Another important strategy is to transmit only the parts of a picture that change from frame to frame. Basic frames (complete images) are transmitted a few times each second. Intermediate frames only contain redefinitions of those groups of pixels that need to be updated. A supplementary process, motion prediction, uses a pattern-recognition algorithm in the encoder to identify a repetitive shape (a person's hand, for example) and then look in a subsequent frame for that shape along a path of likely movement.
This algorithm, DCT (Discrete Cosine Transform), is incorporated by most video compression schemes, including H.261, H.263 and the MPEG series (MPEG-1, MPEG-2 and MPEG-4). The schemes differ in the way they handle intermediate frames and the compression ratios they achieve.
In the "Generic Visual Telephone System" block diagram (above), we show Motion JPEG, MPEG and wavelet as possible alternates to the ITU video codec algorithms, H.261 and H.263. This is intended to predict the appearance of new hybrid systems in the market. Intelect Communications, which targets the telemedicine market, offers a unique wavelet codec that "provides near broadcast quality over a switched IP network." But because the new codec requires substantial amounts of transmission bandwidth, Intelect also provides H.320 and H.323 codecs for interoperation with non-Intelect VT systems.
Watch for similar strategies involving MPEG hardware accelerated boards. Array Microsystems, for example, is developing a Communicator board designed to supercharge NetMeeting by intercepting its H.263 and G.711 software codecs and driving a conference with MPEG-1 audio and video.
Picture Sharpness
Another trick in video compression is to gain smoothness of motion by blurring the picture slightly. Reduced resolution isn't obvious in the small windows used by video telephone systems or NetMeeting. While evaluating VT systems for business use, pay close attention to a parameter called the CIF (Common Intermediate Format). This table shows on-screen pixel resolutions for various CIF multiples:
|
|
R = required for this compression algorithm O = optional
CIF, like MPEG-1, produces "VCR-quality" video displays. It is the default format used in room systems and small group videoconferencing systems. If you join a room system conference from a desktop VT set in Quarter-CIF, you'll be seen, but your image will appear grainy and blocky.
The ITU provided in the Visual Telephone standard for "super size" resolutions, 4CIF and 16CIF. While 4CIF is used in some systems for still-picture (document camera) displays, 16CIF is aimed at telemedicine systems that need to transfer radiology images. In the motion picture arena, the ISO/IEC uses slightly different display formats for MPEG and HDTV.
|
REPORTS
ANALYTICS
Follow 
