels behind a wide-angle lens deliver less at 18 metres than two megapixels behind a telephoto lens.

That calculation is what the cctvfinder takes off your hands: enter the distance, the mounting height and the level you want, and the pixel density is there. Along with the picture the camera delivers at that spot.