This article introduces some of the main technologies of the H.265 video coding standard, some of which are extracted, and there is a standard technical documentation with a H.265.
H.265 is a JCTV organization, approximately three or four months will open a meeting, and all standards formulate members of the members of the respective technical proposals for PK. I am fortunate to participate in the meeting of H.266 standards. The scene is quite fierce. The interests between various companies, JOINT CollaboATITeam on Video Coding are jointly established in ITU and MPEG. They used to set standards independently. Later, these two major organizations have begun to develop more efficient standards, this article summarizes The main technique of H.265, the next blog will be elaborated in detail.
Be
As a new generation of video coding standards, HEVC (H.265) is still a hybrid encoding frame for predicting the increased transformation. However, relative to H.264, H.265 has a revolutionary change in many ways. The technical highlight of HEVC (H.265) is:
1.
Flexible coding structure
exist
H.265
In the size of the macroblock
H.264
of
16
×
16
Extension
64
×
64
To facilitate compression of high resolution video. At the same time, a more flexible encoding structure is used to increase coding efficiency, including coding units.
Coding unit
), Predictive unit (
Predict Unit
) And transform units (
Transform Unit
).
Where the coding unit is similar to
H.264 / AVC
The concept of macroblock in the macroblock is used for encoding, and the prediction unit is a predicted basic unit, and the transform unit is a basic unit for transforming and quantization. The separation of these three units makes the transformation, predicts, and encoding the various processing links more flexible, and also facilitates the division of each link to more in line with the texture characteristics of the video image, facilitating the completion of each unit more optimized.
2.
Flexible block structure - RQT (Residual Quad-tre Transform)
RQT
Is an adaptive transformation technology, this idea is
H.264 / AVC
middle
ABT
(
Adaptive Block-Size Transform
Extension and expansion of technology. For interframe encoding, it allows the size of the transform block to adaptive adjustments according to the size of the motion compensation block; for intra coding, it allows the size of the transform block to adaptive according to the characteristics of the intra predicted residual. adjust. Big block transformations can provide better energy concentrations with respect to small blocks, and can save more image details after quantification, but on the other hand, there will be more ringing after quantification. effect. Therefore, according to the characteristics of the current block signal, the adaptive selection change block size,
It can be obtained from the highest energy concentration, the degree reservation level, and the rings effect of the image.
3. Sample adaptive offset
Sao
In the codewash, located
Deblock
Thereafter, by adding an offset for each type of image pixel value to the reconstructed image, the purpose of reducing distortion is achieved, thereby increasing the compression ratio, reducing the code stream.
use
Sao
After the average can be reduced
2% ~ 6%
Code stream
,
The performance consumption of encoders and decoders increases only
2%
.
4. Adaptive Loop Filter
Alf
In the codewash, located
Deblock
and
Sao
Thereafter, it is used to restore the reconstructed image to achieve a melody between the reconstructed image and the original image (
MSE
The smallest.
Alf
The coefficient is in frame-level calculation and transmission, which can be applied.
Alf
Also can be based on block or quantized trees (
Quadtree
Some areas
Alf
If it is based on partial area
Alf
However, it is necessary to pass additional information indicating regional information. In the late stage of standard, this technology is removed, but at 266 exploration phase, this technology is added to the reference software, I don't know if the first Call for Proposal will not Remove the technology, due to its large complexity, and the subjective performance of the video is small, the objective performance is greatly improved.
5
. Parallelization design
The current chip architecture has gradually developed from single nuclear performance, so in order to adapt to a very high degree of chip implementation, HEVC / H265 introduces many parallel operations, mainly including the following aspects:
(1)
Tile
Divide the image into some rows and columns with vertical and horizontal boundaries, divided into one
Tile
,Every
Tile
Contains integer
Largest Coding Unit
,
Tile
It can be independent of each other and to achieve parallel processing:
(2)
Entropy Slice
Entropy Slice
Allowed in one
Slice
Internal removal into multiple
Entropy Slices
Each
Entropy Slice
Can independent encoding and decoding, thereby improving the parallel processing capability of the codec:
(3)
WPP (Wavefront Parallel Processing)
The second line
LCU
After the processing, the first one is the first
LCU
Entropy encoding (
Cabac
) Probability status parameters are initialized, as shown
5
Indicated. Therefore, only the second line is needed.
LCU
After the codec is completed, you can start the codec of the current line to improve the parallel processing capability of the codec:
6
.
H.264
Improved improvements in
Algorithm complexity relative to H.264, H.265 has greatly improved, with this better compression performance. H.265 has made great improvements in many characteristics, as shown in Table 2:
H.265 Technical documentation can go to JCTVC to download.
Website: http://phenix.int-evry.fr/jct/
Note: Some content from: http://blog.csdn.net/feixiang_john/Article/details/7822830
Our other product:
