Game Development Reference
In-Depth Information
Fig. 6.8 Region index of a
TB. RegIdx 0 consists of the
top-left CG. RegIdx 1
consists of the rest of the first
row and first column of CGs.
RegIdx 2 consists of the rest
CGs
Here, the RIdx reflects the intraluminance prediction mode and region information,
which is defined as:
(
Pos
+
binIdx
) >
11?
if
(
map
(
IntraLumaMode
) ==
2
)
=
0
: (
Pos
+
binIdx
) >
4?1
:
2
.
RIdx
Cy
+
1
otherwise
2
(6.25)
The context model for chroma blocks is defined as follows:
C Run =
min
(
2
,
RIdx
)
3
+
min
(
2
, (
absSum
+
AbsLevel
) /
2
) .
(6.26)
Therefore, total 30 context models are used for coding Run (with 21 for luma and 9
for chroma).
6.4 Summary
Throughout the AVS2 standardization process, entropy coding was targeting at both
coding efficiency, and throughput improvement. In addition to utilize the position
and level information for context modeling as in AVS1, intramode-dependent context
modeling method is also employed to achieve higher coding efficiency. Besides cod-
ing efficiency improving technologies, new techniques were incorporated to improve
throughput such as coefficient grouping. The final design of CBAC in AVS2 shows
that it delivers high coding efficiency while maintaining relative low area cost.
References
Bjontegaard G, Lillevold K (2002) Context-adaptive VLC coding of coefficients. JVT Document
JVT-C028, Fairfax, VA 19
Fan L, Ma S, Wu F (2004) Overview of AVS video standard. In: 2004 IEEE international conference
on multimedia and expo, ICME'04, vol 1, pp 423-426
 
Search WWH ::




Custom Search