Megapixel Cameras - The Future of CCTV

Published Jul 31, 2008 17:38 PM
PUBLIC - This article does not require an IPVM subscription. Feel free to share.

Megapixel IP security cameras deliver one great benefit that traditional analogue CCTV cameras just cannot match, you'll all be using them tomorrow, read on to find out why (Also read my blog for more good information on IP video).

The Analogue CCTV Camera Story So Far

For my first few years working as a company director in a specialist CCTV installer the maximum image resolution that we could provide from a colour CCTV camera was 480 TeleVision Lines (TVL).

TVL is the industry standard for denoting the horizontal resolution of a CCTV image - basically you draw a line across the image and determine how many vertical lines you are able to resolve (of course, we at the coal face never did this - the camera manufacturers simply made claim to a stated resolution in their product specs).

About five years ago, Sanyo launched a new market-leading colour / mono CCTV camera capable of delivering 520 TVL resolution in colour.

By the way, colour / mono switching simply means that the camera has the ability to deliver colour images all day long - whilst there is sufficient light, but will then automatically switch to producing a monochrome picture overnight.

* monochrome cameras are about 20X more light sensitive than colour cameras

* colour cameras deliver more information; blue car, red jacket, etc.

* only monochrome cameras can utilise infra-red lighting

* modern colour cameras can almost 'see-in-the-dark' anyway, and simply need a small amount of white light

I digress, I guess about three years ago Sony stole the top-spot in terms of resolution, by producing colour cameras with a claimed 530TVL resolution.

It was an ongoing battle; Samsung, Panasonic, Baxall, and many other CCTV equipment manufacturers were all competing.

About a year ago, Bosch produced a very good camera, colour / mono switching, 15-bit digital processing, fantastic dynamic range, we were all wowed by its low-light capability. This camera was able to resolve 540 TVL.

To the best of my knowledge, nobody has claimed greater resolution from an analogue CCTV camera in the last year - you see, as I understand it, there's a limit to what you can resolve using the PAL video standard. Aplogies to our friends in the USA, you have your own NTSC limitation, and I'm sure that somebody really clever could write a separate Knol article to explain this to us all - but for now, I want to keep you awake and following my main theme!

You will see that, in a competitive market, with major international technology manufacturers competing for the crown, the resolution bar was raised by 4% over 5 years, and is now at a level where the technology used caps what can be delivered from the devices - don't expect higher resolution security cameras (analogue CCTV) any day soon.

Along Comes The IP Camera

IP is an acronym for Internet Protocol - simply put, these cameras come with a network socket on the back instead of the traditional BNC connector used on analogue CCTV cameras.

Manufacturers such as Axis and Mobotix have been producing Network Cameras for almost ten years now. Initially, the majority of their products were simply treated as webcams on the network. End-users would literally buy them and point them out the window!!

The early IP cameras had limitations which prevented them from competing in the security sector:

* low resolution (images that did not contain as much detail as CCTV cameras)

* poor performance in low-light situations

* a low 'images per second' performance (CCTV uses the term frames (for images), and analogue CCTV cameras produce 25 frames per second), some early network cameras only produced 3fps

* a worsening combination of the two above i.e. in low-light situations the camera's output video would slow down as it struggled to render and deliver images

* furthermore, all of the affordable models used CMOS imaging chips and could not control an auto-iris lens, so they could only be used indoors

Just to confuse us all, as IP cameras are from the world of IT, they define their digital resolution claims in terms of pixels, as do most computer monitors.

A typical budget IP camera such as the Axis 205 (in 2003 - now discontinued) provided a maximum resolution of 640 X 480 pixels. The title of this article says I'm here to talk about magapixel cameras, so let's just convert that performance to megapixels, just multiply the one number by the other to get 307,200 pixels - approximately 0.3 megapixels.

OK, so how far have we come with this technology in the past five years ........

Just about everybody has a 1.3 megapixel camera in their range - that's just 'me too' technology. Most manufacturers of any standing will have a 2 megapixel camera. The leading IP camera manufacturers are now fighting it out in the 3 megapixel plus arena. The show-offs can provide you with 5, 9, 11, 16 or even 21 megapixels!!!!

It's not that surprising, technology has moved on, my mobile phone has a 3 megapixel camera built-in.

The digital camera that we've used for several years has a 4 megapixel sensor, the smaller more recent one that we bought because the other was too big to carry around has a 7 megapixel sensor, and the one I'd like to buy next has 12!!

Kodak have just developed a 50 megapixel sensor, and another lab is claiming 60 megapixels.

Oh, I forgot to tell you what 540TVL in 'CCTV-speak' equates to in megapixels. A 540TVL image equates to 720 X 540, which if you do the multiplication is - 0.389 megapixels.

OK, apart from bragging rights, what does this mean in the real world .......

The UK Home Office Guidelines for CCTV Evidence

If you are a CCTV professional, then the Home Office Guidelines for CCTV Evidence should be your main point of reference when attempting to meet the Operational Requirement for any CCTV camera position.

I am going to gloss over this quickly to try and keep your attention on the main thread. There are basically four uses of CCTV cameras which the guidelines set image challanges for e.g. if you are aiming to achieve this outcome, then the resultant image must be to this standard:

* monitoring an area, a person should appear at least 5% of screen height

* detection of intruders into an area, persons should be at least 10% of screen height

* for recognition of a known individual, persons should be 50% of screen height

* to identify an unknown person, present an image at 120% (knees to head)

* The standard height for a person that they based these standards on was 1.6m.

Always Remember - the format is fixed!

(if you make the image wider, you make it taller, you make people in the image appear smaller, with less detail each)

The standard screen format for a CCTV monitor has always been, as it always was for domestic TV, 4 X 3. Quite simply meaning that the height of the screen (or scene) is always going to be 3/4 of the width of the scene, and there is nothing you can do about that!

The Home Office set the guidelines above because they knew where they stood with analogue CCTV resolution performance; they knew that to achieve useful resolution (fit for those stated purposes) as long as the professionals installing the CCTV cameras acheived those percentages, any evidence obtained would be fit for purpose.

If we consider the 50% challenge:

* we need to present a 1.6m person at 50% of screen height

* so the total screen height will therefore be 3.2m

* therefore, because the format is fixed, the maximum scene width that a camera can cover and meet the 50% challenge is 4.3m

* that's 14 feet in old money, and, in the customer's eyes, it's not a very wide scene to cover

* if you shoot for the 120% goal, for identification of an unknown person, you need to limit the scene width covered by an analogue CCTV camera to just 1.8m (or a bit less than 6 feet).

End-users seldom allow installers to actually set their cameras up to cover these narrow fields of view, they like to look at their CCTV monitors and have the warm feeling of satisfaction that comes from knowing that the cameras are covering everything and everywhere!

The end result, is images that quite simply do not contain the detail required to meet the purpose for which they were bought and installed. Please, just do a double-take, if your analogue CCTV cameras are covering a scene width greater than 4.3m, then you will not achieve a person to screen height ratio of 50%, so your system can only meet the guidance for the purpose of monitoring or detection. Most system owners investing in business surveillance systems expect more, but do not appreciate the limtations until after an incident, then simply end-up disappointed with CCTV!

I'm not going to go into it in this article, but bear in mind that this image fresh from the camera is as good as it can ever be - recording it on a digital video recorder cannot make the image any better (it's actually more likely to lose detail when it's recorded). And, contrary to what you may have seen on Spooks, there is very little you can do to improve a recorded CCTV image - if the detail was not in the pixels recorded, where are you going to get it from?

So, if you're still with me, we've worked through what scene a camera can cover, and why the UK Home Office experts set those guidelines. How is this any different for megapixel security cameras?

It's that greater resolution that makes all the difference. Let's not even begin to push the boundaries in this example, let's just take a current model 2 megapixel network camera. It might have a resolution of 1920 X 1080 (multiply this up to get 2.07 megapixels). If you recall that the equivalent pixels for a 540 TVL CCTV camera is 720 X 540 pixels, you will see that we've got twice the vertical resolution (number of pixels), and 2.66 times the horizontal resolution.

If we were aiming to achieve the same resolution within the image with this camera (bearing in mind that the above guidelines are all about the height in the scene), then we could cover a scene which would be twice as tall, and 2.66 times as wide - instead of being limited to a scene width of 4.3m, we could deliver the same resolution within the scene if the scene were 11.4m wide (a scene of 37.5 feet width is far more useful).

NB Did you notice that the modern 2 megapixel camera that I used for this illustration has also changed format? Just like your TV at home, this one's got a format of 16 X 9 (or wide screen), that's better too - more width without quite so much increase in height - we don't want the sky in the image!!

If you like analogies with TV and computer specs, you might like to think of CCTV evolving with the use of IP cameras through resolution standards that you'll have heard of for computer monitor and TV performance; VGA, SVGA, SXGA, HD Digital, and beyond ... Remember, we don't have to broadcast our high-resolution images, nobody needs a new tuner to see it, we just need sufficient bandwidth, via ethernet standards (RJ45 plugs & sockets, cat 5 cable, gigabit networks, the internet etc) to send and share this video.

The above example was worked through for a 2 megapixel camera. By extending this you can readily see that one camera with 3 megapixels resolution can cover an area four times as big as a high-resolution analogue camera and deliver the same image quality.

The rules have been rewritten (literally) the height guidance in the UK Home Office Guidelines no longer applies, with the variable resolution performance capabilities of these megapixel security cameras, we now need to think in terms of pixels per inch, cm, foot, metre ..... as a new performance standard for evidential image worthiness.

Considering that most installed CCTV cameras failed to meet the guidelines for performance when they were as simple as stated above (e.g. make man = 50% on screen!), it will be interesting to see how we get on now that some effort needs to be put into calculating what image resolution performance will actually be delivered, from which camera (1, 2, 3, 5 megapixels), with how many pixels horizontally & vertically, covering what scene width.

Oh, by the way, let me just tease you with the fact that the lens selected needs to be up to the job - specialist high-quality lenses (made for the purpose) are necessary to resolve the multi-megapixels claimed by the camera manufacturer. This principle is well accepted by photographers, they tend to debate 'quality glass' on an equal or greater importance basis as they give to digital camera bodies and sensor megapixels. Sorry, I digressed again, but it is worth expanding the explanation a little just to illustrate that there is much to be considered ....

Of course, I would advocate not to stretch the scene coverage to its fullest permitted width and height - this is 2008, let's actually have a bit more detail in that image!! After there's been an incident wouldn't it be nice to be able to zoom-in, see what the suspect actually had in his hand ..... I won't go on to work an example of this, but hope that you can readily see that if you just cover a slightly larger scene with a multi-megapixel camera, you will end up with a lot more available detail within the image than you could ever possibly achieve with an analogue CCTV camera.

Of course, there is a balance to be struck; size of scene covered versus pixels per metre within the scene and the number of cameras deployed versus the megapixel performance of each camera.

Please bear in mind that not all IP security cameras are megapixel cameras, if it has an RJ45 socket on the back then it is an IP camera, but if the manufacturer is still claiming 540 TVL, then you're no better off than you were with an analogue CCTV camera (in terms of image resolution at least).