CN107493475A - Quantization level capture device and method - Google Patents
Quantization level capture device and method Download PDFInfo
- Publication number
- CN107493475A CN107493475A CN201710823352.0A CN201710823352A CN107493475A CN 107493475 A CN107493475 A CN 107493475A CN 201710823352 A CN201710823352 A CN 201710823352A CN 107493475 A CN107493475 A CN 107493475A
- Authority
- CN
- China
- Prior art keywords
- quantization
- quantization level
- interception
- scope
- capture device
- 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.)
- Granted
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/124—Quantisation
- H04N19/126—Details of normalisation or weighting functions, e.g. normalisation matrices or variable uniform quantisers
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/132—Sampling, masking or truncation of coding units, e.g. adaptive resampling, frame skipping, frame interpolation or high-frequency transform coefficient masking
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
- H04N19/18—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being a set of transform coefficients
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/44—Decoders specially adapted therefor, e.g. video decoders which are asymmetric with respect to the encoder
Abstract
The present invention provides a kind of quantization level capture device and method.According to multiple embodiments of the present invention by adaptively intercepting quantization level after quantization to avoid the spilling of the conversion coefficient after quantifying.In one embodiment, methods described includes:The conversion coefficient of converter unit is quantified according to quantization matrix and quantization parameter, to produce the quantization level of conversion coefficient.It is determined that interception condition, and according to interception condition interception quantization level, to produce the quantization level after intercepting process.Interception condition includes empty interception condition.Quantization level interception under the conditions of empty intercept is the fixed range represented with n positions, and wherein n is corresponding to 8,16 or 32.Quantization level can also be intercepted under the conditions of sky intercepts in the range of by m to m 1, and wherein m may correspond to 128,32768 or 2147483648.
Description
The relevant information of divisional application
It is on December 14th, 2012, Application No. 20128005463.6 and entitled " quantization electricity the applying date that this case, which is,
It is truncate to take device and method " invention patent application case divisional application.
Technical field
The invention relates to Video coding, more specifically, the invention relates to efficient video coding (High
Efficiency Video Coding, HEVC) quantization level (quantization level) interception (clipping).
Background technology
HEVC is by Video coding joint group (Joint Collaborative Team on Video Coding, JCT-
VC a kind of) new International video coding standard of exploitation.HEVC is used based on mixing block (hybrid block-based)
Transition coding (the DCT- of the similar discrete cosine transform (Discrete Cosine Transformation, DCT) of motion compensation
Like transform coding) framework.Elementary cell for compression is referred to as coding unit (Coding Unit, CU), is
2N × 2N squares, and each CU can (recursively) cutting be recursively four smaller CU, until reaching predetermined minimum
Size.Each CU includes the variable predicting unit (Prediction Unit, PU) of one or more resource block sizes and converter unit
(Transform Unit,TU).For each PU, intra-frame image prediction or inter frame image prediction are selected.Each TU carries out space
Block transform processing, then quantifies to each TU conversion coefficient.The minimum TU sizes that HEVC allows are 4 × 4.
The quantization of conversion coefficient plays an important role in the bit rate of Video coding and quality control.Quantization step collection shares
In being quantization level by quantization of transform coefficients.Larger quantization step size can cause relatively low bit rate and relatively low quality.
On the other hand, less quantization step size can produce higher bit rate and higher quality.The direct implementation of quantification treatment
Mode includes cutting operation, and cutting operation is more complicated in hardware based embodiment, and in the embodiment party based on software
More computing resources can be consumed in formula.Correspondingly, this area develops the various technologies for freely splitting quantification treatment.
In HEVC test models version 5 (HEEV Test Model Revision 5, HM-5.0), quantification treatment is described as follows.Parameter
Set is defined as:
The bit wide (bit width) or bit depth (bit depth) in B=input videos source
DB=B-8
N=TU transform size
M=log2 (N)
Q [x]=f (x), wherein f (x)={ 26214,23302,20560,18396,16384,14564 }, x=0 ..., 5,
And IQ [x]=g (x), wherein g (x)={ 40,45,51,57,64,72 }, x=0 ..., 5.
Q [x] and IQ [x] are referred to as being referred to as quantization step and inverse quantization step-length.Quantization operation performs according to following formula:
Qlevel=(coeff*Q [QP%6]+offset)>>(21+QP/6-M-DB), wherein
Offset=1<<(20+QP/6–M–DB), (1)
Wherein " % " is modulus operator (modulo operator).Inverse quantization operation performs according to following formula:
CoeffQ=((qlevel*IQ [QP%6]<<(QP/6))+offset)>>(M-1+DB), wherein
Offset=1<<(M-2+DB). (2)
Variable qlevel in formula (1) and formula (2) represents the quantization level of conversion coefficient.Variable coeffQ in formula (2)
Represent inverse quantization parameter.IQ [x] instruction inverse quantizations step-length (also referred to as inverse quantization step size), and QP represent quantization parameter.
" QP/6 " in formula (1) and formula (2) represents the later integer parts of QP divided by 6.As shown in formula (1) and formula (2), quantization and inverse
Change operation by carrying out arithmetic shift (arithmetic again after multiplication of integers (integer multiplication)
Shift) implement.Deviant (offset value) is added in formula (1) and formula (2), is rounded (rounding) with use
To implement integer conversion (integer conversion).
For HEVC, the bit depth of quantization level is 16 (including 1 bit flag positions).In other words, quantization level is expressed as 2
Byte (byte) or 16 words (word).Due to IQ (x)<=72 and QP<=51, IQ [x] dynamic range are 7, and "<<
(QP/6) " operator performs left side arithmetic shift to 8.Correspondingly, inverse quantization conversion coefficient coeffQ (i.e. " (qlevel*IQ
[QP%6])<<(QP/6) dynamic range ") is 31 (16+7+8).Therefore, the inverse quantization operation described by formula (2) will not be led
Cause to overflow (overflow), because inverse quantization operation uses 32 data representations.
However, when introducing quantization matrix, inverse quantization operation may be modified such that formula (3) to formula (5) Suo Shi:
IShift=M-1+DB+4. (3)
If (iShift>QP/6),
CoeffQ [i] [j]=(qlevel [i] [j] * W [i] [j] * IQ [QP%6]+offset)>>(iShift-QP/6),
Wherein offset=1<<(iShift-QP/6-1), wherein i=0 ... nW-1, j=0..nH-1 (4)
Otherwise
CoeffQ [i] [j]=(qlevel [i] [j] * W [i] [j] * IQ [QP%6])<<(QP/6-iShift) (5)
Wherein, " [i] [j] " indicates the position (also referred to as pointer) of conversion coefficient in converter unit, and W indicates quantization matrix,
NW and nH is the width and height of conversion.If the dynamic range of the quantization level of n representation transformation coefficients, dynamic range n are necessary
Meet following condition to avoid overflowing:
n+w+iq+QP/6-M+DB-3≤32, (6)
Wherein w is quantization matrix W dynamic range, and iq is IQ [x] dynamic range, and the bit depth of inverse quantization or again
The conversion coefficient built is 32.
If quantization matrix W dynamic range is 8, when working as QP=51, M=2 and DB=0, formula (3) to formula (5) is retouched
The dynamic range of the conversion coefficient for the reconstruction stated is changed into 34 (16+8+7+3).When inverse quantization operation uses 32 data representations
When, the conversion coefficient of the reconstruction obtained according to equation formula (3) to formula (5) may overflow and cause thrashing.Therefore, it is necessary to
A kind of conversion coefficient re-establishment mechanism is developed, to avoid possible spilling.
The content of the invention
In view of this, the present invention provides a kind of quantization level capture device and method.
Avoid quantifying becoming by adaptively intercepting quantization level after quantization according to multiple embodiments of the present invention
Change the spilling of coefficient.In an embodiment of the present invention, this method is included according to quantization matrix and quantization parameter to converter unit
Conversion coefficient is quantified, to produce the quantization level of the conversion coefficient of converter unit;Based on quantization matrix, quantization parameter, regard
Frequency source bit depth, the transform size of converter unit or its any combination, it is determined that interception condition, wherein interception condition includes empty intercept
Condition;And according to interception condition interception quantization level, to produce the quantization level after intercepting process.The quantization level is cut
The quantization level after the interception that may correspond to unconditional fixed range, and interception is taken to represent that wherein n can be right using n positions
Should be in 8,16 or 32.The interception of the quantization level can also correspond to the interception of unconditional fixed range, and quantization level
Intercepted in the range of by-m to m-1, wherein m may correspond to 128,32768 or 2147483648.
Quantization level intercept method according to an embodiment of the invention, suitable for video encoder, the quantization level is cut
Method is taken to include:The conversion coefficient of converter unit is quantified according to quantization matrix and quantization parameter, to produce the conversion
The quantization level of coefficient;Determine to be applied to fixed range interception condition and dynamic range interception condition in the video encoder
In one or combination interception scope;The quantization level is intercepted according to the interception scope, after producing intercepting process
Quantization level;And wherein, under the conditions of fixed range interception, the quantization level is truncated to-m to m-1 fixation
Scope is intercepted, and m corresponds to 128,32768 or 2147483648.When it is determined that the interception scope is to be applicable the dynamic range
During interception condition, the determination of the interception scope is based on the quantization matrix, the quantization parameter, inverse quantization step-length, video
Source bit depth, the transform size of the converter unit or above-mentioned any combination.Based on the video source bit depth, by the quantization
Level interception is the first scope or the second scope.The interception scope is by the way that the first weighted value is made comparisons to determine with threshold value
, wherein, first weighted value corresponds to the quantization matrix, the quantization parameter, the video source bit depth, the change
Change the transform size of unit or the first linear function of above-mentioned any combination.The threshold value corresponds to fixed value or the second power
Weight values, wherein second weighted value corresponds to the quantization matrix, quantization parameter, the video source bit depth, described
The transform size of converter unit or the second linear function of above-mentioned any combination.The fixed range intercepts the described of condition
Intercept the video source bit depth that scope is applied to be more than or equal to 10.The quantization level after intercepting process uses n positions
To represent.N can correspond to 8,16 or 32.
Quantization level capture device according to an embodiment of the invention, suitable for video encoder, the quantization level is cut
Device is taken to include:The conversion coefficient of converter unit is quantified according to quantization matrix and quantization parameter, to produce the conversion
The device of the quantization level of coefficient;Determine that be applied to fixed range interception condition cuts with dynamic range in the video encoder
Take the device of the interception scope of one in condition or combination;The quantization level is intercepted according to the interception scope, to produce
The device of quantization level after intercepting process;And wherein, under the conditions of fixed range interception, the quantization level is cut
- m to m-1 fixed interception scope is taken as, and m corresponds to 128,32768 or 2147483648.
Computer-readable recording medium according to an embodiment of the invention, computer program is stored, described program is in quilt
In being realized during computing device the step of above-mentioned quantization level intercept method.
Quantization level capture device according to an embodiment of the invention, suitable for video encoder, the quantization level is cut
Device is taken to include:Processor, is coupled to memory, and the processor performs in program stored in performing the memory
The step of above-mentioned quantization level intercept method.
It is that conversion coefficient can be avoided using quantization level capture device provided by the present invention and one of method, its advantage
Overflowed in inverse quantization or reconstruction operation.
Brief description of the drawings
Fig. 1 show the interception conversion coefficient with reference to one embodiment of the invention to avoid the preferable flow chart overflowed.
Embodiment
As it was previously stated, when introducing quantization matrix, above-mentioned parameter inverse quantization (or reconstruction) operation may overflow.For
Avoid rebuilding the phase in conversion coefficient and potentially overflow, limited according to multiple embodiments of the present invention before inverse quantization operation is performed
Determine the quantization level of conversion coefficient.The dynamic range of the quantization level of conversion coefficient is represented by Integer n.In formula (3) to formula (5) institute
In the embodiment of description, if inverse quantization (or reconstruction) conversion coefficient uses 32 data representations, n dynamic range is no more than 32
Position.Correspondingly, n needs to meet following limit:
n+8+7+(QP/6-(M-1+DB+4))≤32, (7)
It is derived there
n≤20+M+DB-QP/6. (8)
In this case, the quantization level qlevel of conversion coefficient can be intercepted according to formula (9):
Qlevel=max (- 2n-1,min(2n-1-1,qlevel)) (9)
In order to avoid overflowing, the dynamic range of the quantization level of conversion coefficient needs to be limited according to formula (8).According to formula
(8), n is necessarily less than or avoids overflowing equal to (20+M+DB-QP/6).However, because quantization level in the present embodiment is by 16
Position represents (that is, bit depth=16 of quantization level), and therefore, n is no more than 16.Correspondingly, if (20+M+DB-QP/6) is more than
16, then it is a scope for being no more than 16 data representations that the quantization level of conversion coefficient, which needs to intercept,.Following pseudo-code (pseudo
Codes) (pseudo-code A) display quantization level according to an embodiment of the invention that conversion coefficient is intercepted during conversion coefficient is rebuild
Qlevel with avoid data overflow example:
Pseudo-code A:
As shown in pseudo-code A, two interception scopes are used for two different interception conditions.First interception condition corresponds to " 20+
M+B-8-QP/6 >=16 ", and the second interception condition correspond to " 20+M+B-8-QP/6<16”.First interception scope corresponds to solid
Fixed interception scope, i.e., (- 215,215- 1), and the second interception scope corresponds to (- 220+M+DB-QP/6-1,220+M+DB-QP/6-1).When
When test condition " if (20+M+DB-QP/6 >=16) " is used for above-mentioned preferably pseudo-code A, other test conditions can also be used.For example,
The bit depth B of video source can be used to replace parameter DB in test condition.Test condition is changed into " if (20+M+B-8-QP/6>=
16) ", i.e. " if (12+M+B-QP/6>=16) ".Corresponding pseudo-code (pseudo-code B) is changed into:
Pseudo-code B:
If the bit depth of video source is 8 (DB=0), and transform size is 4 × 4, and formula (8) can be reduced to:
n≤22-QP/6 (10)
Therefore, test condition " if (12+M+B-QP/6 >=16) " in this case is changed into " if (22-QP/6 >=16) ".
Test condition can be further simplified as " if (QP<=36) ".Therefore, for the video source with fixed dynamic range, root
QP is only dependent upon according to the intercept operation of the quantization level of the conversion coefficient of another embodiment of the present invention.As follows is one preferably pseudo-
Code (pseudo-code C):
Pseudo-code C:
When the bit depth of video source is 10 or higher, i.e. DB >=2, the condition in formula (7) always meets.In this respect
Under, 16 interceptions, that is, qlevel=max (- 215,min(215- 1, qlevel)) or qlevel=max (- 32,768, min
(32,767, qlevel)), it is always unconditional to use.When being equal to 10 or higher unconditional execution interception for bit depth, become
Changing the quantization level of coefficient also can unconditionally be intercepted to the bit depth of required bit depth rather than video source.Required locating depth
Degree can be 8,16 or 32, and corresponding interception scope can be [- 128,127], [- 32768,32767] and [-
2147483648,2147483647]。
Described as above with reference to three preferable pseudo-codes of the embodiment of the present invention.These pseudo-codes are intended to indicate that in conversion coefficient
The preferable operation for avoiding data to overflow during reconstruction.Other test conditions can be used to realize the present invention for those skilled in the art.
For example, compared to test " if (QP<=36) ", it is possible to use test condition " if (QP/6<=6) ".In another embodiment, cut
Other functions can be used to implement for extract operation, and such as cutted function clip (x, y, z), wherein variable z is intercepted between x and y
(x<y).Pseudo-code C intercept operation can be expressed as:
Qlevel=clip (- 215,215- 1, qlevel), and
Qlevel=clip (- 221-QP/6,221-QP/6-1,qlevel)
In above-mentioned multiple embodiments, design parameter is used to illustrate with reference to the multiple embodiments of the present invention to avoid data from overflowing
Inverse quantization operation.Used design parameter is not construed as limiting the invention.Those skilled in the art can be based on offer
Parameter changes the test of interception condition.For example, if inverse quantization step-length has 6 dynamic ranges rather than 7 dynamic ranges, formula
(8) limitation is changed into n≤19+M+DB-QP/6.Corresponding interception condition test in pseudo-code A is changed into " if (19+M+DB-QP/6>
=16) ".
Above-mentioned quantization level intercept operation performs in decoder end, and quantization level intercept operation can also compiled after quantization
Code device end performs.Overflowed in order to avoid potential, change is limited after quantization operation is performed according to multiple embodiments of the present invention
Change the quantization level of coefficient.Interception condition can be based on quantization matrix, quantization parameter, inverse quantization step-length, video source bit depth, conversion
The transform size of unit or its any combination.Interception condition may also comprise be not provided with interception condition sky cut (null) take condition.
In other words, empty condition corresponds to and always intercepts quantization level to the unconditional interception of a scope.In one embodiment, measure
Change level can under the conditions of the first interception interception to the first scope, and quantization level can under the conditions of the second interception interception to the
Two scopes.First scope may correspond to the fixed range relevant with quantization level bit depth, and the second scope can be with quantifying electricity
Flat dynamic range is relevant.Interception condition can be by the way that the first weighted value be made comparisons to determine with threshold value, wherein the first weighted value
It is linear corresponding to the first of quantization matrix, quantization parameter, video source bit depth, the transform size of converter unit or its any combination
Function.In addition, threshold value may correspond to fixed value or the second weighted value, wherein the second weighted value corresponds to quantization matrix, quantifies ginseng
Second linear function of number, the transform size of video source bit depth, converter unit or its any combination.
The quantization level of conversion coefficient also can be intercepted unconditionally to the bit depth of required bit depth rather than video source.It is required
Bit depth can be 8,16 or 32, and corresponding interception scope can be [- 128,127], [- 32768,32767] and [-
2147483648,2147483647]。
Fig. 1 is shown with reference to one embodiment of the invention for the interception quantization level of Video Decoder to avoid what is overflowed
Preferable flow chart.In step 110, the conversion coefficient of converter unit is quantified according to quantization matrix and quantization parameter, with
Produce the quantization level of conversion coefficient.In the step 120, based on quantization matrix, quantization parameter, inverse quantization step-length, video source position
Depth, the transform size of converter unit or its any combination, it is determined that interception condition, wherein interception condition includes empty interception condition.
Then, in step 130, according to interception condition interception quantization level, to produce the quantization level after intercepting process.
Flow chart in Fig. 1 is intended to indicate that the quantization level for video encoder is intercepted to avoid quantization level from overflowing
The example gone out.Those skilled in the art can by rearrange these steps, split one or more steps or combination one or
Multiple steps realize the present invention.
Foregoing description can make those skilled in the art implement the present invention with application-specific presented above and requirement.
Those skilled in the art can understand the various deformation of above-described embodiment, and basic principle defined here can be applied to other realities
Apply example.Therefore, the present invention is not limited in above-mentioned illustrated and described specific embodiment, but describes to meet and taken off with this place
The consistent most wide scope of the principle and notable feature of dew.In foregoing detailed description, various details can be provided to this
The comprehensive understanding of invention.However, those skilled in the art work as and are appreciated that how the present invention is implemented.
The above embodiment of the present invention can be embodied as hardware, software code or its combination.For example, embodiments of the invention can be
The circuit integrated in video compress chip, or the program code integrated in video compression software, for performing this place
The processing stated.One embodiment of the invention is alternatively the digital signal processor (Digital for performing processing described herein
Signal Process, DSP) on performed program code.The present invention can be also included by computer processor, Digital Signal Processing
It is multiple performed by device, microprocessor or field effect programmable gate array (Field Programmable Gate Array, FPGA)
Function.The machine that these processors can be configured to define presently embodied ad hoc approach by performing according to the present invention can
Software code or firmware code are read to perform particular task.Different programming languages can be used to mention not for software code or firmware code
With form or type develop.Software code also may conform to different target platforms.However, the different codes of software code
Form, type and language and configuration code are with the other modes according to execution task of the present invention, all without departing from the spirit of the present invention
And scope.
The present invention can be presented as other concrete forms in the case where not departing from spirit or key character.Above-mentioned implementation
Example is for illustration purposes only, and is not used to limit.Therefore protection scope of the present invention, which is worked as, regards appended claims institute defender
It is defined.Any equalization made met in the spirit and scope of the present invention sexually revises, and each falls within the scope of the present invention.
Claims (18)
1. a kind of quantization level intercept method, suitable for video encoder, the quantization level intercept method includes:
The conversion coefficient of converter unit is quantified according to quantization matrix and quantization parameter, to produce the amount of the conversion coefficient
Change level;
In the video encoder determine be applied to fixed range interception condition and dynamic range interception condition one or
The interception scope of combination;
The quantization level is intercepted according to the interception scope, to produce the quantization level after intercepting process;And
Wherein, under the conditions of fixed range interception, the quantization level is truncated to-m to m-1 fixed interception scope,
And m corresponds to 128,32768 or 2147483648.
2. quantization level intercept method according to claim 1, it is characterised in that when it is determined that the interception scope is applicable
During the dynamic range interception condition, the determination of the interception scope is based on the quantization matrix, the quantization parameter, inverse
Change step-length, video source bit depth, the transform size of the converter unit or above-mentioned any combination.
3. quantization level intercept method according to claim 2, it is characterised in that, will based on the video source bit depth
The quantization level interception is the first scope or the second scope.
4. quantization level intercept method according to claim 2, it is characterised in that the interception scope is by by first
Weighted value makes comparisons to determine with threshold value, wherein, first weighted value corresponds to the quantization matrix, the quantization ginseng
Several, described video source bit depth, the transform size of the converter unit or the first linear function of above-mentioned any combination.
5. quantization level intercept method according to claim 4, it is characterised in that the threshold value corresponds to fixed value or the
Two weighted values, wherein second weighted value correspond to the quantization matrix, the quantization parameter, the video source bit depth,
The transform size of the converter unit or the second linear function of above-mentioned any combination.
6. quantization level intercept method according to claim 1, it is characterised in that the institute of the fixed range interception condition
State the video source bit depth that interception scope is applied to be more than or equal to 10.
7. quantization level intercept method according to claim 1, it is characterised in that the quantization level after intercepting process
Represented using n positions.
8. quantization level intercept method according to claim 7, it is characterised in that n corresponds to 8,16 or 32.
9. a kind of quantization level capture device, suitable for video encoder, the quantization level capture device includes:
The conversion coefficient of converter unit is quantified according to quantization matrix and quantization parameter, to produce the amount of the conversion coefficient
Change the device of level;
In the video encoder determine be applied to fixed range interception condition and dynamic range interception condition one or
The device of the interception scope of combination;
The quantization level is intercepted according to the interception scope, to produce the device of the quantization level after intercepting process;And
Wherein, under the conditions of fixed range interception, the quantization level is truncated to-m to m-1 fixed interception scope,
And m corresponds to 128,32768 or 2147483648.
10. quantization level capture device according to claim 9, it is characterised in that when it is determined that the interception scope is suitable
When intercepting condition with the dynamic range, the determination of the interception scope is based on the quantization matrix, quantization parameter, anti-
Quantization step, video source bit depth, the transform size of the converter unit or above-mentioned any combination.
11. quantization level capture device according to claim 10, it is characterised in that further include:
Based on the video source bit depth, the device by quantization level interception for the first scope or the second scope.
12. quantization level capture device according to claim 10, it is characterised in that the interception scope is by by the
One weighted value is compared to what is determined with threshold value, wherein, first weighted value corresponds to the quantization matrix, the quantization
Parameter, the video source bit depth, the first linear function of the transform size of the converter unit or above-mentioned any combination.
13. quantization level capture device according to claim 12, it is characterised in that the threshold value correspond to fixed value or
Second weighted value, wherein second weighted value corresponds to the quantization matrix, the quantization parameter, the video source locating depth
Second linear function of degree, the transform size of the converter unit or above-mentioned any combination.
14. quantization level capture device according to claim 9, it is characterised in that the fixed range interception condition
The interception scope is applied to the video source bit depth more than or equal to 10.
15. quantization level capture device according to claim 9, it is characterised in that the quantization electricity after intercepting process
It is flat to be represented using n positions.
16. quantization level capture device according to claim 15, it is characterised in that n corresponds to 8,16 or 32.
17. a kind of computer-readable recording medium, stores computer program, described program realizes power when being executed by processor
Profit is required in 1-8 described in any one the step of quantization level intercept method.
18. a kind of quantization level capture device, suitable for video encoder, the quantization level capture device includes:
Processor, memory is coupled to, perform claim will in program stored in performing the memory for the processor
Ask in 1-8 described in any one the step of quantization level intercept method.
Applications Claiming Priority (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CNPCT/CN2011/084083 | 2011-12-15 | ||
PCT/CN2011/084083 WO2013086724A1 (en) | 2011-12-15 | 2011-12-15 | Method of clippling transformed coefficients before de-quantization |
CN201280054630.6A CN103959780B (en) | 2011-12-15 | 2012-12-14 | Quantization level capture device and method |
Related Parent Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201280054630.6A Division CN103959780B (en) | 2011-12-15 | 2012-12-14 | Quantization level capture device and method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107493475A true CN107493475A (en) | 2017-12-19 |
CN107493475B CN107493475B (en) | 2019-05-17 |
Family
ID=51334889
Family Applications (2)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710823352.0A Active CN107493475B (en) | 2011-12-15 | 2012-12-14 | Quantization level capture device and method |
CN201280054630.6A Active CN103959780B (en) | 2011-12-15 | 2012-12-14 | Quantization level capture device and method |
Family Applications After (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201280054630.6A Active CN103959780B (en) | 2011-12-15 | 2012-12-14 | Quantization level capture device and method |
Country Status (1)
Country | Link |
---|---|
CN (2) | CN107493475B (en) |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104602014B (en) * | 2015-01-22 | 2018-04-03 | 复旦大学 | A kind of quantization suitable for HEVC standard and inverse quantization hardware multiplexing algorithm and hardware configuration |
CN106028032B (en) * | 2016-05-24 | 2019-03-26 | 西安电子科技大学 | A kind of coefficient level adaptation quantization method |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP2154895A1 (en) * | 2008-08-15 | 2010-02-17 | Thomson Licensing | Method and apparatus for entropy encoding blocks of transform coefficients of a video signal using a zigzag scan sequence, and method and apparatus for a corresponding decoding |
US7778813B2 (en) * | 2003-08-15 | 2010-08-17 | Texas Instruments Incorporated | Video coding quantization |
CN102037729A (en) * | 2006-06-26 | 2011-04-27 | 高通股份有限公司 | Efficient fixed-point approximations of forward and inverse discrete cosine transforms |
CN102271258A (en) * | 2011-08-03 | 2011-12-07 | 中山大学深圳研究院 | Method and device for video coding aiming at low coding rate |
CN102577379A (en) * | 2009-10-05 | 2012-07-11 | 汤姆逊许可证公司 | Methods and apparatus for embedded quantization parameter adjustment in video encoding and decoding |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8606023B2 (en) * | 2006-06-26 | 2013-12-10 | Qualcomm Incorporated | Reduction of errors during computation of inverse discrete cosine transform |
-
2012
- 2012-12-14 CN CN201710823352.0A patent/CN107493475B/en active Active
- 2012-12-14 CN CN201280054630.6A patent/CN103959780B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7778813B2 (en) * | 2003-08-15 | 2010-08-17 | Texas Instruments Incorporated | Video coding quantization |
CN102037729A (en) * | 2006-06-26 | 2011-04-27 | 高通股份有限公司 | Efficient fixed-point approximations of forward and inverse discrete cosine transforms |
EP2154895A1 (en) * | 2008-08-15 | 2010-02-17 | Thomson Licensing | Method and apparatus for entropy encoding blocks of transform coefficients of a video signal using a zigzag scan sequence, and method and apparatus for a corresponding decoding |
CN102577379A (en) * | 2009-10-05 | 2012-07-11 | 汤姆逊许可证公司 | Methods and apparatus for embedded quantization parameter adjustment in video encoding and decoding |
CN102271258A (en) * | 2011-08-03 | 2011-12-07 | 中山大学深圳研究院 | Method and device for video coding aiming at low coding rate |
Also Published As
Publication number | Publication date |
---|---|
CN107493475B (en) | 2019-05-17 |
CN103959780A (en) | 2014-07-30 |
CN103959780B (en) | 2017-10-13 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Meyer et al. | Glicbawls-Grey Level Image Compression by Adaptive Weighted Least Squares. | |
Lee et al. | Energy-efficient image compression for resource-constrained platforms | |
US9749635B2 (en) | Method and apparatus for quantization level clipping | |
EP1980022B1 (en) | Variable length coding for sparse coefficients | |
US10630983B2 (en) | Transform coefficient dequantization method and apparatus, and decoding device | |
CN103947206A (en) | Region-based image compression | |
US7499596B2 (en) | Amplitude coding for clustered transform coefficients | |
US8179972B2 (en) | Image encoding device, image decoding device, and integrated circuit used therein | |
JP2007142823A (en) | Decoding apparatus, inverse quantization method, and program thereof | |
CN107493475B (en) | Quantization level capture device and method | |
Hayne et al. | Using neural networks for two dimensional scientific data compression | |
CN103975592B (en) | The method and device of inverse quantization conversion coefficient | |
JP2012147290A (en) | Image coding apparatus, image coding method, program, image decoding apparatus, image decoding method, and program | |
WO2014167609A1 (en) | Information processing device, control method, program, and recording medium | |
KR102250088B1 (en) | Method and Apparatus for decoding video stream | |
De Neve et al. | GPU-assisted decoding of video samples represented in the YCoCg-R color space | |
JP2008236090A (en) | Color space conversion apparatus and method, information processor, computer program, and information recording medium | |
CN101742298B (en) | Virtual decimal inverse discrete cosine transforming method for embedded real-time decoder | |
Nesamani et al. | Implemenation of simplified architecture of JPEG 2000 MQ coder | |
JPWO2021139572A5 (en) | ||
Borges et al. | Segmented spline hardware design for high dynamic range video pre-processor | |
CN100543769C (en) | Save the method and apparatus of memory source | |
Aritspoundy et al. | Machine Instruction Analysis For Dct Algorithm Using Dlx Architecture | |
RU2500032C1 (en) | Method of compressing graphics files | |
Tang et al. | Network Aware Compression Based Rate Control for Printing System |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |