Graphics Reference
In-Depth Information
NAL units consist of coded slices which, as will be discussed in more detail in
Sect.
3.3.1
below, represent grouped blocks of samples of the video picture. The
non-VCL NAL units contain associated data such as, e.g., parameter set NAL units
or supplemental enhancement information (SEI) NAL units. Parameter sets contain
data that are essential for the decoding process, while the SEI syntax enables the
optional support of supplemental data. Besides the sequence parameter set (SPS)
and the picture parameter set (PPS), as known from H.264
j
MPEG-4 AVC, HEVC
also includes the novel video parameter set (VPS) for conveying additional metadata
about the characteristics of the coded video sequences to be used at the systems
layer.
Each NAL unit consists of a NAL unit header and a NAL unit payload. The two-
byte NAL unit header in HEVC contains information about the type of payload and,
similar to the scalable video coding (SVC) extension of H.264
j
MPEG-4 AVC, a
temporal identifier, which indicates the level in the temporal hierarchical prediction
structure. This information can be conveniently accessed by media gateways, also
known as media-aware network elements (MANEs), for intelligent, media-aware
operations on the stream, such as bitstream thinning using temporal scalability. Note
that this can be achieved without the need of parsing the NAL payload data. The
payload data is also interleaved with emulation prevention bytes when necessary to
ensure that no byte-aligned start code prefix within the NAL payload is emulated.
Details about the NAL concept and parameter sets can be found in Chap.
2
.
3.3.1
Slices and Their Fragmentation into Slice Segments
and Slice Segment Subsets
The high-level segmentation of a picture in HEVC is achieved similar to that in
H.264
j
MPEG-4 AVC based on the slice concept. The slice concept provides a
partitioning of a picture in such a way that each slice is independently decodable
from other slices of the same picture, where decoding refers to entropy, residual,
and predictive decoding. A slice may consist of a complete picture as well as parts
thereof. In HEVC, the minimum block structure unit of a picture contained in a slice
is a single coding tree unit (CTU).
Picture partitioning by slices serves the following three purposes:
1.
Error Robustness
: To partition the picture into smaller self-contained entities in
order to gain error robustness by the ability to re-synchronize both the decoding
and parsing process in case of data losses. This typically implies that the slices
are transported packet-wise, i.e., a loss of a transport packet results in a loss of a
slice.
2.
MTU Size Matching
: To adapt to the network constraint of maximum trans-
mission unit (MTU) size commonly found in IP networks. Such a packetization
scheme is also referred to as MTU size matching and restricts the maximum
number of payload bits within a slice regardless of the size of the coded picture.
