DVRs: For your consideration

Selecting the right equipment in the rapidly changing digital video recorder market can be daunting, yet buyers can protect their investments by carefully analysing a few key factors.

The primary consideration when choosing a DVR is storage capacity. Storage requirements are governed by a number of factors:

• What is the quality of the cameras picking up the images?
• What is the video camera resolution?
• How many frames per second will I be capturing?
• What is the compression type?
• How many days worth of videos will I be storing?

For analogue cameras, image resolution is determined by the number of horizontal video lines that the camera needs in order to reproduce the desired image. Resolution is often referenced in terms of the Common Intermediate Format (CIF) standard. For IP or digital cameras, resolution is measured in digital pixels.

The resolution you choose will depend upon what you are looking to record. If you simply want to capture an image of a car in a car park using an analogue camera, for example, one CIF will do. With one CIF, you would be able to determine if there is a vehicle in the camera's view, but would not be able to capture finer details, such as the license plate number.

Two CIF should be used to record more detailed images, such as a licence plate number or a person's face. However, two CIF also requires twice as much disk space, so it's important to take those storage costs into consideration.

There is a common misconception that four CIF is preferable in all environments. However, while four CIF provides twice the image resolution of two CIF, this is only beneficial for static images due to the limitations in the way a TV picture is transmitted and displayed. A TV picture displays two images, which are a fraction of a second between each other. The first frame image consists of all the odd numbered lines while the second image has all the even lines.

Moving images

A four CIF camera captures images at twice the resolution of a two CIF camera, but if the image is moving then it is capturing these images at slightly different points in time. This is not an issue when watching live video, but it produces a jagged or blurred video image when displaying a freeze-frame. De-interlacing algorithms can be used to remove jagged image artifacts from still 4 CIF images. However, most common techniques achieve the "improved" image results by effectively reducing the resolution to two CIF in the areas of movement. Non-moving image areas still retain the higher four CIF resolution.

Because the ability to freeze a single frame for identification purposes is an important aspect for a CCTV system, this limitation, coupled with higher storage costs, means four CIF is only used in certain security areas.

It is important to also remember there are many different cameras designed for specific purposes and applications. The overall performance of a CCTV system is frequently governed by the quality of the camera attached to it. Using a low quality camera could significantly increase the amount of storage required by the system and possibly defeat the purpose of even trying to record the video.

TV pictures display 30 frames per second. However, most security applications don't require the full 30 frames per second. Typical security cameras record 7.5 frames per second, unless they need to capture subtle details, such as gaming environments that may need to watch for the sleight of hand of a player, for example.

Compressing video

Storing raw (uncompressed) video would require a tremendous amount of disk space, so vendors employ various compression schemes to minimise storage requirements. There are two primary types of compression. Frame based formats, such as M-JPEG, compress individual frames. Change-only formats, such as MPEG, use certain frames as reference frames, recording and transmitting only the information that changes from one frame to the next (such as when a person enters the camera's field of vision).

Change-only formats can minimise hard drive storage requirements more so than frame-based formats. However, if the system will need to record a substantial amount of motion (such as customers entering a bank), the change-only format would have to operate at virtually full-picture rates, negating potential storage benefits.

Processing power

MPEG is an internationally agreed standard created for the entertainment industry. MPEG2 is used on all commercial DVD disks and is capable of very high quality, with a typical movie requiring over 3GB of disc space per hour. MPEG2 is an asymmetrical compression scheme, which means it takes more processing power to compress the video than to play it back.

It works by compressing a key frame and then computing the differences between the key frame and one or more video frames. During decoding the key frame is decoded and the difference information applied to that image to create subsequent frames. Depending on the compression settings, key frames may be generated every five video frames.

The quality of the final video is greatly influenced by how much processing power can be used during the compression stage to calculate difference-data.

MPEG4 is a newer version of the MPEG standard, requiring yet more processing power, but capable of producing much smaller video files than MPEG2. While both MPEG standards are very good at compressing files, they were not designed for the security marketplace and are less than ideal for displaying freeze frame images.

MJPEG is a frame based compression technique used in the first generation of DVRs. It works by compressing each individual video frame using the JPEG standard. The JPEG standard was designed by the Joint Photographic Expert Group as a means for compressing still images. Depending on the compression settings JPEG is also very good at compressing video but typically cannot compress to the same levels as the MPEG standards.

While choosing the right DVR may seem complex, it really comes down to thinking through what the system will be used for...

Active Content Compression (ACC) was developed by American Dynamics expressly for the security market. ACC is a derivative of the MJPEG standard and could be considered a cross between MJPEG and MPEG and provides similar video quality images while using much less disk space. It achieves this by taking advantage of the fact that most security cameras record no motion for large parts of the day (or night) and adapting the compression scheme to suit the type of content being recorded. This is a major benefit in reducing storage costs.

Storage procedures

It's a good idea to ask potential DVR vendors what storage options are available.

DVRs provide a certain amount of internal hard drive space, but generally not enough to store large amounts of video. It's important to plan for offline storage and retrieval. Some companies make backup copies and store them at an offsite location, while others transmit the data via phone lines to another site. Video requires large amounts of space and is relatively expensive to transmit. For example, transmitting multiple camera feeds of high quality video 24/7 from New York to Los Angeles would be cost prohibitive for most companies.

As the industry has slowly moved toward IP-based devices, organisations will have the option to purchase cameras that compress video internally and send the compressed files via IP to storage devices on the network.

This reduces installation costs by leveraging the company's existing computer network infrastructure.

IP is still a relatively new technology that requires more up-front costs than analogue. However, the larger the system, the more cost-effective IP becomes since multiple video streams can be transmitted over a single cable allowing for reduced costs.

For now, analogue still offers better performance and economy for smaller systems.

Ensuring security

The biggest concerns with IP-based cameras are ensuring privacy, security and reliability.

Many IP cameras provide a web interface so employers must ensure unauthorised persons cannot access video they are not supposed to see, or hack in to the system to falsify a video stream.

In addition, Ethernet networks have a 300ft distance limitation which means all devices must be within 300ft of a network hub; a network hub that needs to be protected to avoid providing an easy access point for potential hackers. Since video data also requires relatively large amounts of bandwidth it is unwise to route this across a corporate network where it may impact other core commercial functions.

For these reasons, you may want to consider using a separate IP video network in addition to using a proprietary compression scheme such as ACC which is extremely difficult to replicate or falsify.

Interoperability

Today, surveillance cameras and DVRs can be mixed and matched fairly easily. However, in the IP realm, there are no clear interoperability standards. Just because two devices use MPEG4 does not mean that they will work together. It's important to confirm that the DVR supports the full range of cameras that you are considering, particularly if you anticipate moving to IP.

Additionally, make sure the DVR environment can support networked and non-networked video systems simultaneously.

DVR vendors should have a mechanism to enhance the features of their devices as new functionality becomes available. Inquire about the availability and cost associated with software upgrades.

Specific choices

While choosing the right DVR may seem complex, it really comes down to thinking through what the system will be used for. Talk with companies tasked with similar security issues to learn about their experiences. Finally, look into the history of your potential vendors.

There are many late entrants to the DVR market due to the relatively low development costs and standardised PC platform. Be sure your vendor has the stability to support and enhance their products into the future.