Information Technology Reference
In-Depth Information
Table 7
Experimental condition
- Class A: (2560x1600)
Traffic (30 fps)
- Class B: (1920x1080)
Kimono1 (24 fps)
Test sequence
(YUV4:2:0, 8bpp)
- Class A: 300 frames
- Class B: 240 frames
Number of frames
-IPPP
- I picture interval :
Class A: 28 frames
Class B: 24 frames
GOP composition
- I picture: 16x16
- P picture
Proposed: 32x32
Anchor: 16x16
MB size
- Motion search method: EPZS
- Motion search range:
128 pixels, 1/4-pixel accuracy
- Intra/Inter switching ON
- Block partition
Proposed: rectangle and diagonal
Anchor: only rectangle
Motion compensation prediction
- Class A: 25, 29, 33, 37 (fixed)
- Class B: 25, 28, 31, 34 (fixed)
Qp
ON
R-D optimization
CABAC
Entropy Coding
coding bits are shown also in Fig 58. Moreover, the improvement of coding effi-
ciency is depicted as a comparison of R-D curve with AVC/H.264 is shown in
Fig. 59.
6.3.1.3 Other Possibilities of Performance Improvements for UHD Video
Coding
In addition to the improvement of KTA tools and the expansion of the block size
for UHD video, other possibilities of performance improvement of AVC/H.264
are assumed as follows.
(1)
Examination of pre filter processing.
(2)
The design of the effective probability table of arithmetic coding in CABAC.
(3)
Adaptation of wavelet transform to Intra picture.
(4)
Dynamic Vector Quantization.
