Digital video distribution is a critical function in a variety of defense applications including ship-wide naval distribution systems, local situational awareness systems, airborne distribution systems and simulation gear. An example along those lines is the Navy's Shipboard Protection System (SPS), which provides a fully integrated and seamless sensor, analysis and response system with 360 degree situational awareness and engagement capability. The complexity of such systems is magnified by the recent introduction of higher-resolution, point-to-point digital video/audio interfaces with their restricted transmission distances and cable routing issues. These include the PC standards such as High-Definition Multimedia Interface (HDMI) found in modern flat-panel TVs and monitors. Some military applications achieve uncompressed digital video transmission using the ARINC-818 standard, which is based on Fibre Channel signaling (specifically FC-AV). Video compression and distribution as IP packets over an Ethernet network is a solution being adopted for large platforms and the same principles are being applied to many smaller platforms.

However, compression can cause loss of clarity and detail and introduce delays into signal paths. Delay is usually unacceptable when used for fire direction or for driving a ground vehicle, making the choice of compression algorithm critical. The most familiar standards for video compression, MPEG-2 and MPEG-4, add delays of typically 300 mS by using multiple frames to discriminate changes. There may also be a loss of picture integrity plus some additional recovery time if one frame is lost or corrupted during transmission. The alternative is to use frame-by-frame compression of JPEG2000, where coding and decoding operate at frame rate and any delays are limited to transmission times through a network. JPEG2000 also offers selectable compression ratios for optimum use of bandwidth.

To integrate a successful video distribution system requires a number of elements to be in place. An ideal video distribution solution would handle video resolutions from NTSC/PAL through high-resolution component video up to 2560 x 1600. It would support industry-standard video compression algorithms, network infrastructures and network protocols and bandwidth is critical for high-resolution video.

The components of a digital video system include:

• Video source (for example, SAR image, FLIR, camera or computer display)
• Method for capturing and encoding this video
• Network over which the digitized video is transmitted
• Decoder and regeneration step to reconstruct the video
• Destination where the video will be used-this could be for display, storage or further processing.

JPEG2000

JPEG2000 provides an optimal match of quality, robustness and compression and has been an industry standard since approval by the Joint Photographic Experts Group in 2000. In a deployed video distribution system there is a chance that bit errors may be induced between source and destination or that there may be an interruption in the video stream. It is in these situations that JPEG2000 shows its great error resilience. Compared to JPEG, a bit error in a highly compressed frame of JPEG2000 causes little degradation.

A more extreme difference can be observed when comparing with MPEG2 and MPEG4: bit errors introduce "blocky" artifacts that can have a negative effect on following frames. It is the reliance on compression of sequences of frames in MPEG techniques that lead to their relatively poor performance when the video stream is interrupted-when the stream starts again a significant number of frames can pass before a reasonable image is obtained. In contrast, both JPEG and JPEG2000 are per-frame compression techniques so that an error in a frame affects only that particular frame-subsequent ones are unaffected.

The use of wavelet-based algorithms in JPEG2000 means that much higher compression ratios can be achieved for a given quality requirement; at very high ratios JPEG2000 images may appear blurry as opposed to MPEG and JPEG, which exhibit blocky artifacts. Conversely, when using lower ratios the compression can be virtually lossless if a two-way integer wavelet transform is employed.

In contrast to the discrete cosine transforms employed in MPEG, JPEG2000's wavelet-based approach means it works equally well on frames consisting of continuous gradients as it does on two-tone images. Side channel spatial information is also supported thereby providing support for including transparency levels (if present). Within a typical video distribution system there may be multiple destinations for the video each of which may use video at a different resolution: JPEG2000 allows for progressively increased resolution and quality depending on the amount of the compressed video stream used in decompression.

Another aspect of JPEG2000 that may be of future benefit to systems that employ  very high-resolution sensors is the ability to compress different areas of the screen at different rates-this means that the region of interest can be sent in very high quality with the rest of the frame transmitted in a more compressed form. It is also possible to divide a captured frame of video into strips that can be immediately processed.

Capture, encoding and decoding

A typical video distribution system is comprised of three main elements. There's the Capture, Encoding and Decoding hardware. The Encoder captures and compresses input video, which it then formats and outputs as a network data stream. Meanwhile, the Decoder takes the network data stream and extracts the compressed video, which it then decompresses and re-generates either as uncompressed digital data or, via appropriate interface hardware, as native video.

The other two main elements of a video distribution system are the software and the video processing, recording and display hardware.

The software component of a video distribution system is made up a several pieces. A video distribution client/server framework provides a middleware layer that facilitates the integration of video distribution into arbitrarily complex architectures. Also critical on the software side is RTP (Real-Time Transport Protocol), a real-time network specification that facilitates a standards-based implementation of several of the main capabilities of video distribution. Basic UDP over Ethernet is needed for relatively simple and lightweight implementations with more modest QoS requirements. Meanwhile, software-based sensor and Encoder/Decoder controls facilitate the integration of control-loops into our clients' applications. Software CODECs are included in video distribution for application areas where the real-time capabilities provided by hardware are not needed.

An example of a complete video distribution system solution is Curtiss-Wright's VDS. It uses the JPEG2000 compression algorithm, and enables the transmission of multiple video streams over standard gigabit Ethernet or other suitable network connections in either lossless or lossy modes, thereby providing the flexibility to distribute video to match quality and bandwidth requirements. JPEG2000 compression provides advantages in terms of latency, making VDS ideal in applications where the minimum time from capture to display is critical.

Curtiss-Wright Controls | www.cw.com

Andrew Haylett is Head of Technology and Andrew Hipperson, Business Development Manager, at Curtiss-Wright Controls Embedded Computing, UK

Printer friendly version

Email this article to a friend