CN104185024B - A kind of HEVC quantization parameter optimization methods based on total bitrate and entropy model - Google Patents
A kind of HEVC quantization parameter optimization methods based on total bitrate and entropy model Download PDFInfo
- Publication number
- CN104185024B CN104185024B CN201410470959.1A CN201410470959A CN104185024B CN 104185024 B CN104185024 B CN 104185024B CN 201410470959 A CN201410470959 A CN 201410470959A CN 104185024 B CN104185024 B CN 104185024B
- Authority
- CN
- China
- Prior art keywords
- quantization parameter
- hevc
- parameter
- optimization methods
- total bitrate
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Abstract
The present invention relates to a kind of HEVC quantization parameters optimization method design based on total bitrate and entropy model, for different video sequence signature, total bitrate and the quantization parameter QP in entropy model adjustment HEVC cataloged procedures are used in rate control process, the purpose for improving distortion performance and more accurate Rate Control is reached, and encoder complexity will not be caused to significantly affect.Described quantization parameter QP optimization methods are equally applicable H.264/AVC, and the QP of the video encoding standard such as AVS is calculated.
Description
Technical field
Patent of the present invention is related to a kind of based on H.265/MPEG-H HEVC (High EfficiencyVideo Coding)
The quantization parameter optimization method of video encoding standard, more particularly to a kind of accurate correction side of video frequency coding rate control process
Method.
Background technology
Video coding with transmission application, high-performance video coding standard of new generation H.265/MPEG-H HEVC (High
Efficiency Video Coding) by regarding that ISO-IEC/MPEG and ITU-T/VCEG Liang great International Organization for standardization sets up
Frequency coding joint development group (JCT-VC) exploitation, compared with H.264/AVC, under identical visual quality, HEVC can make ratio
Special rate reduces half.
As video encoding standard of new generation, HEVC still falls within the hybrid encoding frame of prediction plus conversion, and it is also included
The coding modules such as infra-frame prediction, inter prediction, orthogonal transformation, quantization, filtering, entropy code, but all enter in each coding link
Careful optimization and improvement is gone, HEVC standard coding method is as shown in Figure 1.In video frequency coding rate control algolithm, rate
Distortion R-D (Rate-Distortion) performance is the problem of needing to consider.One good rate control algorithm can be accurate
As far as possible small coding distortion is reached while reaching target bit rate.Rate Control problem can be converted to such as formula (1) institute
The rate-distortion optimization problem shown, the situation of target bit will be no more than by this optimization problem encoder in number of coded bits
Under, selection makes the parameter of distortion minimization as optimal coding parameter.Wherein { Para } presentation code parameter sets, including mould
Formula, movable information, quantization parameter QP (Quantization Parameter) etc..
λ in formula (1) is Lagrange multiplier, represents R-D slope of a curve absolute values.Video coding provides very big
Encoding flexibility, encoder be free to select the combination of various coding parameters.The different parameter of selection can be to finally regarding
The coding bit rate of frequency produces very important influence.Therefore, rate control algorithm can cause encoder in some discrete conjunctions
Suitable coding parameter is selected in the coding parameter sets of method, and then reaches target bit rate.
HEVC carrys out the accurate R-D code rate distortion models portrayed in encryption algorithm using hyperbolic model.Such as formula (2) institute
Show, wherein D represents the video distortion after compressed encoding;R represents the bit rate after compression, with every pixel consumption bit bpp
(bit per pixel) is unit;C is the model parameter related to sequence characteristic to K, different video sequence C, K value
It is different.
D (R)=CR-K (2)
In Rate Control, HEVC employs a kind of novel λ domains rate control algorithm based on R- λ models.This
In rate control algorithm, on the basis of R-D code rate distortion models by the code check R and Lagrange multiplier λ that uses of coding it
Between set up mathematical relationship, and reach desired target bit rate using the method for adjusting λ., can be by this as shown in formula (3)
Formula calculates Lagrange multiplier λ, wherein α=CK, β=- K-1.Therefore the two parameters of α and β are also related to the characteristic of sequence,
Different sequences have different values.
Code check R and λ relations are further obtained by formula (3), shown in such as formula (4).
Understand that code check R is determined by Lagrange multiplier λ completely by formula (4).The relation schematic diagram of λ and R-D curves is such as
Shown in accompanying drawing 2.λ is the R-D slope of a curve absolute values determined by the convex closure network of all real work points, and code check R and glug are bright
One-to-one relationship is there is between day multiplier λ.Because R-D curves are convex functions, based on some λ value computational minimization formula
(1) it is equivalent to and goes to approach R-D curves for the straight line of λ value using slope absolute value, and this straight line only can be with R-D contact of a curve in one
Point.Therefore, λ value can determine code check R and video distortion D.
In rate control process, HEVC distributes conjunction according to the situation of occupying of buffering area in the coding unit to each rank
The bit of suitable quantity, generally includes picture group GOP (Group of Pictures) level, picture level and elementary cell level (Coding
Unit).In order to reach some target bit rate R distributed, encoder will determine associated λ value according to formula (3), and by its
For cataloged procedure.After the λ value that coding is used is determined, every other coding parameter all should be by rate-distortion optimization RDO
(Rate-Distortion Optimization) is determined.
QP is one of coding parameter to be optimized in rate-distortion optimization problem.It can generally be optimized by many QP
The mode of (Multiple-QP Optimization) determines optimal QP.In many QP optimizations, generally using formula (5) as excellent
Change target.
{QPiBe QP to be selected set.Ideally, { QPiThe QP values of all permissions can be included, can in HEVC
The QP values of choosing are included from 0 to 51 52 totally.But if encoder attempts all possible QP values, it can greatly increase coding side
Complexity.Even if the method therefore optimized using many QP, also only can rule of thumb include limited several QP values.
Because coding side complexity and the QP quantity used are directly proportional, in order to not increase the situation of coding side complexity
Under, i.e., lift the distortion performance of coding in the case where optimizing without using many QP, experiments of the HEVC based on different cycle tests,
Using the linear relationship between optimal QP values and ln (λ), the QP values used in coding are modified, such as formula (6) institute
Show.Then encoded using revised QP values.
QP=a × ln (λ)+b (6)
In formula (6), slope a is 4.2005, and intercept b values 13.7122, the two parameters are based on different tests
Sequence, the average value after experimental fit.In fact, for the video sequence of different content, the two parameters should be adaptive
Value is answered, and obviously there is certain limitation using the value mode of this fixation in an encoding process.
The content of the invention
It is an object of the invention to provide a kind of HEVC quantization parameter optimization methods based on total bitrate and entropy model, sheet
Algorithm can be directed to different video sequences in rate control process, adaptively accurately be counted with entropy model using total bitrate
Quantization parameter QP values are calculated, the distortion performance of coding is lifted.Designed equally applicable other of quantization parameter QP optimization methods are regarded
The QP of frequency coding standard is calculated.
A kind of HEVC quantization parameters optimization method based on total bitrate and entropy model of Patent design of the present invention is used
Following scheme is realized:(as shown in Figure 3)
Wherein λ be Lagrangian, R be total bitrate, H be residual error code check, parameter ε,It is Laplce for constant, Δ
Distributed constant.
Further, described parameter ε,For B frames and P frames, ε,Different values are taken,
Further, described Δ is laplacian distribution parameter, the video sequence of different Δ value to embody different characteristic
Row.
Further, the QP of described other equally applicable video encoding standards of quantization parameter QP optimization methods is calculated.
Brief description of the drawings
Fig. 1 is HEVC video encoding standard structured flowcharts.
Fig. 2 is the relation schematic diagram of λ and R-D curves in Rate Control.
Fig. 3 is the quantization parameter optimization method of Patent design of the present invention.
Embodiment
Below in conjunction with the accompanying drawings and design principle illustrates the technical scheme that patent of the present invention is related to.
In HEVC, the total bitrate after coding is residual information entropy and side information code check sum, therefore except residual information
In addition, side information also has large effect to total bitrate, especially at low bit- rate.When high code check, usual quantization step
QstepSmaller, total bitrate R and residual error code check H is relatively;And in low bit- rate, quantization step QstepValue is than larger, total code
Rate R then differs larger with residual error code check H, and at this moment information code check in side is just relatively notable.
Conversion residual error generally obeys the laplacian distribution of zero-mean, if Δ is laplacian distribution parameter, then ln (R/H)
With Δ QstepIt can be approximately linear relationship, shown in total bitrate and entropy model such as formula (7), or be write as formula (8) institute
Show.Wherein, parameter ε,For constant, for B frames and P frames, ε,Different values are taken, shown in such as formula (9).The computational methods of Δ
As shown in formula (10), σ is the standard deviation of conversion residual error, embodies the content characteristic of current video.
In HEVC, quantization step QstepRelation such as formula (11) with quantization parameter QP is shown.
Qstep=(21/6)QP-4 (11)
Simultaneous formula (8) and (11), you can obtain shown in quantization parameter QP such as formula (12), due to different Laplces
Distributed constant Δ embodies the video sequence of different characteristic, thus the QP optimization methods designed by the present invention compared with primary standard more
Function with adaptive different video.
In actual cataloged procedure, for current target bit rate R, it is necessary to prediction residual code check H and Laplce point
Cloth parameter, Δ.Because vision signal has very strong correlation in time, change very between frame and frame in the short period of time
It is small, therefore the average of three frames encoded before use of the present invention predicts the residual error code check of current i-th frameWith La Pula
This distributed constantAs shown in formula (13) and formula (14).
In addition, in low bit- rate, the selection frequency of SKIP patterns can more increase, so as to cause the residual error in formula (12) to be believed
Entropy is ceased close to 0.Because side information is set up based on residual information entropy, therefore, the model can cause side information to be estimated at this moment
Meter is less than normal so that final code check model is inaccurate.All encoded at all pieces by SKIP patterns and cause residual information entropy
For under 0 extreme case, QP at this moment, which is just used conventional methods, to be calculated, shown in such as formula (6).
Because using temporal correlation, come prediction residual code check H and laplacian distribution parameter, Δ, the present invention also needs to examine
Consider the situation of scene change.Using the standard deviation of present frame and previous reconstructed frame difference whether scene change is produced to embody.Such as
Shown in formula (15), wherein IiFor current i-th frame,The reconstructed frame of the i-th -1 frame is represented, (x, y) is pixel index value, W and H points
Not Wei video width with height.
When scene change, σiChange is just relatively more violent, therefore uses the σ of present frameiWith former frame σi-1Rate of change
To judge whether to produce scene change, whenDuring less than threshold value 0.3, it is considered as video scene content and has changed, QP at this moment is just
Use conventional methods and calculated, shown in such as formula (6).
In summary it is described, the present invention carry out quantization parameter optimization when, first determine whether all coding units whether be
SKIP patterns or scene change, and are calculated if using the optimization of the quantization parameter based on total bitrate and entropy model without if
Method, otherwise obtains QP, shown in total optimization method such as formula (17) using traditional computational methods.After the completion of present frame coding
The residual error code check H and laplacian distribution parameter, Δ of renewal present frame are needed, for the coding of next frame.Due to different content
Vision signal, its residual error coefficient distributed constant Δ is also different, therefore, and the QP optimization methods designed by the present invention can be adaptive to
Different video sequences, obtains more preferable distortion performance, and the QP of other equally applicable video encoding standards of the method is calculated.
Claims (3)
1. a kind of HEVC quantization parameter optimization methods based on total bitrate and entropy model, it is characterised in that:In HEVC videos
In the rate control process of coding standard, for different video sequence signature, total bitrate and entropy model adaptometer are used
Quantization parameter QP values are calculated, the quantization parameter QP optimization methods based on HEVC video encoding standards include:
Quantization parameter QP optimization methods:
Wherein λ be Lagrangian, R be total bitrate, H be residual error code check, parameter ε,It is laplacian distribution for constant, Δ
Parameter;
Described parameter ε,For B frames and P frames, ε,Different values are taken,
2. a kind of HEVC quantization parameter optimization methods based on total bitrate and entropy model according to claim 1, its
It is characterised by:Described Δ is laplacian distribution parameter, the video sequence of different Δ value to embody different characteristic.
3. a kind of HEVC quantization parameter optimization methods based on total bitrate and entropy model according to claim 1, its
It is characterised by:Described quantization parameter QP optimization methods are equally applicable H.264/AVC, and the QP of the video encoding standard such as AVS is calculated.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410470959.1A CN104185024B (en) | 2014-09-16 | 2014-09-16 | A kind of HEVC quantization parameter optimization methods based on total bitrate and entropy model |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410470959.1A CN104185024B (en) | 2014-09-16 | 2014-09-16 | A kind of HEVC quantization parameter optimization methods based on total bitrate and entropy model |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104185024A CN104185024A (en) | 2014-12-03 |
CN104185024B true CN104185024B (en) | 2017-09-15 |
Family
ID=51965716
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410470959.1A Active CN104185024B (en) | 2014-09-16 | 2014-09-16 | A kind of HEVC quantization parameter optimization methods based on total bitrate and entropy model |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104185024B (en) |
Families Citing this family (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10742708B2 (en) | 2017-02-23 | 2020-08-11 | Netflix, Inc. | Iterative techniques for generating multiple encoded versions of a media title |
US11166034B2 (en) | 2017-02-23 | 2021-11-02 | Netflix, Inc. | Comparing video encoders/decoders using shot-based encoding and a perceptual visual quality metric |
US11153585B2 (en) | 2017-02-23 | 2021-10-19 | Netflix, Inc. | Optimizing encoding operations when generating encoded versions of a media title |
US11184621B2 (en) | 2017-02-23 | 2021-11-23 | Netflix, Inc. | Techniques for selecting resolutions for encoding different shot sequences |
US10666992B2 (en) | 2017-07-18 | 2020-05-26 | Netflix, Inc. | Encoding techniques for optimizing distortion and bitrate |
CN107592535B (en) * | 2017-08-18 | 2020-09-01 | 西安邮电大学 | H.265/HEVC image layer code rate control method |
CN109120934B (en) * | 2018-09-25 | 2020-06-23 | 杭州电子科技大学 | Frame-level quantization parameter calculation method suitable for HEVC video coding |
CN113438479B (en) * | 2021-06-13 | 2022-09-20 | 北方工业大学 | Code rate control based on CTU information entropy |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8693535B2 (en) * | 2009-12-14 | 2014-04-08 | Thomson Licensing | Method and apparatus for bit allocation in offline video coding |
CN101835042B (en) * | 2010-03-19 | 2013-01-23 | 西安电子科技大学 | Wyner-Ziv video coding system controlled on the basis of non feedback speed rate and method |
WO2012162806A1 (en) * | 2011-06-01 | 2012-12-06 | Zhou Wang | Method and system for structural similarity based perceptual video coding |
CN103888765B (en) * | 2014-03-27 | 2015-09-30 | 天格科技(杭州)有限公司 | A kind of modified model R-λ model H.265 bit rate control method |
-
2014
- 2014-09-16 CN CN201410470959.1A patent/CN104185024B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN104185024A (en) | 2014-12-03 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104185024B (en) | A kind of HEVC quantization parameter optimization methods based on total bitrate and entropy model | |
CN104796705B (en) | A kind of HEVC Video codings rate-distortion optimization and rate control algorithm based on SSIM | |
US10070135B2 (en) | Encoding/decoding apparatus and method using flexible deblocking filtering | |
CN103945222B (en) | Code rate control model updating method based on HEVC standards | |
CN102334337A (en) | Apparatus and method for coding/decoding image selectivly using descrete cosine/sine transtorm | |
US20110142124A1 (en) | Perceptual-based video coding method | |
CN104333761B (en) | A kind of HEVC elementary cells level bit-rate distribution method | |
CN107257464B (en) | High-definition video coding rate control method based on Sobel operator and linear regression | |
CN103281530A (en) | HEVC (High Efficiency Video Coding) rate control method based on rate distortion optimization | |
CN104767999B (en) | A kind of HEVC Rate Controls model parameter more new algorithm based on distortion measurement | |
CN103929652A (en) | Intra-frame prediction fast mode selecting method based on autoregressive model in video standard | |
CN103916675B (en) | A kind of low latency inner frame coding method divided based on band | |
WO2018095890A1 (en) | Methods and apparatuses for encoding and decoding video based on perceptual metric classification | |
CN104754335B (en) | A kind of code rate controlling method for video coding | |
CN102364950A (en) | H.264/advanced video coding (AVC)-standard-based intra-frame prediction mode rapid selection method and device | |
CN104038769A (en) | Rate control method for intra-frame coding | |
Ringis et al. | Per Clip Lagrangian Multiplier Optimisation for HEVC | |
CN105187832B (en) | Based on 2.5G wireless network mobile video bit rate control methods | |
Yan et al. | Adaptive distortion-based intra-rate estimation for H. 264/AVC rate control | |
Zhao et al. | An SSIM-motivated LCU-level rate control algorithm for HEVC | |
KR100807330B1 (en) | Method for skipping intra macroblock mode of h.264/avc encoder | |
KR101630871B1 (en) | Method and apparatus for deciding intra prediction mode | |
CN104320660B (en) | Rate-distortion optimization method and coding method for lossless video encoding | |
EP1940176A2 (en) | Method and device for coding a sequence of digitized images | |
Li et al. | Optimal frame-level bit allocation in HEVC with distortion dependency model |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
TR01 | Transfer of patent right |
Effective date of registration: 20190916 Address after: 350002 Room 203, Xihong 397, Hongshan Town, Gulou District, Fuzhou City, Fujian Province Patentee after: Fuzhou Shichi Technology Co., Ltd. Address before: Minhou County of Fuzhou City, Fujian province 350108 Street Town Road No. 2 University City School District of Fuzhou University Patentee before: Fuzhou University |
|
TR01 | Transfer of patent right |