CN105049850B - HEVC bit rate control methods based on area-of-interest - Google Patents
HEVC bit rate control methods based on area-of-interest Download PDFInfo
- Publication number
- CN105049850B CN105049850B CN201510130752.4A CN201510130752A CN105049850B CN 105049850 B CN105049850 B CN 105049850B CN 201510130752 A CN201510130752 A CN 201510130752A CN 105049850 B CN105049850 B CN 105049850B
- Authority
- CN
- China
- Prior art keywords
- mrow
- interest
- area
- lcu
- frame
- 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
Landscapes
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
The present invention provides a kind of HEVC bit rate control methods based on area-of-interest, and it comprises the following steps：The spatial domain Saliency maps of present frame are generated according to GBVS models；The time domain Saliency maps of present frame are generated by motion vector information；Time domain and spatial domain Saliency maps are merged to obtain final Saliency maps using based on consistent normalized method；Region division is carried out to current frame image using Saliency maps, is divided into area-of-interest and regions of non-interest；Bit distribution is carried out to area-of-interest and regions of non-interest respectively；Bit distribution is carried out to each LCU in present frame according to conspicuousness；Calculated according to the code check of distributionWith QP values and carry out cutting amendment；Utilize what is finally givenEncoded with QP values.The present invention can improve the subjective quality of encoded video, while accurately control output bit.
Description
Technical field
The invention belongs to multimedia communication technology field, more particularly to a kind of HEVC ((High based on area-of-interest
Efficiency Video Coding, efficient video coding) bit rate control method.
Background technology
In video coding process, the video quality and code check of output are closely related, if it is desired to the video matter of output
Amount is better, then the code check exported will be higher.But due to being limited by bandwidth or memory capacity, it is necessary to by the defeated of video encoder
Go out bit number control in certain scope, so as to while meeting that bandwidth or memory capacity limit, in decoding end as far as possible
Best video image is obtained, control strategy adopted here is exactly Rate Control.Human visual system (Human Visual
System, HVS) it can quickly locate in video on " significant " or " interesting " object, these are then identified again
Object, be primarily due to HVS can under conspicuousness guiding the rapidly object in scan video scene.Rate Control can lead to
Cross and reasonably distribute code check raising video quality.But the foundation of most rate control algorithm distributing bit is the pre- of region
Encoder complexity is surveyed, such as uses MAD (Mean Absolute Difference, mean absolute difference).From the angle of human eye vision
For, it is difficult to the region of prediction not necessarily most can attracting notice.In order to reach best's eye subjective quality, than
Vision mode must be introduced during spy's distribution.
HEVC (High Efficiency Video Coding, efficient video coding) is ISO-IEC/MPEG and ITU-
The video encoding standard of new generation that T/VCEG Liang great International Organization for standardization joint is formulated, it has continued to use conventional video coding standard
Hybrid video coding basic framework, but all improved and reformed on each coding module.With existing Video coding
H.264/AVC, algorithm is compared, and under comparable applications condition and video quality, the code check encoded using HEVC will reduce nearly
Half.Rate control algorithm in itself and is not belonging to a part for video encoding standard, but in the practical application of Video coding
Rate Control plays an important role, and also directly affects the performance of Video coding.People are to every generation Video coding mark
Accurate rate control module has all done substantial amounts of research, formulates and proposes corresponding model and algorithm, for next
The video compression encoding algorithm HEVC in generation, also there is corresponding motion.
Bin.Li of Chinese University of Science and Technology et al. proposes a kind of HEVC rate control algorithm, and wherein lambda is RDO
An important parameter during (Rate Distortion Optimization).R-lambda models such as following formula (1)：
λ=α × R3……………………………………………………………(1)
Wherein α and β is the parameter relevant with encoded video, can be constantly updated with cataloged procedure.R is current encoded frame
Or the target bits of coding unit.After λ is calculated, quantization parameter QP (Quantization are calculated by following formula (2)
Parameter)：
QP=4.2005 × ln λ+13.7122 ... ... ... ... ... ... ... ... (2)
LCU bits distribution link in the rate control algorithm, each LCU target bits TCurrCUIt is calculated as follows
Formula (3)：
In formula, TCurrPicThe target bits for the present frame being calculated before being, BitheaderIt is that the frame originating point information can
The bit number that can be accounted for, it is drawn by the bit number prediction of the header actual consumption of the frame of above encoded same level
, CodedPicIt is the bit number that the encoded LCU of the frame is consumed.ω represents current LCU weight, and its calculating must combine
The above MAD (Mean Absolute Difference) of the LCU at the same position of the frame of encoded same level, specifically
Process such as following formula (4) and (5)：
Although the rate control algorithm achieves more preferable effect on HM platforms compared to conventional algorithm, yet
There is its weak point.Traditional video coding technique is entered mainly for spatial domain redundancy, time-domain redundancy and statistical redundancy
Row compressed encoding, but human visual system's characteristic and psychologic effect are seldom considered, therefore a large amount of visual redundancy data are encoded
And transmit.Document " K.Minoo, T.Q.Nguyen, Perceptual video coding withH.264, Proc.39th
Asilomar Conf.Signals, Systems, andComputers, Nov.2005. " think that HVS can tolerate to a certain degree
Distortion, this depends on the human eye susceptibility different to image different zones.It is theoretical based on this, it should to be done in Rate Control
The region of distortion to more bits are concentrated on human eye easily discovered, less bit is assigned to what distortion was difficult to be aware
Region.And existing Rate Control does not consider this aspect.
The content of the invention
The invention aims to overcome the shortcomings of prior art, there is provided a kind of HEVC code checks based on area-of-interest
Control method, this method can improve the subjective quality of encoded video, while accurately control output bit.
In order to achieve the above object, the present invention proposes a kind of HEVC bit rate control methods based on area-of-interest, its
Comprise the following steps：
Step 1, the spatial domain Saliency maps of present frame are generated according to GBVS models；
Step 2, the time domain Saliency maps of present frame are generated by motion vector information；
Step 3 is final notable using merging to obtain by time domain and spatial domain Saliency maps based on consistent normalized method
Property figure；
Step 4, region division is carried out to current frame image using Saliency maps, be divided into area-of-interest and non-sense is emerging
Interesting region；
Step 5, bit distribution is carried out to area-of-interest and regions of non-interest respectively；
Step 6, bit distribution is carried out to each LCU in present frame according to conspicuousness；
Step 7, λ and QP values are calculated according to the code check of distribution and carry out cutting amendment；
Step 8, encoded using λ the and QP values finally given, according to the data actually obtained to R- after coding
Parameter in lambda models is updated.
Preferably, the step 2 comprises the following steps：
Step 2 11, the block of 16 × 16 sizes is divided the image into, then to each macro block in present image previous
Blocks and optimal matching blocks corresponding to being found in frame, obtain motion vector；
Step 2 12, pass through obtained motion vector computation global motion vector；
Step 2 13, global motion vector is subtracted in each motion vector and obtains the motion of global motion vector compensation
Vector；
Step 2 14, the size of the motion vector compensated according to global motion vector are correspondingly made available each picture in each piece
The significance value of element.
Preferably, the calculating formula of the step 3 is such as following formula：
SF=θ1Sm+θ2Sp+θ3SmSp
Wherein SmRepresent the conspicuousness of each pixel time domain, SpRepresent the conspicuousness in spatial domain, θ1,θ2,θ3It is weight system
Number.
Preferably, step 4 is stated to comprise the following steps：
Step 4 11, obtaining the Saliency maps S of entire imageFAfter, for each LCU in image, it
Conspicuousness is determined using following formula：
Wherein ws (i) is i-th of LCU conspicuousness size, SF(i, m, n) is the picture that i-th of LCU internal coordinate is (m, n)
The significance value of vegetarian refreshments, M and N are the sizes of the corresponding blocks with Saliency maps of current LCU；
Step 4 12, the ws (i) of all LCU in present frame is obtained, to their sequences of progress from big to small, and selected
The ws at a quarter is taken as threshold value T；
Step 4 13, LCUs of the conspicuousness ws (i) more than T will be divided into area-of-interest, and conspicuousness ws (i) is less than T
LCU will be divided into regions of non-interest.
Preferably, the step 5 uses and formula is calculated as below：
T=TROI+TNROI
TROI=K × TNROI
Wherein T, TROIAnd TNROIPresent frame, the area-of-interest of present frame, the regions of non-interest of present frame are represented respectively
Target bit, K is the quality adjustment factor.
Preferably, the step 6 uses and formula is calculated as below：
WhereinWithArea-of-interest and the remaining bit number of regions of non-interest, R (p) and R are represented respectively
(q) target bits of q-th of LCU in p-th LCU target bit and regions of non-interest in area-of-interest are represented respectively
Number, MleftAnd NleftRepresent respectively when encoding current LCU, it is still remaining in area-of-interest and regions of non-interest region to treat
The LCU numbers of coding.
Preferably, the step 7 is in order to keep the uniformity of quality between frame and frame, it should limits resulting λ and QP
Span, calculating formula such as following formula：
λXlastSameLevel·R-1≤λXcurrPic≤λXlastSameLevel·R
QPXlastSameLevel-ΔQP≤QPXcurrPic≤QPXlastSameLevel+ΔQP
Wherein X is area-of-interest either regions of non-interest, and currPic and lastSameLevel represent to work as respectively
Coefficient corresponding to previous frame or a upper frame with present frame ad eundem, R and Δ QP are the coefficients for regulation；
In LCU layers, the scope where λ and QP should be ensured that it is such as following formula：
λlastLCU·R1 -1≤λXcurrLCU≤λlastLCu·R1
QPlastLCU-ΔQP1≤QPXcurrLCU≤QPlastLCU+ΔQP1
λXcurrPic·R2 -1≤λXcurrLCU≤λXcurrPic·R2
QPXcurrPic-ΔQP2≤QPXcurrLCU≤QPXcurrPic+ΔQP2
Wherein X is area-of-interest either regions of non-interest, currLCU and lastLCU represent respectively current LCU and
Coefficient corresponding to a upper encoded LCU, R and Δ QP are the coefficients for regulation.
The present invention compared with prior art, has the following technical effect that：The present invention can improve the subjective matter of encoded video
Amount, while accurately control output bit., being capable of flexible modulation as needed the invention provides a general processing framework
The bit number that control area-of-interest obtains, so as to control the subjective quality of general image.
Brief description of the drawings
Fig. 1 is the flow chart of the HEVC bit rate control methods of the invention based on area-of-interest；
Fig. 2 is the original image of coded sequence Tennis the 207th frame in present example；
Fig. 3 is spatial domain Saliency maps corresponding to coded sequence Tennis the 207th frame in present example；
Fig. 4 is time domain Saliency maps corresponding to coded sequence Tennis the 207th frame in present example；
Fig. 5 is final Saliency maps corresponding to coded sequence Tennis the 207th frame in present example；
Fig. 6 is the 207th frame ROI/NROI division schematic diagrames of coded sequence Tennis in present example；
Fig. 7 is the schematic diagram of the encoded output image of coded sequence Tennis the 207th frame in present example.
Embodiment
Technical scheme is described in further detail below with reference to accompanying drawing.
Exemplified by encoding the cycle tests Tennis of JCT-VC recommendations, HM10.0, encoder GOP are used in present example
For hierarchical B-frame structure, in a GOP B frames account for quantity be 7, I frames or P frames to account for quantity be 1, Video coding frame per second is that 24 frames are per second,
Coding frame number is 240 frames, and target bit rate is arranged to 1000kbps.As shown in figure 1, one kind of the present embodiment is based on area-of-interest
HEVC bit rate control methods be：Pass through GBVS (Graph Based Visual first before the frame of coded sequence one is started
Saliency) model obtains the Saliency maps of the frame spatial domain, then obtains each frame time domain by motion vector information
Saliency maps, two width Saliency maps are merged to obtain final Saliency maps according to certain algorithm.Pass through obtained conspicuousness
Figure carries out region division to present frame and obtains area-of-interest and regions of non-interest.Two regions are carried out with bit point respectively
Match somebody with somebody, most bit resources is concentrated on human eye region interested, while accurate control output bit.The present invention's is specific
Step is as follows：
Step 1, the spatial domain Saliency maps S of present frame is generated according to GBVS modelsp：Specific method and step borrow
Jonathan Hard and Christ of Koch are published in 20th Annual Conference on Neural
The method in article Graph-based visual saliency on Information Processing Systems.Make
Characteristics of image has color characteristic, strength characteristic, contrast metric, Gabor local orientation features, the weight system of various features
Number is equal, thus obtains the Saliency maps S in spatial domainp.It is assuming that current just in coded sequence Tennis the 207th frame, artwork example
As shown in Fig. 2 obtained spatial domain Saliency maps SpExample is as shown in Figure 3；
Step 2, the time domain Saliency maps S of present frame is generated by motion vector informationm, it is comprised the following steps that：
Step 2 11, the block of 16 × 16 sizes is divided the image into first, then each macro block in present image is existed
Blocks and optimal matching blocks corresponding to being found in former frame, displacement spatially between the two is exactly motion vector, here in order to save
Complexity is saved, motion estimation algorithm is used as using Adaptive Path search method；
Step 2 12, by obtained motion vector computation global motion vector, the method for calculating global motion vector
IEEE Trans.on Circuits and Systems for Video are published in reference to Rath G.B. and Makur A.
Article Iterative least squares and compression based estimations on Technology
for a four-parameter linear global motion model and global motion
Method in compensation；
Step 2 13, global motion vector is subtracted in each motion vector and obtains the motion of global motion vector compensation
Vector (GMC-MV, Global Motion-compensated Motion Vector)；
Step 2 14, the GMC-MV size correspondingly significance value as each pixel in each piece, is thus obtained
The Saliency maps S of time domainm, sample result is as shown in Figure 4.
Step 3, using based on consistent normalized method by time domain Saliency maps SmWith spatial domain Saliency maps SpMerge
To final Saliency maps SF, its calculating formula is such as following formula (6)：
SF=θ1Sm+θ2Sp+θ3SmSp…………………………………………………(6)
θ1, θ2, θ3It is respectively set to 0.5,0.3,0.5.Obtained final Saliency maps SFSample result is as shown in Figure 5.
Step 4, region division is carried out to current frame image using Saliency maps, is divided into area-of-interest (Region
Of Interest, ROI) and regions of non-interest (NROI), it is comprised the following steps that：
Step 4 11, obtaining the Saliency maps S of entire imageFAfter, encoded for each maximum in image
Unit (Largest Coding Unit, LCU), its conspicuousness can use following formula (7) to determine：
Wherein ws (i) is i-th of LCU conspicuousness size.SF(i, m, n) is the picture that i-th of LCU internal coordinate is (m, n)
The significance value of vegetarian refreshments.M and N is the size of the corresponding blocks with Saliency maps of current LCU；
Step 4 12, the ws (i) of all LCU in present frame is obtained, to their sequences of progress from big to small, and selected
The ws at a quarter is taken as threshold value T；
Step 4 13, LCUs of the conspicuousness ws (i) more than T will be divided into region of interest ROI, and conspicuousness ws (i) is small
Regions of non-interest NROI will be divided into T LCU.The instantiation result of division as shown in fig. 6, white is ROI region,
Black is NROI regions.
Step 5, bit distribution, its calculating formula such as following formula (8) and formula (9) are carried out to ROI and NROI respectively：
T=TROI+TNROI…………………………………………………(8)
TROI=K × TNROI…………………………………………………(9)
Wherein T, TROIAnd TNROIRespectively represent present frame, the ROI region of present frame, present frame NROI regions target
Bit number, K are the quality adjustment factors, and value is 4 herein.K value is chosen bigger, and ROI quality will be better.
Step 6, bit distribution, its calculating formula such as following formula (10) and formula are carried out to each LCU in present frame according to conspicuousness
(11)：
WhereinWithThe remaining bit number in ROI and NROI regions is represented respectively, and R (p) and R (q) are represented respectively
In ROI in p-th of LCU target bit and NROI q-th of LCU target bit.MleftAnd NleftRepresent and compiling respectively
During the current LCU of code, still remaining LCU numbers to be encoded in ROI region and NROI regions.
Step 7, after ROI, NROI of present frame and each LCU target bits are obtained, according to the code check meter of distribution
Calculate λ and QP values and carry out cutting amendment, i.e., using λ and QP values corresponding to R-lambda models and its calculating of QP calculating formulas, calculate
Formula such as following formula (12) and formula (13)：
λ=α × Rβ………………………………………………………………………(12)
QP=4.2005 × ln λ+13.7122 ... ... ... ... ... ... ... ... ... ... (13)
In order to keep the uniformity of quality between frame and frame, it should the span of λ and QP obtained by limiting, calculating formula
Such as following formula (14) and formula (15)：
λXlastSameLevel·R-1≤λXcurrPic≤λXlastSameLevel·R……………(14)
QPXlastSameLevel-ΔQP≤QPXcurrPic≤QPXlastSameLevel+ΔQP…………(15)
Wherein X can be ROI either NROI.CurrPic and lastSameLevel represents present frame or upper one respectively
Coefficient corresponding to the individual and frame of present frame ad eundem.R and Δ QP is the coefficient for regulation, and value is 2 and 3 respectively.
In LCU layers, the scope where λ and QP should be ensured that it is such as following formula (16), (17), (18) and formula (19)：
λlastLCU·R1 -1≤λXcurrLCU≤λlastLCU·R1………………………(16)
QPlastLCU-ΔQP1≤QPXcurrLCU≤QPlastLCU+ΔQP1……………(17)
λXcurrPic·R2 -1≤λXcurrLCU≤λXcurrPic·R2………………………(18)
QPXcurrPic-ΔQP2≤QPXcurrLCU≤QPXcurrPic+ΔQP2……………(19)
Wherein X can be ROI either NROI.CurrLCU and lastLCU represents current LCU respectively and upper one has been compiled
Coefficient corresponding to the LCU of code, R and Δ QP are the coefficients for regulation, R1With Δ QP1Value is 2 and 3, R respectively2With Δ QP2
Value is 2 respectively2/3With 2.The sample result for encoding the 207th frame of Tennis sequences is as shown in Figure 7.
Step 8, encoded using λ the and QP values finally given, according to the bit number actually obtained to R- after coding
Parameter in lambda models is updated, specific calculating formula following (20), (21), (22)：
αnew=αold+δα·(lnλreal-lnλcomp)·αold…………………(21)
βnew=βold+δβ·(lnλreal-lnλcomp)·lnbppreal……………(22)
Wherein αoldAnd βoldRepresent current parameter value, bpprealRepresent making for average each pixel that actual coding obtains
Bit number, αnewAnd βnewIt is the parameter value after the renewal being calculated, they are used for adjusting the code check of subsequent frame
Control.
Claims (7)
1. a kind of HEVC bit rate control methods based on area-of-interest, it is characterised in that it comprises the following steps：
Step 1, the spatial domain Saliency maps of present frame are generated according to GBVS models；
Step 2, the time domain Saliency maps of present frame are generated by motion vector information；
Step 3, time domain and spatial domain Saliency maps are merged to obtain final conspicuousness using based on consistent normalized method
Figure；
Step 4, region division is carried out to current frame image using Saliency maps, is divided into area-of-interest and non-region of interest
Domain；
Step 5, bit distribution is carried out to area-of-interest and regions of non-interest respectively；
Step 6, bit distribution is carried out to each LCU in present frame according to conspicuousness；
Step 7, λ and QP values are calculated according to the code check of distribution and carry out cutting amendment；
Step 8, encoded using λ the and QP values finally given, according to the data actually obtained to R-lambda moulds after coding
Parameter in type is updated.
2. the HEVC bit rate control methods according to claim 1 based on area-of-interest, it is characterised in that the step
Two comprise the following steps：
Step 2 11, the block of 16 × 16 sizes is divided the image into, then to each macro block in present image in former frame
Blocks and optimal matching blocks corresponding to searching, obtain motion vector；
Step 2 12, pass through obtained motion vector computation global motion vector；
Step 2 13, global motion vector is subtracted in each motion vector and obtains the motion arrow of global motion vector compensation
Amount；
Step 2 14, the size of the motion vector compensated according to global motion vector are correspondingly made available each pixel in each piece
Significance value.
3. the HEVC bit rate control methods according to claim 1 based on area-of-interest, it is characterised in that the step
Three calculating formula is such as following formula：
SF=θ1Sm+θ2SP+θ3SmSp
Wherein SmRepresent the conspicuousness of each pixel time domain, SpRepresent the conspicuousness in spatial domain, θ1, θ2, θ3It is weight coefficient.
4. the HEVC bit rate control methods according to claim 1 based on area-of-interest, it is characterised in that the step
Four comprise the following steps：
Step 4 11, obtaining the Saliency maps S of entire imageFAfter, for each LCU in image, its conspicuousness
Determined using following formula：
<mrow>
<mi>w</mi>
<mi>s</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mrow>
<msubsup>
<mi>&Sigma;</mi>
<mrow>
<mi>m</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>M</mi>
</msubsup>
<msubsup>
<mi>&Sigma;</mi>
<mrow>
<mi>n</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>N</mi>
</msubsup>
<msub>
<mi>S</mi>
<mi>F</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>m</mi>
<mo>,</mo>
<mi>n</mi>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<mi>M</mi>
<mo>&times;</mo>
<mi>N</mi>
</mrow>
</mfrac>
</mrow>
Wherein ws (i) is i-th of LCU conspicuousness size, SF(i, m, n) is the pixel that i-th of LCU internal coordinate is (m, n)
Significance value, M and N are the sizes of the corresponding blocks with Saliency maps of current LCU；
Step 4 12, the ws (i) of all LCU in present frame is obtained, to their sequences of progress from big to small, and choose four
Ws at/mono- is as threshold value T；
Step 4 13, LCUs of the conspicuousness ws (i) more than T will be divided into area-of-interest, and conspicuousness ws (i) is less than T's
LCU will be divided into regions of non-interest.
5. the HEVC bit rate control methods according to claim 1 based on area-of-interest, it is characterised in that the step
Formula is calculated as below in five uses：
T=TROI+TNROI
TROI=K × TNROI
Wherein T, TROIAnd TNROIRespectively represent present frame, the area-of-interest of present frame, present frame regions of non-interest mesh
Bit number is marked, K is the quality adjustment factor.
6. the HEVC bit rate control methods according to claim 1 based on area-of-interest, it is characterised in that the step
Formula is calculated as below in six uses：
<mrow>
<mi>R</mi>
<mrow>
<mo>(</mo>
<mi>p</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msub>
<mover>
<mi>T</mi>
<mo>^</mo>
</mover>
<mrow>
<mi>R</mi>
<mi>O</mi>
<mi>I</mi>
</mrow>
</msub>
<mo>&times;</mo>
<mfrac>
<mrow>
<mi>w</mi>
<mi>s</mi>
<mrow>
<mo>(</mo>
<mi>p</mi>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<msubsup>
<mi>&Sigma;</mi>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<msub>
<mi>M</mi>
<mrow>
<mi>l</mi>
<mi>e</mi>
<mi>f</mi>
<mi>t</mi>
</mrow>
</msub>
</msubsup>
<mi>w</mi>
<mi>s</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
</mrow>
<mrow>
<mi>R</mi>
<mrow>
<mo>(</mo>
<mi>q</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msub>
<mover>
<mi>T</mi>
<mo>^</mo>
</mover>
<mrow>
<mi>N</mi>
<mi>R</mi>
<mi>O</mi>
<mi>I</mi>
</mrow>
</msub>
<mo>&times;</mo>
<mfrac>
<mrow>
<mi>w</mi>
<mi>s</mi>
<mrow>
<mo>(</mo>
<mi>q</mi>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<msubsup>
<mi>&Sigma;</mi>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<msub>
<mi>N</mi>
<mrow>
<mi>l</mi>
<mi>e</mi>
<mi>f</mi>
<mi>t</mi>
</mrow>
</msub>
</msubsup>
<mi>w</mi>
<mi>s</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
</mrow>
WhereinWithArea-of-interest and the remaining bit number of regions of non-interest are represented respectively, and R (p) and R (q) are respectively
Represent the target bit of q-th of LCU in p-th LCU target bit and regions of non-interest in area-of-interest, Mleft
And NleftRepresent respectively when encoding current LCU, it is still remaining to be encoded in area-of-interest and regions of non-interest region
LCU numbers.
7. the HEVC bit rate control methods according to claim 1 based on area-of-interest, it is characterised in that the step
Seven in order to keep the uniformity of quality between frame and frame, it should which the span of λ and QP obtained by limiting, calculating formula are as follows
Formula：
λXlastSameLevel·R-1≤λXcurrPic≤λXlastSameLevel·R
QFXlastSameLevel-ΔQP≤QPXcurrPic≤QFXlastSameLevel+ΔQP
Wherein X is area-of-interest either regions of non-interest, and currPic represents the coefficient corresponding to present frame,
LastSameLevel represents the coefficient corresponding to a upper frame with present frame ad eundem, and R and Δ QP are to be for regulation
Number；
In LCU layers, the scope where λ and QP should be ensured that it is such as following formula：
λlastLCU·R1 -1≤λXcurrLCU≤λlastLcU·R1
QPlastLCU-ΔQP1≤QPXcurrLCU≤QPlastLCU+ΔQP1
λXcurrPic·R2 -1≤λXcurrLCU≤λXcurrPic·R2
QPXcurrPic-ΔQP2≤QPXcurrLCU≤QPXcurrPic+ΔQP2
Wherein X is area-of-interest either regions of non-interest, and currLCU and lastLCU represent current LCU and upper one respectively
Coefficient corresponding to individual encoded LCU, R and Δ QP are the coefficients for regulation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510130752.4A CN105049850B (en) | 2015-03-24 | 2015-03-24 | HEVC bit rate control methods based on area-of-interest |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510130752.4A CN105049850B (en) | 2015-03-24 | 2015-03-24 | HEVC bit rate control methods based on area-of-interest |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105049850A CN105049850A (en) | 2015-11-11 |
| CN105049850B true CN105049850B (en) | 2018-03-06 |
Family
ID=54455990
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510130752.4A Active CN105049850B (en) | 2015-03-24 | 2015-03-24 | HEVC bit rate control methods based on area-of-interest |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105049850B (en) |
Families Citing this family (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017120776A1 (en) * | 2016-01-12 | 2017-07-20 | Shanghaitech University | Calibration method and apparatus for panoramic stereo video system |
| CN105979304A (en) * | 2016-05-31 | 2016-09-28 | 北京奇艺世纪科技有限公司 | Video pushing method and video pushing device |
| CN106157306B (en) * | 2016-06-30 | 2019-05-07 | 北京理工大学 | A kind of method for registering and device based on multiple spectrum pictures |
| CN106358040B (en) * | 2016-08-30 | 2020-07-14 | 上海交通大学 | Code rate control bit distribution method based on significance |
| CN106488241A (en) * | 2016-10-13 | 2017-03-08 | 千目聚云数码科技（上海）有限公司 | A kind of area-of-interest method to set up |
| CN106604031A (en) * | 2016-11-22 | 2017-04-26 | 金华就约我吧网络科技有限公司 | Region of interest-based H. 265 video quality improvement method |
| CN106937118B (en) * | 2017-03-13 | 2019-09-13 | 西安电子科技大学 | A kind of bit rate control method combined based on subjective area-of-interest and time-space domain |
| CN107396108A (en) * | 2017-08-15 | 2017-11-24 | 西安万像电子科技有限公司 | Code rate allocation method and device |
| CN107835422B (en) * | 2017-10-30 | 2021-07-09 | 河海大学 | HEVC multi-description image coding algorithm based on significance |
| CN111492657B (en) * | 2017-12-08 | 2023-06-30 | 上海科技大学 | Light field video compression based on face region detection |
| CN109451316B (en) * | 2018-11-21 | 2020-11-24 | 北京航空航天大学 | QP selection algorithm based on CU significance |
| CN109309834B (en) * | 2018-11-21 | 2021-01-05 | 北京航空航天大学 | Video compression method based on convolutional neural network and HEVC compression domain significant information |
| CN109451309B (en) * | 2018-12-04 | 2022-07-26 | 南京邮电大学 | CTU (China train unit) layer code rate allocation method based on significance for HEVC (high efficiency video coding) full I frame coding |
| CN109922343A (en) * | 2019-04-12 | 2019-06-21 | 杭州电子科技大学上虞科学与工程研究院有限公司 | A method of conspicuousness, which is extracted, from a small amount of user's eye movement data carries out video compress |
| CN110087075B (en) * | 2019-04-22 | 2021-07-30 | 浙江大华技术股份有限公司 | Image coding method, coding device and computer storage medium |
| CN112106362A (en) * | 2019-09-30 | 2020-12-18 | 深圳市大疆创新科技有限公司 | Image processing method and device for movable platform, movable platform and medium |
| CN111866504B (en) * | 2020-07-17 | 2022-06-21 | Oppo广东移动通信有限公司 | Encoding method, encoder and computer readable storage medium |
| CN111918066B (en) * | 2020-09-08 | 2022-03-15 | 北京字节跳动网络技术有限公司 | Video encoding method, device, equipment and storage medium |
| CN112422965B (en) * | 2020-11-16 | 2022-08-30 | 深圳市嬴圳科技有限公司 | Video code rate control method and device, computer equipment and storage medium |
| CN113038127B (en) * | 2021-03-30 | 2022-03-22 | 福州大学 | ROI-based real-time video multi-path parallel transmission code rate control method |
| CN113923398A (en) * | 2021-09-30 | 2022-01-11 | 中国工商银行股份有限公司 | Video conference realization method and device |
| CN115314722B (en) * | 2022-06-17 | 2023-12-08 | 百果园技术(新加坡)有限公司 | Video code rate distribution method, system, equipment and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101325711A (en) * | 2008-07-16 | 2008-12-17 | 上海大学 | Method for controlling self-adaption code rate based on space-time shielding effect |
| CN101827267A (en) * | 2010-04-20 | 2010-09-08 | 上海大学 | Code rate control method based on video image segmentation technology |
| CN103079063A (en) * | 2012-12-19 | 2013-05-01 | 华南理工大学 | Video encoding method of vision attention region under low bit rate |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2013059504A1 (en) * | 2011-10-21 | 2013-04-25 | Dolby Laboratories Licensing Corporation | Hierarchical motion estimation for video compression and motion analysis |
-
2015
- 2015-03-24 CN CN201510130752.4A patent/CN105049850B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101325711A (en) * | 2008-07-16 | 2008-12-17 | 上海大学 | Method for controlling self-adaption code rate based on space-time shielding effect |
| CN101827267A (en) * | 2010-04-20 | 2010-09-08 | 上海大学 | Code rate control method based on video image segmentation technology |
| CN103079063A (en) * | 2012-12-19 | 2013-05-01 | 华南理工大学 | Video encoding method of vision attention region under low bit rate |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105049850A (en) | 2015-11-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105049850B (en) | HEVC bit rate control methods based on area-of-interest | |
| TWI445411B (en) | Method for performing local motion vector derivation during video coding of a coding unit, and associated apparatus | |
| CN103546749B (en) | Method for optimizing HEVC (high efficiency video coding) residual coding by using residual coefficient distribution features and bayes theorem | |
| CN108989802B (en) | HEVC video stream quality estimation method and system by utilizing inter-frame relation | |
| CN103327325B (en) | The quick self-adapted system of selection of intra prediction mode based on HEVC standard | |
| CN105120282B (en) | Domain-dependent code check control bit distribution method when a kind of | |
| CN103517069A (en) | HEVC intra-frame prediction quick mode selection method based on texture analysis | |
| CN105721878A (en) | Image Processing Device And Method For Intra-Frame Predication In Hevc Video Coding | |
| CN107846593B (en) | Rate distortion optimization method and device | |
| CN106604031A (en) | Region of interest-based H. 265 video quality improvement method | |
| CN102204254A (en) | Methods and apparatus for prediction refinement using implicit motion prediction | |
| CN105791877A (en) | Adaptive loop filter method in video coding and decoding | |
| CN100591136C (en) | Video frequency intraframe coding method based on null field decomposition | |
| CN103634601A (en) | Structural similarity-based efficient video code perceiving code rate control optimizing method | |
| CN108810549B (en) | Low-power-consumption-oriented streaming media playing method | |
| CN108769696A (en) | A kind of DVC-HEVC video transcoding methods based on Fisher discriminates | |
| Hu et al. | Fvc: An end-to-end framework towards deep video compression in feature space | |
| CN110677644B (en) | Video coding and decoding method and video coding intra-frame predictor | |
| CN107343202A (en) | Feedback-less distributed video decoding method and mobile terminal based on additional code check | |
| CN114040211A (en) | AVS 3-based intra-frame prediction rapid decision-making method | |
| CN106060555A (en) | Encoder based on multi-core processor | |
| CN100586186C (en) | Quick inter-frame forecast mode selection method | |
| CN109688411B (en) | Video coding rate distortion cost estimation method and device | |
| Liu et al. | Video coding and processing: a survey | |
| CN102595137B (en) | Fast mode judging device and method based on image pixel block row/column pipelining |
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 |