Video Coding Basics
Video signals differ from image signals in several important characteristics. Of
course the most important difference is that video signals have a camera frame rate
of anywhere from 15 to 60 frames/s, which provides the illusion of smooth motion
in the displayed signal.1
Another difference between images and video is the ability
to exploit temporal redundancy as well as spatial redundancy in designing compression
methods for video. For example, we can take advantage of the fact that objects
in video sequences tend to move in predictable patterns, and can therefore be
motion-compensated from frame-to-frame if we can detect the object and its motion
trajectory over time.
Historically, there have been five major initiatives in video coding [1–5] that
have led to a range of video standards.
s
Video coding for ISDN video teleconferencing, which has led to the ITU video
coding standard called H.261 [6]. H.261 is also the baseline video mode for most
multimedia conferencing systems.
s
Video coding for low bitrate video telephony over POTS2
networks with as little
as 10 kbits/s allocated to video and as little as 5.3 kbits/s allocated to voice
coding, which led to the ITU video coding standard called H.263 [7]. The H.263
low bitrate video codec is used at modem rates of from 14.4 to 56 kbits/s, where
the modem rate includes video coding, speech coding, control information, and
other logical channels for data.