In order to specify the necessary functions and also to later select a technical solution, it is helpful to have some background knowledge about basic features. Here I explain some of the options on offer today.
Analogue or digital or IP?
Until the late ‘90s this question did not arise. All components were fitted with so-called composite outputs using coax connectors, through which the standardised analogue video signal was transmitted between components.
The first step towards digitalisation was the appearance of digital recorders which converted analogue video signals into digital signals so that they could be compressed and stored – usually on hard disks. Since then digitalisation and network technology has comprehensively invaded the video surveillance world – from the camera, via transmission methods to video analysis.
Analogue matrices (switchers for multiple video inputs to multiple video outputs) are now rarely used in control centre systems. Their function has been taken over by network infrastructures. These ensure that the video picture from a selected camera is routed direct or via a digital video management server to the desired monitor. Today’s systems no longer use pure analogue matrix technology. Unfortunately though this new technology does have some disadvantages. For instance, if the network does not have the necessary bandwidth available this can lead to the generation of artefacts – errors in the images – during display. But I will talk more about network infrastructure in a later section.
The main differences between control centre systems relate to the picture compression process used. Although there is much talk about standards, systems from different manufacturers who implement the same standard can still be very different in terms of performance.
The question of whether to use analogue or IP technology is really only relevant to the choice of the camera. Here we need to differentiate between the internal image capture and the external video output.
There are two basic technologies for image-capturing chips which are usually known as analogue (CCD) and digital (CMOS). Usually you can assume that CCD image capture is more light sensitive, whereas CMOS chips have a greater dynamic range, ie, they can cope better with backlighting and large variations in brightness in the picture. The real image quality though is influenced decisively by the camera’s internal image processing, so that the choice of the type of image capture system does not provide information about the realisable image quality with different lighting conditions. Basically it is always advisable to test cameras on site with the existing or expected light conditions.
A camera’s external video output is either analogue (CVBS), network or HDCCTV output (using HDMI standards). All analogue cameras deliver the same fixed resolution. Network cameras come in a wide variety of resolutions. The current ones are D1 (comparable with analogue cameras), HD and Full-HD, with those above HD being referred to as megapixel cameras. The term megapixel really just says that a camera has more than a million pixels and gives a clue to the format (aspect ratio) of the picture. The brand new HDCCTV standard tries to combine the advantages of the simple and reliable point-to-point connection of analogue cameras with the advantages of the higher resolution of megapixel cameras. It is still unclear whether it will become established.
The most important selection criteria for a camera is the required image quality, and this is influenced by:
* the location and viewing angle.
* the required detail/resolution.
* the lighting conditions.
There are some anomalies which arise when using network cameras, so I need to say a few words about issues relating to the detailed resolution and selection criteria for network cameras.
Megapixel or standard resolution?
The higher picture resolution (eg, megapixel) is not always ‘better’ than a lower one (eg, D1). You need to consider the following factors:
* The higher the resolution the more pixels there are in the picture to enable it to provide more detailed resolution. However, more bandwidth for transmission to the recorder or screen and more storage capacity are also required.
* If there is no uninhibited view into the distance then it depends on the location of the camera as to whether higher resolution and more detail are useful or beneficial. It depends on the scene whether a higher resolution actually provides additional relevant information.
* Since even the highest resolution camera chips have the same physical dimensions as low resolution camera chips this means that each individual pixel has a smaller area on the chip. This in turn means that the quantity of light reaching this point is proportionally smaller and this explains why high-resolution cameras with the same chip size are less light sensitive.
* In certain applications, there may be data protection reasons for not using a higher resolution to capture more detail. It may not be desirable or permitted to recognise an individual.
* A high-resolution camera needs a lens which can deliver the same resolution. This means that not only do camera costs increase, but also the lens costs.
If you decide to use a high resolution camera, then it is vital to use a matching megapixel lens in order to reach the appropriate detailed resolution. Unfortunately in practice this imperative is often ignored for cost reasons.
Megapixel cameras offer the advantage that a single camera can present a significantly larger scene and this may reduce the number of cameras required. They also offer the facility to enlarge details in the live images as well as in the recorded ones. For many applications this provides relevant information, eg, for reading vehicle number plates, recognising people etc.
Other selection criteria
With network cameras there are other important selection criteria which are not relevant to analogue ones:
* Switching behaviour: The time taken to switch to and from event- or alarm-controlled image quality or picture rate can be a significant issue. Only if the camera reacts to relevant control signals from the control system without any visible hesitation can you be sure that the critical images and sequences have been displayed and recorded at the required quality. In tests with various different cameras we found reaction times vary with the manufacturer and camera type from less than 500 ms up to 10 seconds and more. For most applications a camera reaction time of less than 1 second is recommended.
* Stability: Basically network cameras are tiny computers with a special function. Many cameras even use a Linux operating system and offer a built-in Web server which can be accessed with standard browsers. Unfortunately even these specialist computers are not immune from software or infrastructure errors and it is still best to get references to be sure that these will provide reliable uninterrupted service over a long period.
* Compression: I will deal with differences between compression processes separately below, but here it is worth noting that phrases like 'up to xx frames per second' and similar descriptions of resolution often appear amongst the technical data for cameras. Unfortunately it is often the case that the image resolution, the compression process and the picture rate are all directly related so many cameras cannot deliver the full picture rate listed together with the highest resolution.
* Dual (or multi-) streaming: This term refers to the ability of a camera to deliver two video streams in parallel to different users. What needs to be checked here is that the streams actually flow independently of each other as far as the compression process, the picture rate and the picture quality are concerned.
Centralised or decentralised system structure
A centralised system structure or architecture means that the control, management and often too the ‘intelligence’ of the system is administered from a single point. This is comparable with a central computing centre where the important processes are run and the data managed and processed. The clients connected to this computer centre, the operator computers, are primarily occupied with display tasks and providing the user interface. The intelligence is situated at one particular location, and the peripherals are non-intelligent. A decentralised structure is more like having many computers each running different programs independently, but are nevertheless networked together and exchange data with each other. There is no central intelligence. Each computer is independent of the others and fully functional.
Since the introduction of network cameras it has also been possible to construct decentralised structures. Here each camera is its own computer which runs the relevant logical functions for this camera – such as image capture, compression, video analysis, storage and the independent distribution of images to users on request – all independently of the remaining cameras. The major advantage of this kind of structure is its reliability, as any defects or problems only affect a small part of the whole system.
As ever the advantages here do not come without disadvantages elsewhere. The range of suitably intelligent cameras in the market is limited. Since the cameras are not compatible with each other, the user has to cope with the integration of different camera types on the user-side, which is extremely restrictive on user convenience. Access to the camera functions is usually via the camera’s own Web server, so the operator has to deal with many different browser windows, which may have different designs and layouts, all in parallel. If the requirements for storage or video analysis increase, then the system cannot be expanded from the centre but may need to have its cameras changed. The camera system link to any third-party system has to be completed separately for each individual camera. Generally a decentralised structure is only recommended for video systems with very few cameras.
The centralised structure provides for the integration of different functions and information at one (or several cascaded or redundant) central node(s). This central node is the video management system that in smaller systems is usually integrated into the video recording device.
The video recorders take on the function of providing individual users with camera signals as well as storing video and often they deliver video analysis too. These functions are controlled by the internal event processing. The system is pre-programmed with reactions for each event, which it then has to carry out. These reactions range from the switching of a live picture in response to a user action, to the alarm-controlled switching of compression parameters in the camera and the recording parameters.
For more information contact Geutebrück, +27(0)11 867 6585, email@example.com, www.geutebrueck.com
Katharina Geutebrück studied electronic engineering and worked in Italy and France before returning to the family firm in Germany, where she has been managing director since 1999. She is chair of the video technical committee of the German Association of Security System Manufacturers and Installers (BHE); a board member of the regional Association for Security at Work (VSW-NW), and is active in many security market committees.
© Technews Publishing (Pty) Ltd | All Rights Reserved