IP Masterclass [Part 2] - Network cameras, servers & compression

It's exciting because of all the new opportunities IP Technology opens up within the CCTV industry. And it is important to installers because it is fast becoming an essential area of expertise – installers who cannot undertake IP installations in the near future will see their markets dissolve before their very eyes!

This month we cover capturing the image, and processing that digitised image:

• Compression – all IP Cameras and Servers need to compress the original analogue signal, for practical purposes. There are a number of different compression methods, and understanding these different methods is essential in the selection of equipment for an IP installation.
  
• Network IP Cameras – digital cameras that attach directly to an IP network and supply an IP signal
  
• Video IP Servers – processing units that can carry out a number of functions, including converting the signal from an analogue CCTV camera to IP.

Compression
Before discussing compression in detail, we should ask – Why is data compression necessary? The answer is that without compression, the volumes of data produced per channel would be far too great to manipulate, display and propagate around what is normally considered a busy network.

  Last month we explained how digital data volumes are measured in bits and bytes, so we can now set out a simple example of data volumes.

A camera working at 500 lines and 300 pixels per line uses 150,000 pixels per frame.

Roughly speaking, that equates to 150KB (1 pixel uses 1 byte of data per single frame of video). Even at 16 frames per second (perceived Real-time video) the signal would be 2.4 MB (Megabytes) of data each second, or 19.2 Mbps (Megabits per second).

This is too large a volume of data for practical use – too much to transmit over most digital networks, and recording devices would very soon be swamped by the volume. Thankfully, with the aid of compression, file sizes are normally around 13KB, if not lower, depending on the compression technique employed.

So compression is a process that reduces the amount of information that needs to be transmitted and stored ... which is hardly new, as conventional analogue CCTV signals have always been compressed from the 5Mhz broadcast signal, to suit our purposes.

Similarly, there has long been a need for data compression in the computer industry. Specialist mathematicians have been solving the basic problem of how to reduce the image size to produce the best compromise between image clarity, the data size of the image, and the amount of processing power it takes to run the compression method.

Different applications have different priorities regarding clarity of the image, data volumes, and processing power. For example, identification evidence requires a higher picture quality compared to monitoring the length of a queue. So for each application there is a suitable compression method.

A server will use one, or at most, two compress-ion formats. Different formats cannot comm-unicate with each other, so you'll need to choose the appropriate compression format for the installation, if you are to offer the best solution.

Different methods of compression are described as lossless or lossy. In general, the less compression the better the playback and record image, so in that sense lossless is always better than lossy.

Compression can reduce the signal in three ways. The first is by various mathematical tricks that are lossless to the image. The second is to remove parts of the signal that are redundant to the human eye viewing the image. The third is to start to visibly reduce image quality – definition, frames per second, and colour range – and it is this third method that is called lossy.

The compression formats used varies by man-ufacturer and by product. But the four most commonly used compression formats are:

• H261
  
• Motion-JPEG, also written M-JPEG or M-JPG
  
• Wavelet
  
• MPEG

H261
H261 is a digitisation and compression scheme for analogue video. It is widely used in video conferencing and is aimed at providing digitised video at bit rates of between 64Kbps and 1Mbps, which makes it most suitable for applications that will utilise public telephone networks.

Compression rates as high as 2500:1 are achieved, but of course at the cost of quality. The format is good for high frame rates, showing movement, but the resolution of those frames may not be high. Thus H261 may not be appropriate if, for example, person identification is required. But if the nature of the application is of a lighter monitoring, such as video-conferencing, the quality is likely to be more than suitable.

Uniquely among the compression formats discussed here, H261 encoded signals can be decoded or decompressed by reversing the process(es) from a valid captured reference or I-Frame.

Thus you can convert the digital signal back to an analogue signal for reintroduction of an existing recording system.

Motion JPEG (M-JPEG)
Motion JPEG is an adaptation of the popular JPEG image compression for still digital photos. JPEG is a lossless compression technique, and the image is less compressed than with H261. Motion JPEG creates a video stream from a succession of JPEG compressed still photos. Because it is based on these high quality lossless images, it can deliver a much higher quality image than H261. But at a cost – it requires a considerably greater transmission bandwidth and storage capacity compared with its H261 counter-parts: 1-1.5Mbps compared to 384Kps for a similar H261 image.

An advantage of Motion JPEG is that, because it is based on still images, it can produce any of its frames as a single image for identification purposes.

As we will explain, some compression techniques cannot provide such images.

Wavelet Compression
Like Motion-JPEG, wavelet compression delivers high-quality moving images by starting with still images, applying a compression method to them, and putting them together to form moving pictures.

One difference between the two methods is that the mathematics that wavelet uses is up to four times more effective in reducing the volume of data than those used for JPEG and M-JPEG. Like M-JPEG, Wavelet compression represents motion in the video image accurately, because it is based on still images.

For each application there is a suitable compression method ... identification for evidence needs higher quality than queue monitoring

So if wavelet technology produces a smaller image of equal quality to M-JPEG, is it always the better option? Not necessarily.

The wavelet mathematics requires a high processing power. As a result M-JPEG is more widely used, and this is a major selection factor as a relatively new industry tries to determine its standards

MPEG
MPEG and its derivatives are specifically designed for moving pictures, rather than a series of still images. This means that each frame is defined as the previous frame plus changes, rather than a full frame. The advantage of this is that compression is more efficient.

Whilst this may be the preferred method of compression of the internet or video (DVD) Industry it remains impractical for most CCTV applications of any notable size.

As most people have experienced, DVD quality video is particularly impressive.

Recreating these standard pictures for CCTV purposes would require the use of the very latest 2.8Mhz PC and an awful lot of bandwidth – over 7Mbps per channel!

MPEG is evolving as technology advances. The first implementation, MPEG –1, was designed to output 15 frames per second video from limited bandwidth sources, such as CD-ROMs.

Later versions are designed for high bandwidth applications such as High Definition TV (HDTV), delivering 30 frames per second video at full CCIR 601 resolution but requiring special high speed hardware for compression and playback.

Although limited by its high processing and transmission requirements at present, MPEG is the format positioned to become a dominant format for the future.

Network IP Cameras
Having outlined compression techniques, we should look at the devices that use them – cameras and video servers.

Network cameras capture images and transmit them as a digital signal. They are, of course, different to traditional CCTV cameras, which create an analogue signal. They are also considerably different to WebCams, which require a dedicated PC to communicate across an IP network. Network IP cameras are network devices in their own right, able to provide service to a number of viewers with images of up to 25fps via standard web browser software such as Internet Explorer or Netscape, without the aid of a dedicated PC.

If the application needs extended functions to support the camera, then a video server may be required. This is less true as time goes on – the latest generation of network cameras can support conventional auto iris C/CS Mount lenses, motion detection and are now physically small enough to fit in readily available external housings.

However, for the more sophisticated uses such as pan, tilt and zoom, or fully functional domes to be controlled over a network, a video server will be required.

With a large number of different network cameras of increasing levels of functionality, you might assume that they connect to the network in a number of different ways. In fact, it couldn't be easier. For the physical connection, all you need is a single CAT5 cable from the device to a convenient hub or switch location on the network. Once physically connected, the device can be identified to the network using a small software installation program or wizard.

We will deal with the detail of connecting IP networks together later in the series, after the principles of IP have been explained.

Video IP servers
Before outlining the purpose of video servers, we will explain the term "server": A server is a com-puter that is connected to a network to provide services to other computers or devices on the network. So a web server is a computer that stores and displays web pages on a network.

File servers are computers that store and make available data files. And video servers are there-fore devices that serve the network with video images from cameras or similar imaging devices.

Video servers carry out two main functions: First, they can be used to convert an analogue signal from a conventional camera to a digital signal, which is required if the system needs to use the existing analogue cameras, rather than replacing them for IP equivalent cameras. Second, they support third party devices such as dome devices through RS232/485 Data protocols and allow you to propagate images from other machinery – MRI Scanners in a hospital would be a good example. In this application, doctors and radiologists would be able to view the scan in real-time from their PCs without having to leave their office.

Image resolution …
It all depends on the application We have set out the common compression methods, and the functions of the network cameras and video servers. As an installer, you will need to choose which compression method to use.

We have pointed out that there is no "good" or "bad" in compression methods. The idea of "horses for courses" applies.

The table on the opposite page summarises the pros and cons of each of the four methods we have outlined.