Security & Surveillance
Camera Count for Area Calculator
How many cameras to cover a site perimeter
Updated October 13, 2026 · Live
What this tool does
Estimates how many identical cameras ring a rectangular site from its size, the horizontal lens angle, the coverage distance and an overlap allowance, and checks the pixel density at the far edge of each view against the IEC 62676-4 DORI levels.
Enter the site size, lens angle, coverage distance and overlap; the calculator returns the cameras for a continuous perimeter ring, the side-by-side count, and the pixel density at the far edge for 1080p and 4K sensors.
People also use
Formula Used
Formula Used
A camera set back from the fence by its coverage distance sees a run of fence as wide as the chord of its horizontal field of view at that distance. Lay those runs end to end around the site, let each overlap its neighbour so nobody slips between two views, and the number of cameras is the perimeter divided by the run each camera adds, rounded up. That is all this calculator does, and it shows the working so you can check it.
How the camera count for area calculator works
The width a camera sees at a distance d with a horizontal lens angle θ is 2d × tan(θ/2): two right-angled triangles with the camera at the apex. A 90° lens at 15 m sees 30 m of fence; at 50 ft it sees 100 ft. Take off the overlap, so 15% leaves 25.5 m (85 ft) of new fence per camera, and divide the perimeter, 2 × (length + width), by that run. The metric default site of 40 by 25 m has 130 m of perimeter and needs 130 ÷ 25.5 = 5.1, so six cameras; the US default of 130 by 80 ft has 420 ft and needs 420 ÷ 85 = 4.9, so five. Drop the overlap to zero and the metric site needs 130 ÷ 30 = 4.3, five cameras, which is one reason overlap is worth paying for: it costs a camera, not a blind spot.
The ring count treats the perimeter as one continuous line, so a camera may be credited with fence on both sides of a corner that it cannot actually see. The result also shows the count with each side taken separately, 2 × ⌈L ÷ run⌉ + 2 × ⌈W ÷ run⌉, which for the US default is six rather than five, and the larger figure is the safer one unless a corner camera really can look along both walls.
What you enter
The site length and width are the outside of the area you want ringed, in metres or feet. The lens angle is the horizontal field of view from the camera's datasheet; fixed-lens bullet and dome cameras mostly sit between 80° and 110°, and the slider runs from 20° for a long telephoto to 160° for a panoramic. The coverage distance is the hard one. It is not how far the camera can see, which is to the horizon, but how far away the fence can be while the picture is still good enough for the job, and that depends on the sensor and the lens together. IEC 62676-4, the international application guideline for video surveillance, grades image usefulness by pixel density on the target: 25 pixels per metre to detect that something is there, 62.5 to observe what it is doing, 125 to recognise a known person, 250 to identify a stranger (the DORI levels, reproduced in Axis's pixel-density white paper from the standard). A 1080p sensor spreads 1,920 pixels across the chord, so at 30 m of width it gives 64 px/m, enough to observe; a 4K sensor gives 128 px/m, enough to recognise. To identify a face at the far edge with 1080p and a 90° lens, the width can be no more than 1,920 ÷ 250 = 7.7 m, which puts the camera 3.8 m from the fence. The calculator works this pixel density out for the distance you enter, so you can see which level the far edge of each view actually reaches.
The UK Home Office's CCTV Operational Requirements Manual (publication 28/09) says the same thing in older units: a person should fill at least 10% of the screen height to be detected, 25% to be observed, 50% to be recognised and 100% to be identified, with a table of equivalents for digital resolutions. Either way, the operational requirement comes first and the distance follows from it. The overlap is the share of each view that repeats the neighbour's; 10 to 20% is usual, and the slider stops at 50%.
Reading the result
The headline is the number of identical cameras for a continuous ring at the coverage distance, with the rows underneath showing the view width, the run each camera adds, the perimeter, the side-by-side count and the pixel density at the far edge for 1080p and 4K sensors. It is a floor. Gates and doors want their own camera at identification density, which for 1080p and a 90° lens means within about 4 m (13 ft), blind corners and recesses want one each, and anything inside the ring, from a loading bay to a car park, is not covered at all. Where the density row says "detect" for the sensor you mean to buy, the ring will tell you that someone was there and little more; the fix is a narrower lens, a shorter distance (and more cameras), or a higher-resolution sensor.
Safe by design
This tool plans coverage. It says nothing about conductor size, PoE budgets, circuit protection or mounting loads, and the rules for recording people differ by country: in the UK and the EU a camera system that captures a public space or a neighbour's property engages data-protection law, and the UK Surveillance Camera Code of Practice applies to relevant authorities; in the US, NFPA 731 covers the installation of premises security systems and state law governs audio recording. The electrical side belongs to a qualified installer.
Assumptions and limits
The site is a rectangle with a clear line of sight along each side, every camera is the same model at the same set-back, and the fence is flat. Real sites have an L-shape, a building against one boundary, trees, parked vehicles, a slope and a gate, and a real design is usually a mix of wide lenses on the long runs and narrower ones at the points that matter, which this single-lens model cannot produce. Pixel density is computed at the chord, not at the far corner of the view, and it ignores lens distortion, low light and compression, each of which loses detail before the pixel count does. Treat the figure as the starting number for a site survey and a proper operational requirement.
Using it with the rest of BuildMetricLab
The camera count feeds the security system cost calculator for the equipment and installation budget, and the ring that the cameras watch is the same line the fence calculator prices. Every BuildMetricLab tool runs entirely in your browser, nothing you type is sent to a server, and the formula is shown on the page so the maths can be audited.
Sources & methodology
View width at the coverage distance = 2d × tan(θ/2); run per camera = view width × (1 − overlap); perimeter = 2 × (L + W); cameras = ⌈perimeter ÷ run⌉. The side-by-side count is 2 × ⌈L ÷ run⌉ + 2 × ⌈W ÷ run⌉, which does not credit a camera with fence on both sides of a corner. Pixel density at the far edge = sensor width in pixels ÷ view width, with 1,920 px for 1080p and 3,840 px for 4K, graded against the IEC 62676-4 operational levels of 25 px/m (detect), 62.5 (observe), 125 (recognise) and 250 (identify) as reproduced in Axis Communications' pixel-density white paper; the UK Home Office CCTV Operational Requirements Manual (28/09) states the same levels as 10%, 25%, 50% and 100% of screen height. Imperial inputs are converted at 0.3048 m per foot before the calculation. The model assumes a rectangular site, identical cameras set back by the coverage distance with a clear line of sight, and a lens angle below 180°; overlap must be below 100%.
- IEC 62676-4:2025, Video surveillance systems for use in security applications, Part 4: Application guidelines (IEC webstore; the standard behind the DORI pixel-density levels)
- Axis Communications white paper, Pixel density based on IEC 62676-4:2014 (detection 25 px/m, observation 62.5 or 63 px/m, recognition 125 px/m, identification 250 px/m; 16 cm face width)
- UK Home Office, CCTV Operational Requirements Manual 2009, publication 28/09 (monitor 5%, detect 10%, observe 25%, recognise 50%, identify 100% of screen height; Table 2 digital equivalents), archived by The National Archives
- NFPA 731, Standard for the Installation of Premises Security Systems (US installation standard)
- UK Government, Surveillance Camera Code of Practice (England and Wales)
Frequently asked questions
How many cameras do I need?
For a continuous ring at the coverage distance you enter: the perimeter divided by the run each camera adds, 2d × tan(θ/2) less the overlap, rounded up. The metric default site (40 by 25 m, 90° lenses 15 m back, 15% overlap) needs six; the US default (130 by 80 ft, 50 ft back) needs five. Then add a camera at each gate and door at identification density, one for each blind corner, and whatever the inside of the ring needs; the result also shows the count with each side taken separately, which is the safer planning number.
Why allow overlap?
Overlap between adjacent views stops the gap where someone could pass between two cameras, and it covers the edge of each image, where lens distortion and the fall-off in pixel density are worst. Ten to twenty percent is usual. It is not free: on the metric default site, going from no overlap to 15% takes the count from five cameras to six, because each camera adds 25.5 m of new fence instead of 30 m.
Does one camera cover a corner?
Only if it is mounted at the corner looking along both walls at once, which a single fixed lens rarely does well. The ring count credits a camera with fence on both sides of a corner, so the result also shows the count with each side rounded up separately (2 × ⌈L ÷ run⌉ + 2 × ⌈W ÷ run⌉). The larger figure is the safer basis, and a camera of its own at the corner also removes the blind spot directly below a wall-mounted unit.
Is perimeter cover enough?
No. It is a baseline for seeing that someone approached. The IEC 62676-4 pixel-density levels (25 px/m to detect, 62.5 to observe, 125 to recognise, 250 to identify) mean a 1080p camera with a 90° lens can only identify a face within about 3.8 m (12.6 ft) of the lens, so gates, doors, tills and anything else where the question is who rather than whether want dedicated cameras close up, and the inside of the site is not seen by the ring at all.
Spotted something off?
Calculations or display — let us know.













