CCTV, Why?

We are the most watched nation in the world, we have the most cameras per person and statistics have shown on average we could be caught on a CCTV system a hundred times a day. How many times have people been convicted on the strength of a CCTV image alone? It therefore begs the question CCTV, why?

It's true most CCTV installations are often a knee jerk reaction to an incident or to an environment. For instance a school is broken into and computers are stolen, suddenly CCTV cameras are a must, a local installer is called and a "CCTV system" is installed. The school down the road hears of the incident and calls the head teacher who passes on the details of their installer and another "CCTV system" is installed. It spreads like a virus and suddenly the whole area is saturated by CCTV cameras, all purporting to provide protection from theft or violent crimes, to such an extent the local council feels it needs to install them in the street to make local people feel safe.

So is the school any better protected because of the cameras? Probably not, the cameras them selves may act as a deterrent (no body likes being watched while committing a crime) but can the school afford to have security out of hours watching the video streams constantly? The majority can not. The thieves are aware of this and commit the same crime this time wearing hoodies and masks, the next day the school have some grainy images of shapes that look like people carrying off the schools computers again. CCTV failed, and yet so much faith is put into these systems to protect property and people, so will the CCTV system be updated? No the school can not afford it. So CCTV, why?

The problem is not in the concept of CCTV but largely the application of the CCTV and often the installation of the system, promising more than it is delivering. The Home Office recognised the need to provide a means of giving guidance to CCTV installers, the trouble is its a large document not very easily found and most CCTV installers will never have read it. Find it here. You also have the problem of competition, installers undercutting with cheaper systems which can not provide the cover needed. It has to be said that any system is only as good as the monitoring applied to the system, going back to the school scenario had the video been monitored then the thieves could have been intercepted, so no matter how good the CCTV system was if its not monitored then its not fit for purpose.

Compounding the issue is the recording of the video stream. To record the stream the image must be converted into a file format which can be read by devices after being archived. These range from tape recorders to disk storage devices. However the act of recording the image degrades the quality which is the reason when we see CCTV images on the TV of recorded video streams they always look fuzzy and unrecognisable. This is caused by compression techniques used to increase the total recordable space needed. So CCTV,why?

What about IP?

OK so i have been deliberately negative toward CCTV systems but all of the above really only applies to analogue in general. The home office document still applies to IP cameras systems which we will call IP security cameras systems (IPSCS) as CCTV does not describe an IP based system and is only really similar in its goal but not its application, i want to make the distinction between the two types of systems.

IPSCS's are not the cure all but it does make the concept of video surveillance work. It has lots of nice features which means IPSCS's systems are not limited by the physical limits of CCTV.

Subject Image Size

There are many applications of a security camera, should you be looking at a crowd you will want a large detailed image to see any incidents but should you want an door entry camera you will want the subjects face to occupy as much of the screen as possible. In the Home Office document its referred to as Monitor, Detect, Observe, Recognise and Identify (see doc for more detail) various sizes of the subject relative to the screen size. This describes the function of the camera, if you are looking at a car park then monitor is the goal and a suitable camera can be used or if you are capturing faces at an ATM machine then identify is the goal again a different camera would be specified. With IP cameras this blurs the boundaries of these classifications as a mega pixel camera may be able to monitor and provide identification due to digital zoom. This dual functionality can be achieved with in the analogue world with a PTZ camera, but not in the same way. With an analogue PTZ (pan, tilt and zoom) camera any time you zoom into a scene the surrounding area is lost and only what the camera is looking at is recorded, so while identifying the camera is no longer monitoring, its an either or situation.

Mobotix cameras are a perfect example of a well engineered IP surveillance camera (IPSC) and in many situations one camera can replace three to four analogue cameras. They have no moving parts as only digital zoom is employed and the image that is processed is a true digital 3.1 mega pixel image not a re-engineered analogue camera with an encoder bolted on. Unlike a fixed zoom analogue camera the subject size can be changed to match the required size to provide an all round camera function in line with the Home Office guidelines.

Event Notification.

In our opening example the school could not afford to monitor the CCTV round the clock and so the thieves got away, how can an IPSCS help? There are two ways to help with monitoring you can configure the cameras to monitor the areas of interest for movement and once movement is detected then send warnings with pictures via email, sms or in some cases as with Mobotix cameras the camera can make a call to your mobile alerting you movement has been detected. Doing away with the need to have round the clock or out of hours monitoring.

Brining it all Together

There is another way, its true i am pro IPSC and will always look to fulfil a surveillance system with IPSC but an application like Milestone can take video feeds from various camera manufacturers and even analogue cameras (via encoders) and unify the camera streams under one common interface and management system. So should the camera not have the ability to monitor movement then Milestone can provide that in software helping current analogue system migrate seamlessly into the IP world.

Storage and Loss of Image Quality

When storing the images as can be seen in my previous blog even in the IPSCS world how you record the image can have a massive impact on retention of your initial quality. What you see might not be quite what you get. In the analogue world this is amplified to an even greater extent. The largest image possible in the analogue world is 4 CIF or .4 mega pixel when this is recorded and compressed it will be reduced to .1 mega pixel rendering the images next to useless. Milestone comes to the rescue again, as an encoder will not degrade the image quality when converting it to a digital stream (depending on the codec used, see previous blogs) so Milestone can record the image at close to the .4 mega pixel not a patch on IPSCS but better.

IPSCS Why?

So IP and its associated technologies actually makes the concept of security via video work better. Its not the cure all but its a very large step in the right direction. Compared to CCTV it really does work and makes your investment give you the protection you are paying for. This is just looking at a plain IPSCS system without the addition of analytics packages which can further extend the power of your IPSCS (a topic for future blogs). So if you want your CCTV system to be an investment in protection and actually live up to what is promised then install IP based CCTV an IPSCS or if you have a CCTV system then Milestone can be your champion in putting the analogue streams on an IP based system.

The bottom line is if its not IP why CCTV?

Which Codec? - Why does is Matter?

Choosing the right codec for your camera can be the difference between your camera network being fit for purpose and just being a waste of technology.
What codecs are out there and how do they work?:

Before we start this explanation it must be said this is my simplified view of theses codecs i have added links to fuller explanations should you want some bed time reading.

MJPG or Motion JPG this is essentially JPG one of the standard formats of images used on web pages due to its space saving compression applied to the image. The motion bit of MJPG is the same principle of cartoon drawing a rapid changing image with small changes equals motion. So MJPG is many JPG images displayed rapidly one image per frame.

MPEG4 or MPEG4 part2 is a block based video compression algorithm. What it actually does is it takes a view of the video stream rather than each individual frame then a comparison is made between frames and checks for changes between these frames, then only the block(s) that have changed are updated when sent to the end device. So the initial image is sent as a whole image called the key frame then only changes to the key frame are updated, unless the scene changes to such an extent that the sending of the changes would require more bandwidth than another key frame so a key frame is sent. It can be also said that the codec may also send key frames at fixed intervals to make for a smoother transition and better quality image. Compression is used to decrease the bit stream even further.

H.264/AVC or MPEG4 part 10 is just another version of MPEG4 i hear you say, well that is correct to a certain point but it is offered as an alternative to MPEG4. Its aim was to build on various standards of MPEG and greatly reduce the bit rate required for high quality video transmission/storage. It can achieve 50% or greater saving on standard MPEG4 bit stream through its greater compression and overall image interrogation algorithms.

MxPEG exclusively used by Mobotix cameras, mxpeg has been designed for surveillance cameras by a surveillance camera manufacturer. It builds on the best quality codec Mjpeg and uses the block change approach of Mpeg4, but only using Jpg compression. The result is significant reduction in the bit stream compared to Mjpeg but images of a high quality compared to H.264.

So Which Is Best?

Well there is no "best" codec, some favour reduction of the bit stream over quality of image and some favour quality of image over the reduction the bit stream. It all boils down to what you are trying to achieve. If you are performing analytics on the post event images then you would want best quality images. If you have limited band width you would want a reduced bit stream. The interesting thing about all of these codecs except one is they where not designed for IP surveillance cameras, instead they where aimed at broadcast of movies over the internet/networks. The odd one out being Mxpeg which was designed for surveillance camera streaming.

Individually then how do they stand on their merits:

MJPEG - Is high quality with limited compression, meaning you get the best image quality but at the cost of bandwidth, but because every frame is a key frame then each frame give or take a few bytes will be the same making it very easy to work out the load of a constant stream from each camera. Also because jpg compression is low the camera CPU can dedicate more cycles to other camera funtions.

MPEG4 - Is better at reducing the bit stream but at a slight cost to quality, also making your bandwidth calculation harder due to the none uniform size of the frames due to block change compression.

H.264 - Reduces the bit stream considerably but at a much greater cost to quality. The reduction in quality is noticeable and is reminiscent of analogue loss on recorded images as compared to live images.

Mxpeg - Is a lower bandwidth option better than MPEG4 but without loss of the quality from Mjpeg seen in H.264

So as you can see it matters which one you use. Here are some recommendations for various situations:

PTZ cameras on continuous patrol - In this situation it would be better for all aspects of the network to use Mjpg as each frame will be a key frame due to the motion of the camera, meaning the camera will not be running complex algorithms on each image only to decide to send the whole image. This also applies to very changeable images.

Live view of images - This is a tricky one but for the sake of bandwidth then MPEG4 would probably be best, very little is lost on transmission in the way of quality and the network load is reduced, with the caveat that the images seen are fairly static.

Limited storage facilities - I hate to say it but H.264 would be your best for reduction of storage needs but this flies in the face of IP surveillance since quality is the greatest gain and H.264 will reduce that. Also remember because of the high compression of this codec multiple streams to a server would mean a very high load on the CPU, to the extent that it causes the processing of images to back up and even drop frames.

I have not mentioned it yet but here goes Mxpeg would be the best overall codec to use in all surveillance situations, but sadly only Mobotix cameras to date use this. So if you can use Mobotix cameras then do so and reap the benefits of Mxpeg. Not to mention the high quality 3Mpx images and extended functionality of the cameras.

In Summary the right codec is dependant on what you are trying to achieve and often circumstance will dictate which codec is the right one to use. We must remember the aim of IP video surveillance is to deliver a high quality usable image that can be used rather than these fuzzy blurred out lines we are used to seeing from analogue systems. So the codec we use should try not to compromise the quality but at the same time not impact on other systems.

The ultimate best codec for surveillance in the perfect world would be Mxpeg but since only Mobotix cameras use it only people with these cameras reap its benefits. Who knows it may even become and industry standard one day and available to the masses.

IP Cameras - the most asked question

The single most asked question with IP cameras is also the single hardest question to answer. How much bandwidth/storage will i need for my cameras? Seriously it gets embarrassing when you have to say i do not know. That said however there are reasons why we can not say, such as how many cameras, frame rate image size, image quality, codec used, recording schedule and type of trigger if any to activate recording. There are some guides out there such as the one developed by Overland Storage and Mobotix, but they are only a guide. A rough estimate for Mobotix cameras is about 2 to 3 gig a day based on movement triggers.

Taking it a step further lets look at some numbers to show how difficult it is to pin a number down. The best an analogue camera can do is about the same as VGA(640x480) 0.3M pixels, its actually 4 CIF(704x576) 0.4M pixels, you would not find a mobile with this limited camera abilities today. Mobotix cameras however wade in at an impressive genuine 3M pixels so if we just take that number at 10 frames per second how does it compare to poor analogue cameras? Well 10 x 3Mpx is equal to 100Fps at 4 CIF (BTW CIF stands for Common Intermediate Format used to standardise the horizontal and vertical resolutions in pixels) or 400FPS at 1CIF.

This is a scary thought i here you say.....not really you see with IP cameras you can get the camera to do a lot of the work on the camera and only send what is of real interest. There is no point recording an empty room or a quiet street so the camera is set up to to only send to the recording device when movement is detected, this does 2 things it reduces storage requirements dramaticly even at 3Mpx and makes events easier to find during and after the event. The size and sensitivity of the trigger box can have a big impact on reducing false posatives too as a small trigger box could be constantly set off by any thing passing, so to detect cars entering a gate a car size trigger box is used so a person walking past would not be able to change the pixels sufficiently to cause an event.

So getting back to the question it should really be worded like so - If i had 5 cameras all running at 3Mpx at 12 FPS using MXpeg on an out of hours schedule with motion detection triggered recording stored for 31 days how much diskspace will i need? Answer ABOUT 760GB - 1TB planning this sort of system means worst case scenario a 2TB system will required.

Next time we will discuss the pros and cons of camera codecs.

Welcome to my blog. Its not going to be flashy but rather it will hopefully provide some incite in to using IP cameras such as mobotix, network planning for IP cameras and even some other IP related products such as Netgenium (door entry systems). So till my next update i will carefully consider my first subject and post ASAP.