"Related Index: https://blog.csdn.net/knowledgebao/Article/details/84776869
content
1. Main differences between H.264 and H.265
2, compression performance comparison
3. Each module technology difference summary
4. Block division structure
5. Frame prediction mode
6. Inter-frame prediction
7. Device filtering
8. Sample AdaptiveOffset filtering point adaptiveoffset
9. TILE
10. WPP
11. Dependentslice
12. Other related technologies
H.265 High Efficiency Video Coding
Technical characteristics
Predictive block size
Internal color
Parallel Processing Tools
Entropy Coding
Intra prediction
Motion Compensation
Sports Vector Prediction Motion Vector PREDIX
Inverse Transforms
Loop filter
Unlock filter
Sample adaptive offset
Coding efficiency Coding EffICIENCY
1. The same image quality and the same code rate, the H.265 is 50% less in storage space occupied by H2.64. 2, if the storage space is as large, then means that H.265 is higher than H2.64 in the same rate, 30% ~ 40%
More different tools are provided by H.264 / AVC, H.265 / HEVC to reduce the code rate, in the coding unit, the smallest 8x8 to the maximum 64x64. There are not many areas of information (color variations are not obvious, such as the red part of the vehicle body and the gray part of the ground) are large, and the coded is less code, and the details of the details (tires) are divided into macro The block has a small and more, the encoded codeword is more, which is equivalent to the encoding of the image, thereby reducing the overall code rate, and the encoding efficiency is correspondingly improved. At the same time, the intra prediction mode of H.265 supports 33 directions (H.264 only supports only 8), and provides better motion compensation processing and vector prediction methods.
Repeated quality comparison tests have shown that under the same image quality, the video code stream encoded by H.265 is reduced by approximately 39-44% compared to H.265 compared to H.264. This data will also have corresponding changes due to differences in quality control measurement. According to the data derived by subjective visual test, the quality of the H.265 encoded video can be approximately even better with the H.264 encoded video in the case of a rate of 51-74%, which is essentially more than expected. The signal to noise ratio (PSNR) is good. These subjective visual tests have covered many disciplines, including psychology and human eye visual characteristics, and video samples are very broad, although they can't be the final conclusion, but this is also very inspiring results.
The current HEVC standard has three modes: Main, Main10 and Main Still Picture. The main model supports 8Bit color depth (ie, 256 chromasings each having a total of 167 million colors). The MAIN10 mode supports 10 bit color depth, which will be used in UHDTVs (UHDTV). The top two will limit the color sampling format to 4: 2: 0. It is expected that the standard will be expanded in 2014, which will support 4: 2: 2 and 4: 4: 4 sampling format (ie, higher color reduction) and multi-view encoding (eg, 3D stereo video coding).
In fact, H.265 and H.264 standards have some overlap in various functions. For example, the HI10P part in the H.264 standard supports 10 bit color deep video. Another part of H.264 (Hi444PP) can also support 4: 4: 4 chrominance sampling and 14-bit deep. In this case, the difference between H.265 and H.264 reflects that the former can use fewer bandwidth to provide the same function, the price is the device computing power: H.265 encoded video requires more computing power To decode. There is already a chip that supports H.265 decoding - US Broadcom has released a Brahma BCM 7445 chip in the CES exhibition in early January, which is a quad-core processor using 28 nano-processes. , Can transfer four 1080p video data streams or resolution H.265 encoding ultra HD video with a resolution of 4096 × 2160
The birth of the H.265 standard is to transmit higher quality network video under limited bandwidth. For most professionals, the H.265 coding standard is not unfamiliar, which is the video coding standard established after ITU-TVCEG followed by H.264. The H.265 standard is mainly surrounded by the existing video coding standard H.264. In addition to some other technologies, it has increased the relationship between improved code stream, coding quality, delay and algorithm complexity. Related technology. The main contents of H.265 include increasing compression efficiency, improving robustness and error recovery capabilities, reducing real-time delays, reducing channel acquisition time and random access delay, reducing complexity.
1. Main differences between H.264 and H.265
Cu (CodingUnit), PU (PREDictionUnit) and TU (TransformUnit)
2, compression performance comparison
Be
3. Each module technology difference summary
Be
4. Block division structure
In H.265, the size of the macroblock is extended from H.264 to 16 × 16 of 64 × 64 to facilitate the compression of high resolution video. At the same time, a more flexible encoding structure is used to increase coding efficiency, including coding units, predictunit, and transformunit.
in:
The encoding unit Cu is similar to the concept of macroblocks in H.264 / AVC for encoding the process.
The prediction unit PU is a basic unit for prediction,
The transform unit TU is a basic unit for conversion 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.
RQT is an adaptive transform technology that extends and expands the extension and expansion of ABT (AdaptiveBlock-size transform) technology in H.264 / AVC.
For inter-frame 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 adjustments according to the characteristics of the intra predictive residual.
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 variable block size, as shown in the figure below, can be obtained by the excellent energy concentration, the degree retention level, and the rings effect of the image.
Flexible block structure diagram
5. Frame prediction mode
Be
6. Inter-frame prediction
7. Device filtering
Essentially H.265's deck filtering process is consistent with H.264's deck filtering process, which has made the following most significant changes:
Ø Filter boundary: H.264 minimum to 4x4 boundary filtering; while H.265 adapts to the filter edge of the latest CU, PU and TU division structure, minimum filtering boundary is 8x8,
Ø Filter order: The vertical boundary is used in the H264 prior macro block, and the horizontal boundary of the current macroblock; and the vertical boundary of the H.265 first frame, the horizontal boundary of the whole frame
The ALF is used to restore the reconstructed image after the DEBLOCK and SAO are in the DEBLOCK and SAO, and the minimum difference (MSE) is minimized between the reconstructed image and the original image.
The coefficient of ALF is to apply ALF in frame-level computing and transmission, or may perform ALF based on blocks or quadtree, and if it is based on part of the ALF, it is necessary to pass indicator area information. Additional information.
8. Sample AdaptiveOffset filtering point adaptiveoffset
SAO is in the codeca, after DEBLOCK, by the classification of the reconstructed image, add an offset to each type of image pixel value, reducing the purpose of reducing distortion, thereby increasing the compression ratio, reducing the code stream.
After SAO, the average can reduce the code stream of 2% to 6%, while the performance consumption of the encoder and decoder increases by about 2%.
9. TILE
10. WPP
Be
11. Dependentslice
12. Other related technologies
Ø Transform_skip mode: transfer_skip_flag, this mode does not transform, but it is quantified, this mode has a better effect on text desktop video
Ø Internal bit depth increase: in order to ensure intermediate prediction, transformation, and internal bits in the quantization process to achieve better compression performance
H.265 High Efficiency Video Coding
H.265 is a video compression format of the high compression ratio developed after ITU-T VCEG followed. H.265 Video Format Standard was officially announced by the International Telecommunication Union (ITU) on January 25, 2013, with the highest resolution up to 8192 × 4320. NGVC wants to reduce the bit rate by 50%, while the main image quality and computational complexity are compared with H.264, the computational complexity is increased from three times. The HEVC faces the next generation of HDTV design, features such as plane scanned, support sample rate to 4320P (8192 × 4320), enhanced dynamic range adjustment and noise suppression.
Figure 2. H.265 encoding flow chart
Technical characteristics
Two-dimensional unsubable adaptive interpolation filter Separable AIF orientation AIF no longer uses motion compensation with 1/8-PEL motion vector Supermacroblock structure to 64x64 conversion (H.264 only 32x32) Adaptive prediction error coding organization (APEC) Adaptive Quantization Matrix Selection (AQMS) Sports Vector Selecting and Coding Competition Method for Individual Coded Module Dependently KLT
Predictive block size
HEVC will use the macroblocks defined in the previous standard with a maximum to 64x64 pixels and can be further subdivided into a variable size block. HEVC becomes the coding tree unit (CTUS) into the coding block of brightness and chroma (CTBS). A CTB can be 64x64, 32x32 or 16x16. This is the size of the PREDICTION UNITS, PU, and the size of the PRETITS, PU is sized from 64x64 to 4x4, but only for two-way predictions. To 8x4 to 4x8 size. The conversion block size of the predicted residual encoding can be 32x32, 16x16, 8x8, 4x4.
Internal color
Internal color depth increase (IBDI) allows the encoder to operate in a higher internal state. IBDI can do a maximum of 14-bit bit wide.
Parallel Processing Tools
Images can be divided into independently decoded rectangular blocks and strips, that is, the concept of strips and Tile porcelain. Most of the stripes can be decoded separately, but ultimately need to synchronize into a video stream. The strip can be encoded as a strip without prediction, independent of each other. Of course, the ribbon may still require loop filtering.
Entropy Coding
The HEVC is similar to the context-adaptive binary archmetic code (Cabac), and H.264 is similar. Only HEVC only supports Cabac encoding.
Intra prediction
HEVC's intra prediction has 33 direction modes, while only 8 in H.264, HEVC also specifies Planar and DC intra prediction mode.
Motion Compensation
The HEVC uses a half-pixel or 1/4 pixel precision motion compensation, and a 7 tap or 8 tap filter. H.264 uses half a pixel precision and 6 tap filters. For 4: 2: 0 video, the chromaticity component has 1/8 pixel precision and 4 tap filters. Weighted prediction in HEVC can be a one-way or two-way forecasts.
Sports Vector Prediction Motion Vector PREDIX
HEVC defines the level of 16-bit and vertical motion vectors, supporting the range to [-32768, 32767], up to -8192 to 8191.75 brightness pixels, H.264 only supports -512 to 511.75 pixel points. HEVC's MV mode has advanced motion vector prediction (AMVP)) and merge mode. The merge mode is running from the neighbor block inheriting the MV vector value, thereby there is SKIP and Direct mode.
Inverse Transforms
The transform block size of the predicted residual encoding in HEVC can be 32x32, 16x16, 8x8, 4x4. A CTB can be recursively divided into 4 or more TUs. TU will use basic transform DCT (Discrete Cosine Transform), and the residual of the 4x4 intra prediction brightness block is adopted from integer transformations derived from DST (Discrete Sine Transform). This is reduced by 1% of the codec with the original 4x4 brightness. The chroma block is used in the same TU size as the brightness block.
Loop filter
HEVC has two loop filters, decapsular filters (DBF, DEBLOCIKING FITER) and sample adaptive offset (SAO, SAMPLE Adaptive Offset filter (DBF). Similar to the DEBLOCKING filter and H.264 / MPEG-4 AVC, the DBF in the HEVC can only be used for blocks of 8x8 (improved parallel processing performance), while H.264 is suitable for 4x4 blocks. The intensity of DBF in HEVC is from 0 to 2. It is perpendicular to the vertical boundary, and vertical filtering is perpendicular to the horizontal boundary. After the SAO filter is after the DBF filter, for better reconstruction of the original image. Each CTB's SAO filter can enable or disable the boundary offset mode or sub-offset mode.
Unlock filter
DBF uses H.264 / MPEG-4 AVC similar design, better support and distribution processing is similar. The DBF in HEVC is only suitable for a 8 × 8 sampling mesh, and a 4 × 4 sampling grid for DBF with H.264 / MPEG-4 AVC. DBF uses an 8 × 8 sampling grid because it causes no obvious degradation and significantly increases concurrent processes because DBF no longer leads to the interaction between cascades and other operations. Another change is that hevc only allows three DBFs that are 0? 2. The DBF that HEVC also needs to be applied to the horizontal filtering of the vertical edge of the screen and only the vertical filtering of the horizontal edge is applied, which allows for multiple concurrent threads of DBF.
Sample adaptive offset
Use the SAO filter after DBF and use offset to produce a better reconstruction of the original signal. There are two modes: edge offset mode or with offset mode. In the edge offset mode, through the value of the comparison, according to the two neighbors, the sample is divided into five categories: the smallest, two edges, the maximum, or both, for each first four categories Apply an offset. The pattern that can be offset can be classified into 32 frequency bands and select four consecutive band transfer offset. The SAO filter is designed to improve image quality and reduce the oscillation effect.
Coding efficiency Coding EffICIENCY
Figure 1 Comparison Video Coding Standard Average Code Rate under Various Video Standards of PSNR Decrease H.264 / MPEG-4 AVC HP MPEG-4 ASP H.263 HLP H.262 / MPEG-2 MP HEVC MP 35.4% 63.7% 65.1 % 70.8% HP - 44.5% 46.6% 55.4% MPEG-4 ASP - - 3.9% 19.7% H.263 HLP - - - - 16.2%
Video coding efficiency is typically an objective evaluation index with Peak Signal-to-Noise Ratio (PSNR). HEVC benefits from a larger Coding Tree Block (CTB) size. The HM-8.0 HEVC video resolution is 2560 × 1600, compared with the 64 × 64 CTB size, if the 32 × 32 CTB size is used, the code rate is increased by 5.7%, and if the 16 × 16 CTB size is used, the code rate is increased by 28.2%. Moreover, the larger the resolution, the greater the size of the CTB size, the more the decoding time is also reduced. The table is HEVC Main Profile (MP) and H.264 / MPEG-4 AVC High Profile (HP), MPEG-4 Advanced Simple Profile (ASP), H.263 High Latency Profile (HLP), and H.262 / MPEG -2 Main profile (MP) encoding efficiency comparison. The test sequence includes 5 HD resolutions and 4 WVGA (800 × 480) resolutions. The results of the main view test showed that the HEVC MP was reduced by 49.3% by the HP Code rate of H.264 / MPEG-4 AVC HP.
Reference:
1, https://blog.csdn.net/firell/Article/details/77827156
2, https://blog.csdn.net/owen7500/Article/details/47334929
3, https://wenku.baidu.com/view/93f15abebb4cf7ec4bfed038.html
If you have any questions, please contact:
[email protected] "
Our other product:
