Short answer
DORI stands for Detection, Observation, Recognition and Identification. It is based on the IEC 62676-4 standard and is measured by the number of pixels per meter in the image. Roughly, detection needs 25, observation around 62, recognition 125 and identification 250 pixels/meter.
Why don't megapixels tell you much on their own?
When buying a camera, most people look at megapixels. It is assumed that an 8 MP camera sees better than a 4 MP camera. Yet whether a camera can make out a distant face depends on how wide an area its pixels are spread over.
The same 4 MP camera with a wide-angle 2.8 mm lens sees a garden 20 meters wide; in this case, few pixels fall on each meter. If you point the same camera with a narrow-angle lens at a door 5 meters wide, far more pixels fall on each meter and the face becomes clear. DORI is used to measure this difference.
What do the four levels mean?
The values are based on the classification in the IEC 62676-4 standard, and manufacturers list them in camera catalogs as DORI distance. The pixel values below are the number of pixels falling on one meter of image width at the target's distance.
- Detection (about 25 pixels/meter): the presence of a person or vehicle is noticed in the image. You cannot tell who it is.
- Observation (about 62 pixels/meter): general features such as clothing color, build and what the person is doing are visible.
- Recognition (about 125 pixels/meter): you can recognize someone you know in the image. It may be hard to tell a stranger apart from someone else with certainty.
- Identification (about 250 pixels/meter): the face appears in detail that leaves no room for doubt. This level is the target for establishing a person's identity after an incident.
How do you calculate it for your own camera?
The calculation is simple: divide the camera's horizontal pixel count by the width of the area seen at the target distance. For example, if a 4 MP camera with 2560 horizontal pixels sees an area 10 meters wide at the target distance, about 256 pixels fall on each meter. That is at the edge of the identification level. If the same camera sees a 20-meter width, the value halves and drops to the recognition level.
The visible width is determined by the lens's horizontal field of view and the target distance. In the camera distance calculator on our Tools page, enter lens and the distance to see this width and which DORI level you fall into.
Expert note: Catalog DORI distances are measured in ideal light at the camera's optical limit. At night, under infrared light or in rain, in practice you get the same detail at a shorter distance. Plan somewhat below the catalog value.
Which level is needed at which spot?
Planning every camera for the identification level needlessly increases the number of cameras. What matters is deciding what needs to be seen at each spot.
At building entrances, in front of the cash register and at turnstile passages, where everyone passes one at a time, identification is the target. At these spots, the camera should be close to the person and have a narrow angle. In places such as corridors, the sales floor and inside a warehouse, observation or recognition is usually enough; here the goal is to follow how an incident unfolds. In areas such as perimeter fences, large parking lots and fields, detection is enough. Here the camera does not identify the person; it signals that there is movement, for the alarm or a guard.
Putting this distinction in writing at the quote stage makes your job easier. One line per camera is enough: location, viewing direction, target distance and target level. After installation, you test with the same list; it shows right away which camera hits its target and which needs a lens or position change. The list also becomes a reference document when adding cameras later.
DORI is not enough for reading license plates
Reading license plates is a different task from the levels DORI defines for people. For plate characters to come out clearly, angle, shutter speed and light matter more than resolution. A moving vehicle's plate comes out blurry with a slow shutter, and at night headlights wash out the plate.
That is why a separate camera is planned for plates: at a spot where vehicles slow down, looking at the plate at a near-right angle, and set up for plate reading. Expecting a general overview camera to read plates is the most common disappointment.
At housing estate and parking lot entrances, the area right in front of the barrier is the best place for this. The vehicle stops or slows down there, and the plate comes to the camera from a few meters away at a straight angle. If the road is open and vehicles pass quickly, getting consistent results from a plate camera becomes harder; in that case, adding a speed bump or a barrier to the road makes the camera's job easier.
Mounting height and angle throw off the calculation
The DORI calculation assumes the camera looks straight at the target. In reality, the camera is usually mounted on a ceiling or pole, three or four meters above the ground. The higher the camera, the more it looks down on the person; the image shows hair, a hat or a cap instead of a face. Even if the pixel density is at identification level on paper, the face cannot be made out.
That is why, at spots where identification is the target, the camera is placed as low as possible, facing the direction the person walks in. At doorways, a height of around two and a half meters is often a good balance; the camera is harder to reach by hand, and the face angle does not get too steep. The exact height is determined on site, based on the width of the door and the distance at which the camera will capture the person.
Light also changes the calculation. Strong sunlight from behind turns the face into a dark silhouette. In Şanlıurfa, this problem is common at west-facing glass doors in the afternoon. A camera with wide dynamic range (WDR) balances this difference to some extent, but it does not fully rescue a wrong angle.
How do you ask about DORI in a quote?
When getting a quote, ask these two questions for each camera: what distance is this camera looking at, and what level does it target at that distance? The installer should position each camera with a target. A wide-angle camera mounted on the ceiling 8 meters from the door sees the door but cannot identify the person entering.
Ask for a test after installation as well. One of your employees stands at the target spot, you take the image of that moment from the recording and check how clear the face is. Test daytime and nighttime separately.
In the systems it installs, Desa Bilişim places each camera with this target and runs the test together with you at handover.
Frequently asked
Does an 8 MP camera always identify farther than a 4 MP camera?
With the same lens and the same light, an 8 MP camera gives more pixels, so it can reach the same detail from farther away. But sensor size, night performance and compression also affect the result. In a dark spot, a good 4 MP camera can give a sharper image than an 8 MP camera with a small sensor.
Does a varifocal or motorized lens change DORI?
Yes. As you zoom the lens in, the field of view narrows and more pixels fall on the target. With a motorized lens, fine-tuning the camera to the target point after installation becomes easier.
Where do I find the DORI value?
Most manufacturers give DORI distances in meters on the camera datasheet. This value may have been measured at the camera's narrowest lens setting; check which lens it is given for.
What detail should each camera provide?
Let us measure your spots on site and prepare a quote that states the target DORI level for each camera.
