Graphics Reference
In-Depth Information
Tabl e 5. 7
Average bit rate savings of HEVC interpolation filter for the
chroma component
BD-rate [%]
RA-Main
LB-Main
LD-Main
Class
Cb
Cr
Cb
Cr
Cb
Cr
A(2560 1600)
10:8
11:3
n/a
n/a
n/a
n/a
B(1920 1080)
10:8
12:2
14:7
16:8
5:8
6:5
C(832 480)
10:3
10:8
13:0
13:5
8:9
9:6
D(416
240)
15:3
17:1
20:4
21:7
16:6
18:1
E(1280
720)
2:4
2:5
5:4
4:7
2:5
2:8
Average
11.7
12.8
13.3
14.3
7.4
8.1
become very large and become more than 20 %. Further experiments show that
close to half of this gain is due to high precision filter operations as described in
Sect.
5.3.1.2
and the rest of the gain is due to improved filter coefficients with longer
tap-lengths.
5.4
Weighted Sample Prediction
Similar to H.264/AVC, HEVC includes a weighted prediction (WP) tool that
is particularly useful for coding sequences with fades. In WP, a multiplicative
weighting factor and an additive offset are applied to the motion compensated
prediction. In principle, WP replaces the inter prediction signal P by a linearly
weighted prediction signal P
D
w
P
C
o,where
w
is an Illumination
Compensation weight and o is an offset. Care is taken to handle uni-prediction
and bi-prediction weights appropriately using the flags
weighted_pred_flag
and
weighted_bipred_flag
transmitted in the Picture Parameter Set (PPS).
Consequently, WP has a very small overhead in PPS and slice headers contain
only non-default WP scaling values. WP is an optional PPS parameter and it may
be switched on/off when necessary. The inputs to the WP process are: the width
and the height of the luma prediction block, prediction samples to be weighted,
the prediction list utilization flags, the reference indices for each list, and the color
component index. Weighting factors
w
0
and
w
1
, and offsets o
0
and o
1
are determined
using the data transmitted in the bitstream. The subscripts indicate the reference
picture list to which the weight and the offset are applied. The output of this process
is the array of prediction sample values. The WP process, for the case when only
the list L0 is used, can be written in a simplified form as:
P ŒxŒy
D
Clip3.0; max_val;P
L0
ŒxŒy
w
0
C
o
0
/
(5.9)
where x and y denote spatial coordinates within the prediction block and max_val
represents the maximum value in the considered bit depth. Additionally, a log
