Graphics Reference
In-Depth Information
no parallel decoding is possible and, therefore, it is quite obvious that there are a
number of further restrictions imposed on slice segment subsets, when used together
with the new high-level parallelization tools in HEVC, as will be further explained
in Sect. 3.3.2 .
3.3.1.3
Slice Segment Header
A lot of syntax elements in the HEVC slice segment header are already known from
the corresponding slice header syntax in H.264 j MPEG-4 AVC. This subsection
gives a brief survey of the main syntax elements in the slice segment header
of both independent and dependent slice segments in HEVC by discussing the
commonalities and differences in relationship to the slice header syntax of its
predecessor.
If the slice segment is not the first slice segment in a picture, which is signaled
by the first_slice_segment_in_pic_flag at the beginning of the slice
segment header, the syntax element slice_segment_address determines the
address of the first CTU in the slice segment in CTU raster scan order, analogous
to the address of the first macroblock in the slice header of H.264 j MPEG-4 AVC.
Note that the first CTU of the first (independent) slice segment in a picture is always
given as the CTU that covers the luma CTB containing the top left luma sample of
the picture.
Each slice segment header refers to the specific picture parameter set that is in
use for the picture to which the slice belongs. The corresponding identifier slice_
pic_parameter_set_id must have the same value for all slices of a given
picture. The referred PPS contains further information such as the referred SPS
and information controlling the presence of particular syntax elements in the slice
segment header.
All above-mentioned syntax elements are present for both independent and
dependent slice segment headers. The discrimination between both slice segment
types is signaled by the dependent_slice_segment_flag , if the corre-
sponding fragmentation functionality is enabled in the PPS. Since for dependent
slice segments only a shortened header is transmitted, all values of the remaining
slice header syntax elements, excluding the entry point offset signaling for slice
segment subsets, are derived from the full slice segment header of the preceding
independent slice segment, instead of being explicitly transmitted.
One of these syntax elements unique to the header of independent slice segments
is the slice_type that indicates the coding type of the slice. This syntax element
is also known from H.264 j MPEG-4 AVC and specifies whether the slice is a
B slice, a P slice, or an I slice, depending on whether the use of bi-predictive, uni-
predictive, or only intra-predictive coding is allowed, respectively.
For all pictures which are not instantaneous decoding refresh (IDR) pictures, the
(independent) slice segment header also contains information that allows derivation
of the picture order count (POC) of the enclosing picture. The POC allows to
identify pictures that needs to be present in the decoded picture buffer (DPB) for
Search WWH ::




Custom Search