If your application requires digital video recording, you can design a homegrown solution or, rather than starting from scratch, begin with a board-level DVR. This decision may be complex, depending on what your objectives are and how you set out to accomplish them.

First you must determine what your specific application requires. For example, you may be starting with some form of an analog video signal. Simply recording your analog signal onto tape for archiving is now widely considered an outdated concept, and with the continued decline of memory costs, the economic justification of remaining tied to a tape-based storage process is also considered outdated.

Begin with the input signal. In the analog context, your input signal is probably S-Video, composite, or component. So the first decision that will need to be made is whether or not you will change the signal itself by replacing the camera, or if converting that signal to a digital format would be suitable. Given the capabilities, costs, and image quality available today with most commercial analog-to-digital converters (ADCs), there are few instances where the input itself would need to be digital as opposed to analog. An ADC is an electronic circuit that converts continuous signals into discrete digital numbers.

The type of ADC you select will greatly influence the quality of the image recorded and subsequently viewed. In order to ensure the integrity of the images you are recording, use ADCs with YUV digital color space, 60 fields/second NTSC, 50 fields/second PAL, and pixel resolution of 720 x 486 (525/60) and 720 x 576 (625/50). Otherwise, it will be impossible to deliver high-quality recorded images.

Note that the conversion of analog signals to digital does require a bit more effort than simply selecting an ADC. It's easy to underestimate the task of decoding analog video into digital components, and then manipulating the output of a video encoder into a high-quality analog signal.

Adapt reference designs with care
Careful handling of the analog signal, especially the A to D conversion step, is fundamental to high picture quality. Losses at this stage are the single biggest culprit in degrading the image. While the danger is loss of resolution and signal distortion, a more insidious issue is that high-frequency noise can easily be injected, which the downstream compression encoder sees as high-frequency picture information. This effect exacts an extremely high cost in reduced compression efficiency, increasing the data rate by twenty percent or more, and subsequently exacerbating compression artifacts. In bandwidth-limited applications, the side effect is increased quantization and further reduction of high frequencies to keep the data rate on target. A picture compressed from a noisy input is visibly degraded when compared to a picture compressed to the same size from a clean input.

Monolithic analog video decoders and encoders are available from several well-known vendors, along with reference designs that neatly specify values for supporting discrete components. Regrettably, this cookbook approach is deceptively simple and often provides poor results. For better or for worse, the legacy composite video signal is packed with idiosyncrasies. Frankly, intimate knowledge of the interdependencies of subcarrier, sync, luminance, and chroma belongs to a tiny number of individuals who have invested the time to learn the details, spending countless hours in real-world video situations playing witness to common problems and workarounds.

When applying a reference design to new hardware with the goal of maximum picture fidelity, the designer must navigate the pitfalls of selecting high-performance components and judiciously enhancing or sometimes discarding discrete filtering. During the all-important layout phase, one must pay fanatical attention to such details as component placement, power supply filtering, and proper use of analog and digital planes in close proximity to one another. At this point, the real work begins. After power-up, the engineer's familiarity with video is put to the test. The first issue is adjusting component values and layout. The second is judging the effects of adapting the silicon's bewildering register settings for optimum results in the context of a unique circuit board.

At this point you have taken your analog signal and converted it to digital. Next you need to determine how much data you will need to store. This is where video compression decisions become extremely important. By its very nature, digital video requires high data rates, and the higher the quality of the image, the more data that is required. Video compression is a process that enables a digital video signal to use much less data. There are many forms of compression available today, each with their own advantages and disadvantages. Though we do not have the time to get into all the differences between compression methods in this discussion, it is worth noting that the terminology used to describe some of these compression methods can be a bit misleading.

Next: Motion JPEG Compression One example of such misleading terminology is the use of the word "lossy." Algorithms such as JPEG are perfectly capable of producing images that are indistinguishable from their source. While in fact some data may be irretrievable after compression, in most cases this loss of data is irrelevant as long as no difference can be detected in the image either with test equipment or with the naked eye. By contrast, combining luminance and chrominance into a composite video signal has a profound negative impact on the luminance resolution. While not termed a "lossy" operation, the fact that chrominance irretrievably obliterates all luminance detail around the subcarrier frequency of 3.58 MHz (NTSC) does not concern many video users who insist on lossless compression. Picture quality would be far superior if the same user specified S-Video as the analog source, and used "lossy" compression at half the data rate of lossless compression.

The compression technique chosen will have a direct correlation to image quality and storage requirements, so it is important to know what level of image quality and integrity is important to you. In addition to Wavelet, Cinepak, Fractal compression, and other proprietary techniques, two of the more popular compression methods used today are MPEG (Motion Picture Experts Group) and JPEG (Joint Photographic Experts Group). Given JPEG was originally created for the image compression of naturalistic still pictures, and MPEG for the purpose of standardizing video compression, you would think an MPEG compression technique would be the obvious choice for recording video. However, it turns out that a video version of JPEG, called motion JPEG, actually provides a number of desirable results that cannot be obtained with other compression techniques. JPEG achieves extremely high image quality, which is arguably indistinguishable from the uncompressed original even at 5:1 compression, and continues to maintain image quality even as compression increases.

Another important benefit of motion JPEG is that every frame is a stand-alone image, so any frame can be extracted as a still picture. Not to mention that while JPEG is a common standard, the ability to view and enhance a JPEG picture is ubiquitous using freely available software on any personal computer. Also, since motion JPEG is less complex electronically to encode than MPEG, a video recorder that uses it can be made smaller and more power efficient.

Figure 1: Motion-JPEG allows access to every frame

As mentioned, though each compression method has it own advantages and disadvantages depending on the application, the fundamental difference between JPEG and MPEG schemes is that JPEG compresses each frame independently. MPEG exploits the similarity in sequential video frames to gain additional compression by recording only the differences from one frame to the next between periodic complete key frames.

Motion JPEG's lack of inter-frame dependency means that no details are omitted from frame to frame. Motion JPEG also provides the ability for your encoder to operate as a time-lapse recorder, multiplex with several video sources, and has the option of variable compression rates for the ultimate in flexibility.

Storage media
Having briefly focused on ADC and compression techniques, we turn next to the physical storage media, location, and memory size for your digital recorder. To narrow it down, first determine if you will need to have removable media, hot-swappable drives, or if a built-in hard drive will do. Will you need to be able to download this data from your device, and if so what interface will be used?

The type of media you select will depend largely on how you plan to archive your new video images. If previously you were simply removing analog tapes, labeling them, putting them in a box, and shipping them off to a storage location, would it make sense to now store your digital images on removable media and then continuing with the labeling and shipping to a storage facility? Probably not. So the real question is how will you transfer your video images to your server or network?

Only you can determine what works best in your environment, but in terms of flexibility, using a standard 2.5" laptop hard drive is an effective option or, for more rugged environments, an off-the-shelf compact flash would be a very effective option. Fueled by huge consumer demand, compact flash is advancing rapidly in density while costs are plummeting. Of course, if you desire removable media, you should also plan to include an ejector mechanism, which allows pushbutton removal of the compact flash card. That way, any user can exchange the compact flash in the same manner as a video tape in a conventional recorder. Lastly, the solid-state construction of compact flash means your recorder will have no moving parts. This is a tremendous benefit in portable, mobile, and body-worn devices.

File management
While having your analog images converted and stored digitally may appear to be the goal, it is really only the beginning. Significant thought and effort should be given to managing all this data. File management is perhaps one of the most over-looked and misunderstood aspects of recording digital video.

To start, what data format should be selected? The most popular today, which any reasonably up-to-date computer is able to directly play back, is QuickTime. In fact, QuickTime viewer software is installed on many computers by default. If not already installed, the standard QuickTime player is available for download at no cost, and the majority of the more sophisticated editing software packages support QuickTime.

For removable media, using a standard PC file system so that files are readily stored on the hard disk is an unquestionable advantage, especially when needing to copy files for analysis or archiving. Using a standard PC file system provides straightforward access to video files with no special software.

A USB 2.0 interface can be used as an alternative to, or in conjunction with, removable media. This simple plug-in connection provides high-speed access to your data while providing a very high data rate of 800 megabits per second. USB 2.0 ports are a standard feature of all recent PCs and are now commonly used with mass storage devices with built-in support provided by Windows XP. So when connecting your device, you should have it set up for Windows to see it as just another hard disk. This will allow for easy playback of your video from the recorder with a simple mouse click, or you can choose to copy the video to the PC or network.

Next: DVR Block Diagram and Boards
Figure 2: Hardware block diagram

In Figure 2, you can see the block diagram outlining the core architecture of one of Fast Forward Videos board-level products, known as Outrider. This patent-pending architecture shows a method that we have utilized to manage and process large amounts of data by limiting the use of the processor.

Most designs use the processor to simply move data around. This design uses the processor for its intended purpose, to process data so it can move without touching the processor. This is partly accomplished through our memory optimization scheme and file system, both optimized for file efficiency using arbitrary random access.

Figure 3: FFV Outrider products provide two ways to easily remove and transfer recorded material. The Outrider CF (top) is equipped with compact flash cards, while the Outrider IDE (bottom) utilizes 80-GB, laptop-sized, 2.5-in. IDE hard drives.

There are many hidden challenges and development costs involved in developing a digital video solution. When comparing the cost of developing a homegrown solution to the cost of utilizing existing ones, sometimes it makes more sense to capitalize on the efforts and successes of others.

Video solutions from Fast Forward Video are backed by a twenty-year investment in designing compact, ultra-rugged products that provide the highest-quality images. Our recorders have always been designed for a discriminating audience of customers in a wide array of challenging applications including on-body, on-vehicle, underwater, and even outer space. Fast Forward Video recorders are highly compatible with other equipment, and can be completely integrated to meet the requirements of any recording system.

About the author
Scott Keating is VP of Sales and Marketing for FFV. Previously he served as group sales manager leading the Imaging Division for Panasonic North America, and prior to that, he served as an OEM sales engineer for Saft America, Inc. He can be reached at skeating@ffv.com.