Parking Lot and Car Park CCTV Systems

If a car park system only ever does two things well, they should be the entry lane and the exit lane. Everything else — a scrape in aisle 4, a bike taken from the rack, a confrontation by the stairs — is resolvable with adequate coverage, but a plate that was never captured cannot be recovered later. The lane position is therefore specified first, before the aisles are laid out.

The geometry is not complicated. Mount at 1.5–2.5 m, close to the lane, angled toward the approaching plate so the plate faces the lens as it approaches rather than receding. One camera per lane, dedicated. As a rule of thumb, keep the horizontal angle shallow — a plate viewed at 45 degrees loses digits — and keep the camera low enough that headlights do not flood the frame. Test with a real vehicle on dipped headlights, because the failure mode only appears at night.

Inside a car park the constraint is line of sight. Columns break the level into segments, and a camera mounted high in a corner sees the top of a van and the far wall — not the bay it is supposed to cover. The reliable approach is to treat the aisle, not the level, as the unit: mount cameras high, in line with the aisle, and cover each segment from one end. Where a segment is long, cover it from both ends rather than raising the camera, because raising it makes headlights worse and faces smaller.

A 5 MP frame is enough to identify a person and a vehicle in an aisle-width shot, and not much more than that. Spending the same sensor on a 60 m diagonal view of an open level produces a picture that shows where cars are and not who is with them. Deciding up front what each aisle position is for — vehicle movement, pedestrian identification, or both — is what keeps the channel count honest.

Unlit car parks are where the specification is really tested. The 5 MP full colour low-light series keeps colour in conditions down to 0.1 Lux without external floodlights and without switching on the camera’s own LEDs, so a level with only emergency lighting still yields vehicle colour and plate backgrounds instead of a silhouette. That also means you do not need to keep every level lit all night to keep the CCTV useful.

Two conditions still need to be designed for. Headlights, which are handled by geometry rather than settings: aim ramps downward, keep lane cameras out of the direct beam path, and avoid pointing any camera into the exit lane from the far end. And absolutely no light at all, where the camera’s built-in fill light maintains the colour image. If a level is genuinely unlit, put that in the design notes so the position is planned and tested for it rather than assumed.

The operator of a car park system is usually a booth attendant, a control room with several screens, or both. For a single surface lot or one level, a 4K monitor on the recorder output (HDMI up to 3840 × 2160, layouts of 1/2/4/6/8/9/10/16) with the lanes permanently visible is the right setup, and it is the one that gets checked every morning. For a site with several levels or several lots, the decoder server driving a 2×2 or 3×3 wall — 2 HDMI 2.0 4K@60 Hz outputs per card and up to 36 splits per card in a 14-slot chassis — keeps every entry lane and ramp on screen at once.

Whatever the display, the handover should include three things: a documented night test on each lane with a real vehicle, an export procedure the attendant can follow without an engineer, and channel names that match the signs on site — "Level 2 entry ramp", "surface lot north exit" — so a request for footage is a search rather than a translation.