Insight · 5 min read
How to specify CCTV so the footage is usable
Camera count and resolution describe what is bought. Pixel density at the target describes what the system can do.
The wrong metric
Tender documents often describe a CCTV system by camera count and resolution: "120 cameras, 4 MP minimum". That tells a contractor what to buy. It doesn't tell anyone what the system must achieve, so two compliant bids can deliver very different results.
The metric that matters is pixel density at the target: how many pixels fall across one metre of the scene where the person or vehicle will be. IEC 62676-4 links pixel density to tasks: detect (25 px/m), observe (62.5 px/m), recognise (125 px/m) and identify (250 px/m).
A worked example
A hotel wants to identify everyone entering through a 6 m wide main entrance. Identification needs 250 px/m. Across 6 m that is 1,500 horizontal pixels, so a 1080p camera (1,920 px) can do it if the lens is chosen so the field of view is no wider than about 7.7 m at the entrance line. A 4 MP camera (2,560 px) covering the whole 25 m lobby from the same spot would give about 100 px/m: enough to observe, not to identify.
The better design is usually two cameras: one narrow view at the doors for identification and one wide view for observation. It costs less than the very high-resolution camera that would be needed to do both from one position.
What reduces the real density
- Light. Backlit doorways and low night light reduce detail long before resolution runs out.
- Compression. Low bitrates smear fine detail, especially in moving scenes.
- Angle. A steep downward view shows the top of the head, whatever the pixel count.
- Motion. Slow shutter speeds at night blur moving people and vehicles.
- Dirt. Dust on the camera window degrades images steadily, which is why cleaning belongs in the maintenance plan.
So we design with margin above the minimum, and we check the critical views at night as well as by day before handover.
How to write it into a specification
For each critical position, the specification should state:
- The task (for example, identify persons entering through door D-01).
- The target area, marked on the drawing.
- The minimum pixel density at the target, by day and by night.
- Lighting conditions the view must cope with.
- How compliance will be tested at commissioning.
A test target or a person standing at the target line, measured on the recorded image, settles any dispute. That is what we do at handover.