Page 112 - Video Coding for Mobile Communications Efficiency, Complexity, and Resilience
P. 112
Section 3.5. The MPEG-4 Standard 89
RVLCs, as illustrated in Figure 3.16(c). In this case, when the decoder jumps
to the next resynchronization codeword, instead of discarding all preceding
bits, the decoder can start decoding in the reverse direction to recover and
utilize some of those bits.
3.5.9 Pro>les and Levels
As already discussed, pro/les and levels provide a means of de/ning subsets
of the syntax and semantics of a standard. This in turn provides a means of
de/ning the decoder capabilities required to decode a particular bitstream. Pro-
/les and levels are used to de/ne conformance points that facilitate bitstream
interchange among di1erent applications.
In MPEG-4, object types are used to de/ne pro/les. An object type de/nes
a subset of MPEG-4 tools that provides a single or a group of functionalities.
There are six natural video object types: simple, core, main, simple scalable,
N -bit, and still scalable texture. For example, the main object type includes
the following subset of tools: basic (I- and P-VOP, coe,cient prediction,
4-MV, and unrestricted MV), error resilience, short header, B-VOP, Methods
1 and 2 for quantization, P-VOP-based temporal scalability, binary shape, gray
shape, interlace, and sprite.
A pro4le is a de/ned subset of the entire bitstream syntax. MPEG-4 de-
/nes six natural video pro/les: simple, core, main, simple scalable, N -bit, and
scalable texture. Each pro/le is de/ned in terms of video object types. For
example, the main pro4le includes the following object types: simple, core,
main, and scalable still texture.
A level within a pro/le is a de/ned set of constraints imposed on parameters
in the bitstream that relate to the tools of that pro/le. For example, level 1
(L1) of the simple pro4le has a typical session size of QCIF, a maximum
total number of objects of 4, and a maximum bitrate of 64 kbits=s.