Game Development Reference
In-Depth Information
All technologies for image and video coding are building up based on the
assumption of consistent sampling. If this assumption is broken in the future, then we
might design more efficient systems for image and video representation. However,
other different problems may arise and need to be dealt with.
Consistent sampling an image means to digitize that analog image by equal step
length in the horizontal direction, and in the vertical direction. For example, when
we say a camera image is in resolution of 4,000
2,000, we mean that the camera
image is acquired by sampling by 4,000 equal steps in horizontal and by 2,000 equal
steps in vertical, and sometimes it can be called 2,000 lines and 4,000 rows. It is clear
that this consistent sampling may bring many redundancies in case of the view of
the image being very simple. This kind of redundancy is called spatial redundancy,
since it is due to oversampling in the spatial domain.
For a video, it is a sequence of images, so the process of digitization for video
has no difference with that for image in the spatial domain. Consistent sampling
for a video means to digitize that analog image sequence by the equal step length
in horizontal and in vertical just like above, and by the equal step length between
images in the temporal direction. It is obvious that this consistent sampling may
bring redundancy in case there are few changes between images (e.g., the view is an
announcer in reading news). This kind of redundancy is called temporal redundancy,
since it is due to oversampling in the temporal domain.
Spatial and temporal redundancies are the major parts of redundancies we need
to remove in the task of video coding. There are also some other redundancies, such
as human vision perceptual redundancy and coding scheme redundancy, which will
be discussed later. A good video coding system should consider all these kinds of
redundancies together, to design a tool set which should be a global optimal solution
and not only good or the best for some specific redundancies.
There are so many factors that influence the forming of today's video coding field.
Among them, two threads can be looked into if readerswant to find out themainstream
development of this field: technology and standardization. New technology is the
driving force for research from academic society, and standardization is the driving
force for production from both industry and academy. We will look into the history
of video coding along with these two threads, technology and standardization.
×
1.2 History of Video Coding Technology
Video data usually contains a lot of information representing visual signals in a
period of time. The contents in video data not only include motion objects, but also
still scenes, which may be composed of smooth areas or complex textures with color
information. Without compression, the large volume of video data will exceed the
capacity of any available storage devices easily. For example, one of the formats
defined for HDTV (High-definition Television) in the United States is 1,920 pixels
horizontally by 1,080 lines vertically, at 30 frames per second. If these numbers
were all multiplied together, along with 8 bits for each of the three components,
the total volume of the data for a film about 90min would exceed 930GB, which
