Security camera blind spots: how to find them

A security camera blind spot is any area that matters where no camera captures enough detail to use: no view at all, a view blocked by something in the way, or a view too far away to identify a person. In a commercial CCTV fleet of 100 to 500 cameras, most blind spots are not left out at design time. They appear later, as shelving goes up, cameras are re-aimed, models are swapped, and units fail, and they are usually found on the day someone asks for footage that does not exist.

This post covers where blind spots come from, how to audit for them across a site with a few hundred Axis or Hanwha cameras, and how to keep finding them after the audit is done.

Five places blind spots hide

Blind spot	How it forms	How to find it
Under the camera	A high mount tilted down cannot see the floor directly in front of it	Work out the near edge from mount height, tilt, and vertical field of view
Beyond useful range	The camera sees the area, but with too few pixels to identify anyone	Check pixel density at the distance where people actually stand
Behind an obstruction	Racking, signage, a new wall, a grown tree, or a parked trailer blocks the view	Compare the current view with the view at install
Between cameras	Two views meet short of a door or corner and leave a strip neither covers	Plot every camera’s view on one scaled floor plan
A camera that is down	Offline, black image, frozen image, or knocked off its aim	Watch device health continuously, not at the quarterly walk

The first four are geometry. The fifth is time: any camera that stops working turns everything it covered into a blind spot until someone notices.

Under the camera: the dead zone

A camera only sees what falls inside its vertical field of view. Mount it high and tilt it down, and the floor directly in front of the mount sits below the bottom edge of the image. A 1080p camera with a 90-degree lens has a vertical field of view of about 59 degrees. Mounted 12 feet up and tilted 25 degrees down, the bottom edge of its view points about 54 degrees below horizontal and first meets the floor 8.6 feet out. Anyone who walks along the wall under that camera is not on video.

The fix is placement: a second camera across the room looking back at the first, a lower mount, or a steeper tilt that trades distance for the near field. Entrances matter most here, because the camera above a door is often the one that cannot see the person standing in it.

Beyond useful range

Seeing an area and being useful there are different tests. IEC 62676-4 defines four levels by pixel density on the target: detect at 25 pixels per meter, observe at 63, recognize at 125, and identify at 250. A 1920-pixel camera with a 90-degree lens reaches identify at 12.6 feet, recognize at 25.2, observe at 50, and detect at 126. A face at 60 feet is on video and cannot be identified.

This is the most common blind spot in a fleet that appears to cover everything: the parking lot camera that sees every car and no plate, the warehouse dome that sees a person and no face. List the places where identification matters (entrances, cash handling, loading docks, gates) and check each one against the camera that covers it, at the distance a person actually stands. The worked distances and mounting rules are in security camera placement for commercial sites.

Behind an obstruction

Most obstructions arrive after install: racking, inventory stacked high, seasonal displays, signage, a new interior wall, a tree that was a sapling on handover day, a trailer parked in the same spot every night. None of them changes the camera, so nothing about the camera reports a problem. The view is smaller than it was, and nobody is told.

Two habits catch it. Compare each camera’s current view with a reference image from install, and draw new obstructions onto the floor plan so the coverage math matches the building. A sudden change, a sign hung in front of the lens or a camera knocked sideways, is a tamper event, and camera tamper detection covers how that is told apart from ordinary scene change. A slow change, a tree growing over a season, needs the reference comparison.

Between cameras

Two cameras that each look correct can leave a gap where their views should meet. It shows up at corners, at the far end of a corridor, and at doors set between two cameras that each watch half a wall. Views overlap near the cameras and spread apart with distance, so the gap tends to sit at the door, where it matters most.

The reliable way to find these gaps is to put every camera on the same floor plan, drawn to scale, with its real model and real aim. A spreadsheet of camera names cannot show the gap. A per-camera live view cannot show it either, because each view looks complete on its own.

The camera that is down

Every camera that stops working creates a blind spot the size of its view. The causes are ordinary: power and PoE faults, network changes, a failed storage card, an encoder that hangs and keeps serving a frozen frame, a lens knocked off aim. Why security cameras go offline walks through them. What makes this blind spot expensive is how long it lasts. A camera found dark on a quarterly walk-through may have been dark for most of the quarter, and every incident in that window has no footage.

Two points matter at a few hundred cameras. A camera that answers on the network is not necessarily producing a usable image; a frozen or black image passes most simple checks. And overlap is the backstop: where two cameras cover a critical area, one failure leaves the area covered at a lower level instead of not at all.

How to audit blind spots across a fleet

A practical audit for a site with 100 to 500 cameras:

1. List what must be seen, and at what level. Entrances at identify, cash handling at recognize, general floor at observe, perimeter at detect. Write the required level down per area. Every camera gets checked against this list.
2. Get a current floor plan and put every camera on it. Real position, real model, real aim. The install drawing shows the building on handover day, which is not the building today.
3. Draw what blocks the view. Walls, racking, glass, and anything taller than a person.
4. Check each critical area against the cameras that reach it. At the required level, at the distance a person actually stands. Mark it pass or fail.
5. Check under every high mount and every door camera. Near-field dead zones are the gaps most often missed on a plan.
6. Check overlap on the areas that cannot go dark. One camera on a main entrance is one failure away from none.
7. Verify on site. Walk the fails and the borderline passes with a live view. Lighting, glare, and backlit doorways do not appear on any plan.
8. Repeat after every change. A new camera, a moved one, new racking, a remodel. The audit is only as current as the last time the plan matched the building.

Steps 2 through 6 are the slow part by hand. With a spreadsheet and a PDF floor plan they take days per site, and the result starts aging the moment the next camera is moved.

Finding blind spots across 500 cameras

The MentatNOC camera coverage map does the geometry on the same floor plan the fleet reports to. Each camera is a pin with its model’s datasheet optics, aimed the way it is aimed. Walls clip the views and glass does not, and a mount height and tilt turn a view into a band with its dead zone under the camera. A zone is a named area with a required level, and the map marks it PASS or FAIL with the share covered. Overlap heat shows where one camera is the only one on an area, and the entrance gaps list names every entrance camera that falls short of identify in front of it.

The fifth blind spot shows up on the same map. Each pin carries the fleet board’s live health, so a camera that drops offline changes on the plan within seconds and the area it covered is plain to see. Black or frozen images and a camera knocked off its aim raise their own alerts. It monitors device health, not video. The views on the map are a model-based estimate, not a survey, so the walk in step 7 still happens, on the fails instead of on every camera.

For an integrator, the FAIL list and the entrance gaps are a scoped list of work you can quote, and a camera knocked off its aim reaches you as an alert before the client notices. See it running against a 500-camera fleet on the interactive demo.