When it comes to installing the right camera, the devil is in the detail
Closed circuit television, or more commonly known as CCTV, was invented in the Soviet Union in the 1920’s and was almost immediately used for no good; with some of the earliest CCTV systems being demonstrated to Joseph Stalin, installed in the courtyards of the Moscow Kremlin and used in Nazi Germany to observe the launching of rockets. It later found a more noble purpose in broadcasting live sporting events to venues as a form of pay-for-view. In fact, the Rumble in the Jungle in 1974 between George Foreman and Muhammed Ali attracted 50 Million CCTV viewers worldwide!
Now, nearly a decade on it is estimated that there are around 1 Billion CCTV cameras installed around the world (and a whopping 65% of those are in Asia).
There’s a broad spectrum between good and evil that CCTV has, and not doubt continues to be used for, but the purpose of video surveillance in most applications is to promote a feeling of safety and security, deter criminal activity, and detect and identify those involved in a crime. And it’s the latter of those that we want to focus on here: how to detect and identify the perpetrators of a crime.
You see, it’s not as simple as sourcing a CCTV camera, fixing it to a wall and pointing it in the right direction. There’s some science behind it, and a huge difference between a camera that has the ability to detect a person, and one that has the ability to identify a person.
Firstly, let’s look at the D.O.R.I principle of detect, observe, recognise and identify which is an international standard that defines the criteria for the performance of a camera.
|Detect||The ability to detect if there is some ‘thing’ there vs. nothing there|
|Observe||The ability to observe what that thing is, such as a person, animal or vehicle|
|Recognise||The ability to recognise a person you know or the make and model of a vehicle|
|Identify||The ability to positively identify a specific individual beyond reasonable doubt|
If careful consideration hasn’t been taken before installing that CCTV camera, then there’s a good chance you might find that your investment in a security system has left you unable to clearly recognise the distinguishing features of a person’s face or read the number plates of a passing vehicle. Some of the considerations include the position of the camera and angle of view to the subject, and the direction and intensity of the lighting. But the most vital one is the resolution of the ‘thing’ you are trying to identify on the screen. And there’s some science behind this:
People come in all shapes and sizes, but more or less we have the same size face, which on average is about 16cm wide. Now, if that face is being captured on CCTV, in order to be able to clearly identify the characteristics of that face, such as eye colour, birth marks or tattoos (god forbid 🙄) the resolution needs to be around 40-80 pixels horizontally from one end of the face to the other .
Reduce the amount of pixels and you might be able to recognise someone (Investigator: “oh, look that’s you Bob!”) but you won’t be able to prove that it is that person (Bob: “no, that must be someone that looks a bit like me, but it’s definitely not me…”) in a court of law. Reducing the resolution and pixel count even further and you’ll be able to observe that there’s a person there, but you wouldn’t be able to pick them from Tom, Dick, Harry or Jane. Anything lower than 4 pixels per face, and you’ll be left struggling to distinguish the potential cat-burglar from an actual cat.
Now, this all sounds a bit hard and how on earth are you supposed to measure the number of pixels across someone’s head without seriously invading their personal space? Well, we translate pixels per face into pixels per centimetre and measure the distance between the camera and the object. And as long we stick to the following formula for the resolution we know we’re going to capture the level of detail required:
|Detect||0.25 pixels per centimetre|
|Observe||1.0 pixel per centimetre|
|Recognise||1.5 pixels per centimetre|
|Identify||2.5-5 pixels per centimetre|
Real life conditions are challenging and there’s lots of other factors that can affect the ability of a CCTV camera to recognise or identify a person or vehicle involved in a crime. But if the camera that’s been installed doesn’t measure up with the pixel count, then no matter how hard you squint, you’ll be left staring at a pixelated image of some ‘thing’ moving in the background (and no, unfortunately there is no CSI style lab expert that can magically blow up that grainy image into a full HD poster and save the day).
At Red Flag Systems, we’re experts in the design and installation of CCTV systems. If you’re considering implementing or upgrading the CCTV system at your facility (Nazi’s and Stalinists excluded) then reach out to us on 1300 685 504.
1300 685 504
Get in touch with us
Request a call back
Proud members of ASIAL and we share their commitment to promoting standards and raising the level of professionalism within the security industry