Graphics Reference
In-Depth Information
The CUs inside a CTU are coded in a depth-first order. This coding order is also
referred to as z-scan and is illustrated in Fig.
3.4
. It ensures that for each CU, except
those located at the top or left boundary of a slice, all samples above the CU and left
to the CU have already been coded, so that the corresponding samples can be used
for intra prediction and the associated coding parameters can be used for predicting
the coding parameters of the current CU.
The horizontal and vertical size of a video picture, in luma samples, has to be
an integer multiple of the minimum CU size,
5
in luma samples, transmitted in the
sequence parameter set, but it does not need to be an integer multiple of the CTU
size. If the horizontal or vertical size of the video pictures does not represent an
integer multiple of the CTU size (as it is, for example, the case in Fig.
3.2
b), the
CTUs at the borders are inferred to be split until the boundaries of the resulting
blocks coincide with the picture boundary. For this enforced splitting no splitting
flags are transmitted, but the resulting blocks can be further split using the quadtree
syntax described above. The CUs that lie outside the picture area not coded.
The CUs represent the processing units to which a coding mode is assigned.
For each CU, it is decided whether the luma and chroma samples are predicted
using intra-picture prediction or motion-compensated prediction. In that respect,
CUs in HEVC are similar to macroblocks in older video coding standards. However,
in contrast to macroblocks, CUs have variable sizes. Multiple CUs form a CTU,
which represents the basic processing unit used for partitioning a picture. For intra-
picture prediction, motion-compensated prediction, and transform coding of the
prediction residuals, a CU can be further split into smaller blocks along the coding
tree structure as will be discussed in the following subsections. The main advantage
of introducing the coding tree structure is that in this way an elegant and unified
syntax is obtained for specifying the partitioning of CTUs into blocks that are used
for intra-picture prediction, motion-compensated prediction, and transform coding.
3.2.3
Prediction Blocks and Prediction Units
For each CU, a prediction mode is signaled inside the bitstream. The predic-
tion mode indicates whether the CU is coded using intra-picture prediction or
motion-compensated prediction. If intra-picture prediction is chosen, one of the 35
supported spatial intra prediction modes has to be selected for the luma CB and
signaled inside the bitstream. If the CU has the minimum CU size specified in the
sequence parameter set, the luma CB can also be decomposed into four equally-
sized square subblocks, in which case a separate intra prediction mode is transmitted
for each of these subblocks. Independent of the CU size, a single chroma intra
5
Videos with picture sizes that do not represent an integer multiple of the minimum CU size can be
coded by extending the picture area using arbitrary sample values and specifying a conformance
cropping window in the sequence parameter set.
