Tuesday, June 22, 2010

Video Surveillance for Transit Buses - Keeping Up With Technology

(Source: Mass Transit)

Video surveillance is hardly a new idea to most people; in recent times it's increasingly becoming an accepted and even expected part of our environment. There is an ongoing and often lively debate about the pros and cons of video recording itself, regardless of the technology employed. While the practice may have its detractors, suffice it to say there are many organizations, both public and private, that have found video surveillance to be a valuable component of their overall security and risk management programs.

Video surveillance in the transit industry has also been around for many years. Cameras are commonly installed in fixed-location applications such as passenger terminals or train platforms. Within the past 10 years or so, cameras have also been put on board buses and other passenger transportation vehicles. This article will focus on the second category, commonly referred to as 'mobile surveillance'.

Once the decision has been made to implement or upgrade a video system, transit authorities are faced with a deluge of vendors, each promoting their bundle of must-have features. Many organizations choose to employ an external consultant to help them navigate through the sea of information. Whether or not an outside expert is consulted, it is still very important for the purchasers to have a good understanding of the overall technology.

The Basics
For the benefit of those somewhat unfamiliar with video surveillance, here is a brief summary of the basic technology.

The two major components of a video system are cameras and recorders. It is also possible to set up a closed circuit television (CCTV) system with only a camera connected to a video monitor, but the benefits are greatly reduced without the recording. Such systems are typically used simply to provide enhanced viewing range, such as rear-vision cameras for vehicles with obstructed rear views.

When a video system is referred to as 'digital', it commonly means that it is using a digital video recorder, or DVR. Most DVRs store the video recordings on hard drives contained within the DVR. However, the cameras will usually still be analog, as explained below in the section on cameras. Analog video systems use both analog cameras and analog recorders, which would usually be VCRs.

Analog systems are seldom used in transit applications anymore, except as part of legacy installations. The advantages of digital over analog are numerous:

-Longer recording time
-Better picture quality
-Better reliability
-Ability to search recordings very quickly
-Ability to copy and archive without loss of quality
-Ability to transmit wirelessly without loss of quality
-Ability to access video via computer networks over unlimited distance
-Ability to record multiple cameras in high quality
-Easier integration with other digital and analog vehicle systems

Some of these points may not be relevant to all digital systems, but it is clear why they have all but replaced VCRs. Even the cost of the basic digital systems is approaching close to that of the remaining analog products.

DVR Terminology
There are a few terms and related concepts that are key to understanding digital video recording. The first is "resolution." When speaking about DVRs, resolution refers to the level of detail that can be recorded. In the same way as computer monitors, DVR resolution is usually expressed as the number of horizontal and vertical pixels. For example, '720 by 480' means that the DVR will record the video signal as a digital image of 720 rows by 480 columns of pixels. As you would expect, higher numbers mean more detail and a potentially better quality image. Note that better quality is not guaranteed by higher resolution, because other factors may come into play. For example, if the video image coming from the camera is of low quality to start with, the higher resolution recording may not give any advantage.

The second important term is "compression." All DVRs use some type of compression to reduce the amount of data they must store. For example, a single image of 720 x 480 pixels could take up approximately 690,000 bytes of storage in uncompressed form. At that size, an 80GB hard drive could barely store one hour of video. By using compression, this size can be reduced 10 times, a hundred times or even more. This data compression is crucial to allowing efficient storage of video, providing days and weeks of storage instead of just minutes or hours.
Compression and Quality

Most DVRs allow the user to select the amount of compression applied to the video. This setting is often referred to as "quality," because higher compression will result in a loss of video quality. If you choose a higher quality setting, you are reducing the level of compression, and thereby decreasing the storage capacity.

It is very important to remember that video compression does not imply a reduction in resolution. There are many different compression techniques, but reducing the resolution is generally a separate process. However, decreasing resolution can also be an effective way of reducing the storage requirement for video. For example, a 320 x 240 image would require roughly one-quarter the space of a 720 x 480 image.

The reduced quality associated with higher compression can be visible in different ways, depending upon the specific technology used. However, the net effect is the same; lower quality will mean a decreased ability to identify people or objects in the picture. It is difficult to precisely quantify the level of quality or compression you need without referring to a specific model of DVR, since each manufacturer has its own measurement scale. When evaluating and ultimately implementing a video system, it will be necessary to test the equipment under specific, controlled conditions. Only by viewing actual video and comparing the various quality levels will you be able to determine the settings appropriate for your needs.

What About the Cameras? As mentioned earlier, a digital video system does not necessarily use digital cameras. On the contrary, the vast majority of DVRs take their inputs from analog cameras. The typical reason is not because users have replaced VCRs with DVRs and kept their old cameras, though some have done so. Most new installations are still using analog cameras as well, simply because they provide the best combination of price and performance.

Compared to a good quality analog equivalent, the cost is much higher for a digital camera with the same performance. The most common type of digital camera uses a CMOS image sensor. This technology is advantageous for applications such as cell phone cameras, because it consumes less power, and can be implemented with less circuitry. However, CMOS sensors have some disadvantages compared to the analog sensors, the most common of which are knows as CCDs. One big difference is that CCDs perform much better in low light conditions. For bus applications, the ability to record images in low light is obviously very important, since many incidents occur in poorly lighted scenes.

As with DVRs, resolution is also an important term relating to video cameras. For analog cameras, the most widely used measure of resolution does not refer to its horizontal and vertical pixels, though this specification is also available. Analog cameras, and the video signal they produce, are measured in terms of "horizontal lines of resolution." This number is equivalent to how many alternating black and white vertical lines can be resolved in the resulting image. A typical number would be 330 lines for a color camera, and 410 lines for a black and white camera.

Why do cameras use a different measure? Partly because of the history of the technology, but also because the number of lines correspond to the actual detail that can be resolved in the signal. Just because a sensor has, for example, 500 pixels across, this does not mean the video signal you get at the output can reveal a level of detail down to 500 lines of resolution. Other factors come into play, such as how well the circuitry was designed. The same principle holds true for DVRs; the stated resolution will tell you how the video is measured at the input, but you may not be able to see the same level of detail at the output. Just as with evaluating the compression quality, it is important to see the actual output of a DVR before assessing the picture resolution.

How Important is the Camera?
It is interesting to note that most video system vendors talk almost exclusively about DVRs, while barely mentioning cameras. By contrast, users speak about "camera systems." While the DVR is certainly the most complex and costly system part, simple reasoning tells you that the objective of any video system is to produce high-quality images. Furthermore, the old adage about the weakest link in a chain definitely applies; a DVR cannot produce images better than those coming from the cameras.

Technophiles may argue that digital image processing can do wonders to brighten up dark images, remove noise, etc., but that is missing the point entirely. It makes little sense to invest in an expensive, high-resolution DVR and then cripple the system with low-quality cameras. Buyers have many options when it comes to cameras, and shouldn't hesitate to ask the vendors about what features are offered.

There are several options for camera resolution. The numbers of 330 and 410 lines mentioned earlier would be considered "standard" or "medium" resolution. High-resolution cameras offer 450 lines for color and 500 lines for black and white. A new generation of analog sensor, known as HQ, can deliver more than 500 lines of resolution in color, at a reasonable price level. As a reference point, video with 500 horizontal lines is equivalent to DVD quality resolution, which is also approximately equal to 720 x 480 in terms of pixels.

In addition to greater resolution, there are other camera options available. One particularly useful feature is known as the "day/night" function. A day/night camera has the ability to automatically switch its output from color to black and white in low light conditions. By switching to black and white, the camera can achieve much better sensitivity, providing a useful video image under otherwise insufficient lighting.

A related option is infrared illumination. Using small, electrically efficient infrared LEDs, IR illuminators allow black and white or day/night cameras to record useful images in complete darkness. The range of some of these illuminators can easily reach 20 or 30 feet.
The last word about cameras concerns the enclosures. There are a few reasons to look beyond the enclosure as a mere box for a camera. Firstly, there is no reason why a camera cannot be aesthetically pleasing, designed to fit well with the interior look of the bus. Secondly, an enclosure should offer flexible mounting options, allowing the camera to be placed exactly where it is needed, not forcing the view to be compromised. Lastly, the enclosure should be vandal-resistant, discouraging perpetrators from interfering with the surveillance system.

Enhanced DVR Features We have already mentioned that DVRs can offer a recorder resolution of 720 x 480 pixels. This would currently be termed as a "high," or "full" resolution for a DVR. Technically, higher resolutions are possible, but would not be warranted unless there are cameras to match. At the other end of the scale would be a resolution of 320 x 240 pixels, which is close to that of normal TV broadcast quality.

The reason to choose a lower resolution may be cost-driven, though the difference is becoming quite small. A more common concern is the storage requirement. Regardless of the compression used, higher resolution will require more storage. Looking at it another way, with any given size of hard drive, a higher resolution will result in shorter recording capacity. Between 320 x 240 and 720 x 480, the difference may be as much as four times. In some cases, regulations may force an organization to settle for lower resolution in order to meet a minimum time requirement for retaining recorded video.

Real Time Recording
Another term used with DVRs is "frame rate." This specification indicates how many images are recorded every second. A standard video signal in North America, following the NTSC standard, is comprised of 30 images or frames per second (fps). In Europe and other areas using the PAL standard, the number is 25 fps. These rates are sometimes called "real time," though as you can see from the differences in the standards, the numbers are somewhat arbitrary. In fact, they are derived from the frequency of the AC power in a country, i.e. half of the 60 Hz power in North America, and half of the 50 Hz power in Europe.

It is true that people will perceive a series of images presented 25 times per second or faster as continuous motion. However, it is a mistake to think of the 25/30 fps rate as an absolute determination of real time video. If you have ever seen TV footage of Tiger Woods' golf swing, you were looking at high-speed video, recording at much more than 30 fps. The point is that an image rate of less than 30 fps may be quite suitable for transit surveillance. Many people would be hard pressed to differentiate between 20 fps and 30 fps. Once again, evaluate based on real video samples, before specifying "real time" as an absolute requirement. Just as with resolution, a higher frame rate will increase the storage requirements and likely will also increase cost.
Multiple Cameras

The frame rate specification becomes even more important as the number of cameras per DVR increases. Until quite recently, most DVRs were limited to a total of 30 fps aggregate, i.e. across all cameras. For example, with six cameras, each was limited to a maximum of 5 fps. Newer DVRs offer higher aggregate rates of 60 fps, 120 fps, or more. The advantage is the possibility of smoother video, i.e. higher frame rates, for each camera. The downside is increased storage requirements as well as cost.

One way to mitigate the storage problem is through the use of alarm recording. Specifically, alarm inputs can be used to trigger higher quality recording on specific cameras for a limited period of time. For example, when a driver hits the panic button, the alarm can cause the quality setting, the frame rate or other parameters to change during the alarm recording interval. Other alarm inputs can also be used, like inertia sensors, which detect extreme changes in speed such as during an accident.

Improving Storage
The limitation which seems to come up most often is storage. In most video system implementations, agencies are interested in maximizing the storage time. At the very least, there is a forced compromise between recording time and the parameters which affect image quality: resolution, compression and frame rates. Two areas of advancement continue to reduce the amount of compromise required.

The first area is hard drive storage. Driven by consumer demand for more storage, hard drive manufacturers are continually offering larger capacity drives. Every few months larger drives appear, and the prices of existing drives are lowered.

Furthermore, newer DVR designs are able to incorporate larger 3.5-inch hard drives, as opposed to 2.5-inch drives which are more commonly used in mobile applications. The key is that the DVRs must be properly designed to protect the 3.5-inch drives from the increased shock and vibration of the mobile environment. DVRs have typically used 2.5-inch drives because they were designed specifically for more vibration-prone laptop computers; while 3.5-inch drives are intended for relatively stable desktop or server applications.

The second area of major improvement in storage is video compression. With the huge demand for consumer and other video applications, video compression is a hot topic for researchers. Aided by ever more powerful processors, DVRs can squeeze more video into a fixed space, without having to sacrifice overall video quality. MPEG4 is one type of compression which is becoming increasingly popular, because it offers significantly better performance over older methods such as MPEG2, or MJPEG.

A Moving Target
The technology employed by video systems has advanced significantly over the years. Just as with personal computers, we have seen prices decrease while functionality and capabilities have increased. This scenario of ever-improving products at lower and lower prices presents an attractive opportunity for purchasers, but inevitably leads to the vexing question: "When should we buy?" The best advice is the same as for all technology buyers, base your decision upon your requirements, not on future promises. If you really don't need the system today, then it makes sense to wait. However, don't be afraid to commit to a purchase simply because of the fear of obsolescence.

A related question is whether to buy the absolute newest technology on the market. Again, common sense combined with an analysis of your own needs is the best guide. You may buy one or two more years of useful service by investing in the latest and greatest, however, the tradeoff is almost always higher initial costs, coupled with higher technical risk.

Apart from the obvious potential problems of unproven products, there is also the real possibility of unintended obsolescence. Often times manufacturers are forced to choose a technology before clear standards have emerged. In an attempt to be the first to market, vendors will gamble that their chosen path will end up being adopted as the industry standard. With numerous vendors and multiple competing technologies, it is inevitable that some will end up guessing incorrectly.

This article has touched on only some of the key areas of mobile video technology. However, regardless of the features being considered, the approach should be the same. Start with, and stay with, your needs. The flashy new features can be very appealing, but make sure you have fully assessed the risks as well as the benefits.