CN108347603A - Coded dynamic image device and coded dynamic image method - Google Patents
Coded dynamic image device and coded dynamic image method Download PDFInfo
- Publication number
- CN108347603A CN108347603A CN201710060951.1A CN201710060951A CN108347603A CN 108347603 A CN108347603 A CN 108347603A CN 201710060951 A CN201710060951 A CN 201710060951A CN 108347603 A CN108347603 A CN 108347603A
- Authority
- CN
- China
- Prior art keywords
- summation
- data
- conversion
- circuit
- coordinate values
- 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
- H04N19/103—Selection of coding mode or of prediction mode
-
- 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
- H04N19/134—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or criterion affecting or controlling the adaptive coding
- H04N19/146—Data rate or code amount at the encoder output
- H04N19/147—Data rate or code amount at the encoder output according to rate distortion criteria
-
- 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
- H04N19/189—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the adaptation method, adaptation tool or adaptation type used for the adaptive coding
- H04N19/19—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the adaptation method, adaptation tool or adaptation type used for the adaptive coding using optimisation based on Lagrange multipliers
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
A kind of coded dynamic image device of present invention offer and coded dynamic image method, controller therein include to add up circuit, data volume estimating circuit and assessment circuit.Each pattern in a variety of intra predictions/motion compensation pattern respectively correspond tos residual error data after one group of conversion quantization.The totalling circuit calculates the coordinate values summation of the absolute value summation and these nonzero elements of nonzero element therein relative to a reference point for residual error data after each group of conversion quantization.The data volume estimating circuit is directed to each intra prediction/motion compensation pattern, according to the absolute value summation and the coordinate values summation of residual error data after its corresponding conversion quantization, generates a data volume estimated value.The assessment circuit selects an optimal mode according to multiple data volume estimated value from a variety of intra predictions/motion compensation pattern.
Description
Technical field
The present invention is related to image processing technique, and is especially selected from a variety of image processing patterns with according to data volume
The technology of one optimal mode is related.
Background technology
In recent years, as various electronic related technologies flourish, the multimedia systems such as family's theater become increasingly popular.More
In number multimedia system, most important hardware device just belongs to display device.In order to meet viewer for image true to nature
Demand, the one of the current development trend of display device is to continue the size and resolution of promotion figure frame (frame), thus make
The image data amount for obtaining each figure frame is significantly increased.How while possessing good image quality, through compress technique by image
It is the subject under discussion to merit attention to save storage space and transfer resource that data volume reduces as far as possible.
Fig. 1 presents the functional block diagram for the coded dynamic image system being widely used at present.Each figure frame would generally
Multiple image blocks are divided into, as the base unit of coding.The reference data of one block to be encoded is inputted into intra prediction
(intra-prediction)/motion compensation (motion compensation) circuit 101, foregoing circuit 101 is after operation
Export a reference block.Wherein intra prediction (intra-prediction)/motion compensation circuit also includes the work(of mobile estimating
Can, scope is not discussed in the present invention, this will not be repeated here.Then, residual error generation circuit 102 is responsible for finding out block to be encoded and ginseng
The difference of examination district block.Difference can transfer to conversion circuit 103A progress discrete between this is commonly referred to as the block of residual error (residual) data
Cosine converts (discrete cosine transform, DCT) and sample circuit 103B is transferred to be quantified
(quantization) program.Then, entropy coding circuit 104 is responsible for residual error data and its corresponding intermediary after quantifying to conversion
Data (metadata) impose entropy coding (entropy encoding), to generate a coding result.
Inverse quantization circuit 106A receives residual error data after conversion quantifies with inverse transform circuit 106B analog image decoding ends
Residual error data after the reduction that will produce afterwards.Residual error data is stored in slow after being added with reference block after add circuit 107 will restore
Device 108 is rushed, as the intra prediction used for intra prediction/motion compensation circuit 101/motion compensation reference data.
In practice, 101 adoptable intra prediction of intra prediction/motion compensation circuit/motion compensation pattern is up to tens of
Kind, and the coding result that each self- steering is different.It is existing generally to evaluate and most using Lagrangian method (Lagrange method)
Take into account intra prediction/motion compensation pattern of low data bulk and two kinds of demands of low distortion.In the coded dynamic image system that Fig. 1 is painted
In system 100, control circuit 110A control intra prediction/motion compensation circuits 101 attempt various intra prediction/motion compensations one by one
Pattern.Controller 110 also includes a data volume counting circuit 110B, an amount distortion counting circuit 110C and an assessment circuit
110D.Data volume counting circuit 110B is responsible for calculating the data volume R of a coding result.Amount distortion counting circuit 110C is then difference
The residual error data and residual error data after the reduction of inverse transform circuit 106B generations that residual error generation circuit 102 generates are received, and according to this
Great amount distortion D can be caused by calculating the conversion circuit 103A and sample circuit 103B programs carried out, be supplied to assessment circuit
110D is referred to.Then, assessment circuit 110D is each intra prediction/motion compensation mode computation point according to Lagrangian method
Number.Roughly speaking, the data volume R of coding result is lower, and the score is lower;Amount distortion D is lower, and the score is also lower.Therefore, it comments
The minimum intra prediction of score/motion compensation pattern can be selected as an optimal mode by estimating circuit 110D.
Before assessment circuit 110D selects optimal mode, each coding result that entropy coding circuit 104 generates is by elder generation
It is temporarily stored in temporary storage 109.Until assessment circuit 110D selects optimal mode, temporary storage 109 sends out best mould again
The corresponding coding result of formula (is denoted as forced coding result) in figure, as the output signal of coded dynamic image system 100.
This coding result for calculating each intra prediction/motion compensation pattern one by one in coded dynamic image system 100
Although the way of data volume and amount distortion can accurately select optimal mode, process is relatively time consuming, and needs to employ a large amount of
Calculation resources.
Invention content
To solve the above problems, the present invention proposes a kind of new coded dynamic image device and coded dynamic image method.
A specific embodiment according to the present invention is a kind of coded dynamic image device, wherein including one intra prediction/shifting
Dynamic compensation circuit, a residual error generation circuit, a conversion circuit, a sample circuit and a controller.Intra prediction/the motion compensation
It is that an image block to be encoded finds out a variety of reference blocks that a variety of intra predictions/motion compensation pattern, which is respectively adopted, in circuit.This is residual
Poor generation circuit is according to the image block to be encoded multigroup residual error data corresponding with a variety of reference block generations.The conversion
Circuit carries out a discrete cosine transform program for each group of residual error data, to generate matrix after a conversion.The sample circuit needle
One quantization program is carried out to matrix after each conversion, to generate residual error data after one group of conversion quantization.The controller includes one
Add up circuit, a data volume estimating circuit and an assessment circuit.Residual error data after the totalling circuit quantifies for each group of conversion,
Calculate the coordinate values summation of the absolute value summation and these nonzero elements of nonzero element therein relative to a reference point.The number
It is directed to each intra prediction/motion compensation pattern according to amount estimating circuit, according to residual error data after its corresponding conversion quantization
The absolute value summation and the coordinate values summation, generate a data volume estimated value.The assessment circuit is estimated according to multiple data volume
Value, an optimal mode is selected from a variety of intra predictions/motion compensation pattern.
Another specific embodiment according to the present invention is a kind of coded dynamic image method.First, a variety of frames are respectively adopted
A variety of reference blocks of interior prediction/motion compensation pattern, an image block to be encoded are found.Then, according to the shadow to be encoded
As block and a variety of reference blocks, corresponding multigroup residual error data is generated.Then, each group of residual error data be applied in
One discrete cosine transform program and a quantization program, to generate residual error data after one group of conversion quantization.For each group of amount of translation
Residual error data after change, the coordinate values of the absolute value summation and these nonzero elements of nonzero element therein relative to a reference point
Summation is computed.For each intra prediction/motion compensation pattern, according to residual error data after its corresponding conversion quantization
The absolute value summation and the coordinate values summation, a data volume estimated value is generated.According to multiple data volume estimated value, certainly should
In a variety of intra predictions/motion compensation pattern, an optimal mode is selected.
It can be further understood by following detailed description and accompanying drawings about the advantages and spirit of the present invention.
Description of the drawings
Fig. 1 presents the functional block diagram of a now widely used coded dynamic image system.
Fig. 2 is the functional block diagram according to the coded dynamic image system in one embodiment of the invention.
Fig. 3 presents the data matrix example that a size is 4*4.
Fig. 4 is the functional block diagram according to the coded dynamic image system in another embodiment of the present invention.
Fig. 5 is the functional block diagram according to the coded dynamic image system in another embodiment of the present invention.
Fig. 6 is the flow chart according to the image treatment method in one embodiment of the invention.
Fig. 7 is the flow chart according to the image treatment method in another embodiment of the present invention.
Symbol description
100、200:Coded dynamic image system
101、201:Intra prediction/motion compensation circuit
102、202:Residual error generation circuit
103A、203A:Conversion circuit
103B、203B:Sample circuit
104、204:Entropy coding circuit
106A、206A:Inverse quantization circuit
106B、206B:Inverse transform circuit
107、207:Add circuit
108、208:Buffer
109、209:Temporary storage
110、210:Controller
110A、210A:Control circuit
110B:Data volume counting circuit
210B:Data volume estimating circuit
110C、210C:Amount distortion counting circuit
110D、210D:Assessment circuit
210E:Add up circuit
210F:Amount distortion estimating circuit
S61~S67:Process step
S71~S74:Process step
It should be noted that attached drawing of the invention is not thin portion circuit diagram, and connecting line therein is only indicating signal stream.
A variety of interactive relationship between functional element and/or program are not necessarily intended to reach through the direct electrical connection beginning.In addition, a
The function of other element is not necessarily intended to distribute such as the mode being painted in figure, and distributed block is not necessarily intended to distributed electronics
Element is realized.
Specific implementation mode
A specific embodiment according to the present invention is a coded dynamic image system.Please refer to function block shown in Fig. 2
Figure.Coded dynamic image system 200 includes intra prediction/motion compensation circuit 201, residual error generation circuit 202, conversion circuit
203A, sample circuit 203B, entropy coding circuit 204, inverse quantization circuit 206A, inverse transform circuit 206B, add circuit 207, delay
Rush device 208, temporary storage 209 and controller 210.Also, include a control circuit 210A, a number in controller 210
Circuit 210E is added up according to amount estimating circuit 210B, an amount distortion counting circuit 210C, an assessment circuit 210D and one.
In this embodiment, intra prediction/motion compensation circuit 201, residual error generation circuit 202, conversion circuit 203A, amount
It is prior art to change circuit 203B, inverse quantization circuit 206A, inverse transform circuit 206B, add circuit 207 and buffer 208,
Its function mode can refer to the explanation of each corresponding circuit in Fig. 1.Coded dynamic image system 200 and coded dynamic image system
One of 100 essential difference is, controller 210 is not the accurate data amount according to coding result to select optimal mode, and
It is the data volume estimated value that residual error data generates after being changed to be quantified according to conversion with reference to one, details are as follows.
For each intra prediction/motion compensation pattern, sample circuit 203B can generate residual after one group of conversion quantifies
Difference data.The example of 4*4 data matrixes shown in Fig. 3 represents the 2-D data of residual error data after conversion quantization, in matrix
Each data value respectively has an an abscissa x and ordinate y.The exhausted of nonzero element therein can be calculated by adding up circuit 210E
To being worth summation SUMABSAnd coordinate values summation SUM of these nonzero elements relative to a reference pointCRD.If with the data square of Fig. 3
For battle array, the circuit 210E calculated absolute value summation SUM of institute are added upABSCan be:
54+25+16+4+32+11+10+6+2+8+1=169.
In an embodiment, it is by the longitudinal coordinate values and lateral seat of coordinate where all nonzero elements to add up circuit 210E
Scale value is added, as coordinate values summation SUMCRD.If by taking the data matrix of Fig. 3 as an example, adds up circuit 210E and calculate in this way
The coordinate values summation SUM gone outCRDCan be:
(0+0)+(1+0)+(2+0)+(3+0)+(0+1)+(1+1)+(0+2)+(1+2)+(2+2)+(0+3)+(3+3)=
27。
For each intra prediction/motion compensation pattern, data volume estimating circuit 210B is according to its corresponding data square
The absolute value summation SUM of battle arrayABSWith coordinate values summation SUMCRD, generate a data volume estimated valueIn an embodiment, data volume
Estimating circuit 210B assigns absolute value summation SUM respectivelyABSWith coordinate values summation SUMCRDOne certain weights, then according to the weighting
Absolute value summation generates data volume estimated value with coordinate values summation after the weighting afterwardsFor example, data volume estimating circuit
Following default calculation formula can be used in 210B:
Symbol a, b, c therein represent the fixed parameter of numerical value (parameter a, b is above-mentioned certain weights), using linear
The methods of recurrence generates.More specifically, circuit designers can coordinate data volume counting circuit 110B shown in FIG. 1 and more in advance
Group sample data, finds out absolute value summation SUMABS, coordinate values summation SUMCRDWith the correspondence between accurate actual amount of data,
And it finds out calculation formula appropriate (that is, finding out parameter a, b, c appropriate) using the methods of linear regression and describes this correspondence.
In other words, aforementioned imparting absolute value summation SUMABSWith coordinate values summation SUMCRDCertain weights can be using linear regression etc.
What method was determined.It should be noted that the calculation formula for describing the correspondence also may include absolute value summation SUMABSAnd/or seat
Scale value summation SUMCRDMultiple item, that is, not with only include first order formula one be limited.
In another embodiment, for the 2-D data of residual error data after each group of expression conversion quantization, circuit is added up
210E calculates the lateral coordinate values summation SUM of nonzero element thereinCRD_XWith longitudinal coordinate values summation SUMCRD_Y.Data volume is estimated
Meter circuit 210B assign absolute value summation SUM respectivelyABS, lateral coordinate values summation SUMCRD_XWith longitudinal coordinate values summation SUMCRD_Y
One certain weights, then according to lateral seat after longitudinal coordinate values summation after absolute value summation, weighting after the weighting and the weighting
Scale value summation generates data volume estimated valueFor example, following default calculation formula can be used in data volume estimating circuit 210B:
Parameter a, b1, b2, c therein can utilize the methods of aforementioned linear recurrence to generate.
As shown in Fig. 2, the data volume estimated value that data volume estimating circuit 210B is generatedIt is provided to assessment circuit 210D.
For same image block to be decoded, it is various that control circuit 210A can control intra prediction/motion compensation circuit 201 to carry out one by one
Intra prediction/motion compensation pattern, data volume estimating circuit 210B is then according to the corresponding generation different data amount of various patterns
Estimated value
Similar, amount distortion calculates module 207 and corresponds to the different mistake of generation according to various intra predictions/motion compensation pattern
True amount D.More specifically, inverse quantization circuit 206A and inverse transform circuit 206B can rebuild residual error after the conversion quantization of each pattern
Data, to generate residual error data after corresponding reduction.Amount distortion calculates 207 system of module to according to residual error data after this reduction
Difference value between the residual error data generated with residual error generation circuit 202 determines amount distortion D.Assessment circuit 210D can be according to all
The data volume estimated value of intra prediction/motion compensation patternAnd amount distortion D, utilize Lagrangian method or similar assessment side
Formula, it is a kind of most for current image block selection from a variety of image processing patterns of intra prediction/motion compensation circuit 201
Good pattern (that is, can most take into account the pattern of low data bulk and low distortion amount).
In an embodiment, before assessment circuit 210D selects optimal mode, residual error data can be temporary after conversion quantization
It is stored in temporary storage 209.After assessment circuit 210D selects optimal mode, temporary storage 209 can be by correspondence
In optimal mode conversion quantization after residual error data be supplied to entropy coding circuit 204, with to this conversion quantization after residual error data and
Its corresponding broker data imposes entropy coding, to generate a coding result.
In another embodiment, only stored in temporary storage 209 optimal mode being currently known conversion quantization after it is residual
Difference data.Whenever assessment circuit 210D find another intra prediction/motion compensation pattern be it is better, will be with new conversion
Residual error data after conversion quantization originally temporary in residual error data substitution temporary storage 209 after quantization.Until in all frames
Prediction/motion compensation pattern has been attempted, and stored by temporary storage 209 is the amount of translation corresponding to optimal mode
Residual error data after change.This way is advantageous in that the hardware space that can save temporary storage 209.
In another embodiment, before assessment circuit 210D selects optimal mode, temporary storage 209 does not store any
Residual error data after conversion quantization, but only record (such as in an indexed fashion) current optimal mode is any pattern.Directly
It is all attempted to all intra predictions/motion compensation pattern, control circuit 210A can just control intra prediction/motion compensation
Circuit 201, residual error generation circuit 202, conversion circuit 203A, sample circuit 203B are regenerated out corresponding to this optimal mode
Residual error data after conversion quantization, is encoded for entropy coding circuit 204.
It can be seen that by described above, in coded dynamic image system 200, data volume estimating circuit 210B is according to amount of translation
Residual error data generates data volume estimated value after changeCompared to the prior art that Fig. 1 is presented, entropy coding circuit 204 only needs needle
Residual error data after conversion quantization corresponding to the optimal mode that is selected and broker data are encoded, without to each
Residual error data and broker data are encoded after conversion quantization corresponding to a pattern.Coded dynamic image system 200 can be used
The shorter time generates data volume estimated value with less calculation resourcesOptimal mode is selected as assessment circuit 210D
Reference data.
Please refer to Fig. 4.In another embodiment, data volume estimating circuit 210B is also to various intra prediction/motion compensations
The broker data of pattern carries out data volume assessment.By a certain parameter in broker data by taking multiple bits indicate as an example, this multiple position
If in member some bit be classified as bypass (bypass) data, that is, refer to this part bit can not by Probability into
Row is correctly predicted, so to this part, bit does not impose entropy coding program to entropy coding circuit 204.The broker data of each pattern
In each content of parameter be bypass data or non-bypass (non-bypass) data are it is known that therefore data volume estimating circuit 210B can
A bypass data quantity B of corresponding broker data can be calculated according to these informationPWith a non-bypass data bulk BNP, and according to
This generates a broker data amount estimated value
For example, following default calculation formula can be used in data volume estimating circuit 210B:
Wherein symbol α represents a weighting parameters, and numerical value can rule of thumb be determined by circuit designers, such as can be set
To be equal to 1 or being slightly less than 1.In practice, if the content of various broker datas is fixed, data volume estimating circuit 210B is available
The mode tabled look-up generates bypass data quantity BPWith non-bypass data bulk BNP。
In addition to the absolute value summation SUM of each patternABSWith coordinate values summation SUMCRD, the data volume estimating circuit in Fig. 4
210B is generating its data volume estimated value for each patternWhen, also by its broker data amount estimated valueIt is included in and examines
Amount.For example, formula one can be revised as by the data volume estimating circuit 210B in figure four:
A kind of change type of coded dynamic image system 200 is presented in Fig. 5.In this embodiment, amount distortion calculates module
210C is replaced by an amount distortion estimating circuit 210F.The input signal of amount distortion estimating circuit 210F is that conversion circuit 203A is generated
Conversion after the inverse quantization result that generates of matrix and inverse quantization circuit 206A.For each intra prediction/motion compensation mould
Formula, amount distortion estimating circuit 210F calculates the difference of the inverse quantization result and matrix after the conversion, as an amount distortion estimated valueIt is referred to when selecting optimal mode for assessment circuit 210D.Compared to the prior art that Fig. 1 is presented, done used by figure five
Method is generating amount distortion estimated valueDuring need not carry out inverse transform program, thus be reduced output amount distortion estimation
ValueThe required time, and further save calculation resources.
In addition, amount distortion estimating circuit 210F is designed to only to calculate matrix and the one of its inverse quantization result after the conversion
High bit member difference ignores a relatively low bit difference.For example, it is assumed that each after conversion in matrix and inverse quantization result
First prime system respectively indicates that amount distortion estimating circuit 210F can only calculate two relatively with the binary data of length sixteen bit member
The difference for answering the first eight high bit member of element ignores the difference of rear eight relatively low bits.This way can also reach into one
Step saves the effect of operation time and calculation resources.
In practice, aforementioned totalling circuit 210E, data volume estimating circuit 210B and amount distortion assessment circuit 210F can be respective
It is implemented as but is not limited to fixed and/or programmed numerical digit logic circuit, including programmable logic gate array, specific
Application integrated circuit, microcontroller, microprocessor, digital signal processor, with other necessary circuitries.
Another specific embodiment according to the present invention is a kind of coded dynamic image method, and flow chart is illustrated in Fig. 6.It is first
First, step S61 is that an image block to be encoded finds out a variety of reference areas for a variety of intra predictions/motion compensation pattern is respectively adopted
Block.Secondly, the image block to be encoded multigroup residual error number corresponding with a variety of reference block generations according to step S62
According to.Then, step S63 is to carry out a discrete cosine transform program for each group of residual error data, to generate matrix after a conversion.
Step S64 is then to carry out a quantization program for matrix after each conversion, to generate residual error data after one group of conversion quantization.It connects
, step S65 is residual error data after quantifying for each group of conversion, calculate nonzero element therein absolute value summation and this
A little coordinate values summations of the nonzero element relative to a reference point.Step S66 is to be directed to each intra prediction/motion compensation mould
Formula generates a data volume and estimates according to the absolute value summation and the coordinate values summation of residual error data after its corresponding conversion quantization
Evaluation.Step S67 is then to select one from a variety of intra predictions/motion compensation pattern according to multiple data volume estimated value
Optimal mode.
Another specific embodiment according to the present invention is a kind of image treatment method, and flow chart is illustrated in Fig. 7.First,
Step S71 is to carry out a discrete cosine transform program for an image data, to generate matrix after a conversion.Secondly, step S72
A quantization program is carried out to be directed to matrix after the conversion, to generate data after a conversion quantization.Then, step S73 is for this
Data carry out an inverse quantization program after conversion quantization, to generate an inverse quantization result.Then, according to step S74 after the conversion
The difference of matrix and the inverse quantization result determines an amount distortion estimated value.
By the above detailed description of preferred embodiments, it would be desirable to the feature and spirit of the present invention are more clearly described, and
Not scope of the invention is limited with above-mentioned revealed preferred embodiment.On the contrary, the purpose is to wish
Various changes can be covered and have being arranged in the scope of the scope of the claims to be applied of the invention of equality.
Claims (14)
1. a kind of coded dynamic image device, including:
One intra prediction/motion compensation circuit, it is an image area to be encoded that a variety of intra predictions/motion compensation pattern, which is respectively adopted,
Block finds out a variety of reference blocks;
One residual error generation circuit, according to the image block to be encoded multigroup residual error number corresponding with a variety of reference block generations
According to;
One conversion circuit carries out a discrete cosine transform program for each group of residual error data, to generate matrix after a conversion;
One sample circuit carries out a quantization program for matrix after each conversion, to generate residual error number after one group of conversion quantization
According to;And
One controller, including:
One adds up circuit, residual error data after quantifying for each group of conversion, calculate the absolute value summation of nonzero element therein with
And coordinate values summation of these nonzero elements relative to a reference point;
One data volume estimating circuit, for each intra prediction/motion compensation pattern, according to residual after its corresponding conversion quantization
The absolute value summation of difference data and the coordinate values summation, generate a data volume estimated value;And
One assessment circuit is selected one most from a variety of intra predictions/motion compensation pattern according to multiple data volume estimated value
Good pattern.
2. coded dynamic image device as described in claim 1, which is characterized in that the data volume estimating circuit assigns respectively should
Absolute value summation and one certain weights of coordinate values summation, it is total with coordinate values after absolute value summation after one weighting of generation and a weighting
With;The data volume estimating circuit generates data volume estimation according to coordinate values summation after absolute value summation after the weighting and the weighting
Value.
3. coded dynamic image device as described in claim 1, which is characterized in that for residual error number after each group of conversion quantization
According to a longitudinal coordinate values summation and the lateral coordinate values summation of the totalling circuit counting nonzero element therein;The data volume
Estimating circuit assigns the absolute value summation, the longitudinal direction coordinate values summation and one certain weights of transverse direction coordinate values summation respectively, with
Lateral coordinate values summation after longitudinal coordinate values summation is weighted with one after absolute value summation, a weighting after one weighting of generation;The data
It is total according to lateral coordinate values after longitudinal coordinate values summation after absolute value summation, weighting after the weighting and the weighting to measure estimating circuit
With the generation data volume estimated value.
4. coded dynamic image device as described in claim 1, which is characterized in that the data volume estimating circuit is further:
For each intra prediction/motion compensation pattern, a bypass data quantity and one for corresponding broker data is calculated
Non-bypass data bulk, and a broker data amount estimated value is generated accordingly;And
When generating its data volume estimated value for each intra prediction/motion compensation pattern, also its broker data amount is estimated
Value, which is included in, to be considered.
5. coded dynamic image device as described in claim 1, which is characterized in that further include:
One inverse quantization circuit, residual error data carries out an inverse quantization program after quantifying for each group of conversion, to generate an inverse quantization
As a result;
And the controller further includes:
One amount distortion estimating circuit calculates the inverse quantization result and the conversion for each intra prediction/motion compensation pattern
The difference of matrix afterwards, as an amount distortion estimated value;
Wherein the assessment circuit also estimates the amount distortion of each intra prediction/motion compensation pattern when selecting the optimal mode
Evaluation, which is included in, to be considered.
6. coded dynamic image device as claimed in claim 5, which is characterized in that the residual error data after quantifying for one group of conversion
When generating its amount distortion estimated value, which calculates a high bit of matrix and its inverse quantization result after the conversion
First difference ignores a relatively low bit difference.
7. a kind of coded dynamic image method, including:
(a) it is that an image block to be encoded finds out a variety of reference blocks that a variety of intra predictions/motion compensation pattern, which is respectively adopted,;
(b) according to the image block to be encoded multigroup residual error data corresponding with a variety of reference block generations;
(c) it is directed to each group of residual error data and carries out a discrete cosine transform program, to generate matrix after a conversion;
(d) matrix carries out a quantization program after being directed to each conversion, to generate residual error data after one group of conversion quantization;
(e) it is directed to residual error data after each group of conversion quantization, the absolute value summation of nonzero element therein is calculated and these is non-
Coordinate values summation of the neutral element relative to a reference point;
(f) be directed to each intra prediction/motion compensation pattern, according to its it is corresponding conversion quantization after residual error data this absolutely
It is worth summation and the coordinate values summation, generates a data volume estimated value;And
(g) according to multiple data volume estimated value, an optimal mode is selected from a variety of intra predictions/motion compensation pattern.
8. coded dynamic image method as claimed in claim 7, which is characterized in that step (f) includes to be directed to each group of conversion
Residual error data after quantization:
The absolute value summation and one certain weights of coordinate values summation are assigned respectively, to generate absolute value summation and one after a weighting
Coordinate values summation after weighting;And
The data volume estimated value is generated according to coordinate values summation after absolute value summation after the weighting and the weighting.
9. coded dynamic image method as claimed in claim 7, it is characterised in that:
Residual error data after quantifying for each group of conversion, step (e) include to calculate a longitudinal coordinate values of nonzero element therein
Summation and a lateral coordinate values summation;
Step (f) is specific comprising the absolute value summation, the longitudinal direction coordinate values summation and the transverse direction coordinate values summation one is assigned respectively
Weight, lateral coordinate values summation after being weighted with one with longitudinal coordinate values summation after absolute value summation, a weighting after one weighting of generation;
And
Step (f) includes according to lateral seat after longitudinal coordinate values summation after absolute value summation, weighting after the weighting and the weighting
Scale value summation generates the data volume estimated value.
10. coded dynamic image method as claimed in claim 7, which is characterized in that further include:
For each intra prediction/motion compensation pattern, a bypass data quantity and one for corresponding broker data is calculated
Non-bypass data bulk, and a broker data amount estimated value is generated accordingly;
Wherein step (f) for each intra prediction/motion compensation pattern when generating its data volume estimated value, also by its intermediary
Data volume estimated value, which is included in, to be considered.
11. coded dynamic image method as claimed in claim 7, which is characterized in that further include:
Residual error data carries out an inverse quantization program after quantifying for each group of conversion, to generate an inverse quantization result;And
For each intra prediction/motion compensation pattern, the difference of the inverse quantization result and matrix after the conversion is calculated, as
One amount distortion estimated value;
Wherein step (g) is when selecting the optimal mode, also by the amount distortion estimated value of each intra prediction/motion compensation pattern
It is included in and considers.
12. coded dynamic image method as claimed in claim 11, which is characterized in that for residual error data after one group of conversion quantization
Generate its amount distortion estimated value include a high bit member difference of matrix and its inverse quantization result after calculating the conversion, ignore one compared with
Low level member difference.
13. a kind of coded dynamic image method, including:
(a) it is directed to an image data and carries out a discrete cosine transform program, to generate matrix after a conversion;
(b) matrix carries out a quantization program after being directed to the conversion, to generate data after a conversion quantization;
(c) data carry out an inverse quantization program after being directed to conversion quantization, to generate an inverse quantization result;And
(d) according to the difference of matrix and the inverse quantization result after the conversion, an amount distortion estimated value is determined.
14. coded dynamic image method as claimed in claim 13, which is characterized in that step (d) includes after calculating the conversion
One high bit member difference of matrix and the inverse quantization result ignores a relatively low bit difference.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710060951.1A CN108347603B (en) | 2017-01-25 | 2017-01-25 | Moving image encoding device and moving image encoding method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710060951.1A CN108347603B (en) | 2017-01-25 | 2017-01-25 | Moving image encoding device and moving image encoding method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108347603A true CN108347603A (en) | 2018-07-31 |
CN108347603B CN108347603B (en) | 2020-08-04 |
Family
ID=62963165
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710060951.1A Expired - Fee Related CN108347603B (en) | 2017-01-25 | 2017-01-25 | Moving image encoding device and moving image encoding method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108347603B (en) |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101014128A (en) * | 2007-02-02 | 2007-08-08 | 清华大学 | Method for quick estimating rate and distortion in H.264/AVC video coding |
CN101409845A (en) * | 2008-10-31 | 2009-04-15 | 北京大学软件与微电子学院 | Method and apparatus for estimating video distortion in AVS video encode |
CN101466040A (en) * | 2009-01-09 | 2009-06-24 | 北京大学 | Code rate estimation method for video encoding mode decision |
US20120237132A1 (en) * | 2011-03-15 | 2012-09-20 | Fujitsu Limited | Image-encoding method, image-encoding device, and computer-readable recording medium storing image-encoding program |
CN106034235A (en) * | 2015-03-11 | 2016-10-19 | 杭州海康威视数字技术股份有限公司 | Method for calculating coding distortion degree and coding mode control and system thereof |
-
2017
- 2017-01-25 CN CN201710060951.1A patent/CN108347603B/en not_active Expired - Fee Related
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101014128A (en) * | 2007-02-02 | 2007-08-08 | 清华大学 | Method for quick estimating rate and distortion in H.264/AVC video coding |
CN101409845A (en) * | 2008-10-31 | 2009-04-15 | 北京大学软件与微电子学院 | Method and apparatus for estimating video distortion in AVS video encode |
CN101466040A (en) * | 2009-01-09 | 2009-06-24 | 北京大学 | Code rate estimation method for video encoding mode decision |
US20120237132A1 (en) * | 2011-03-15 | 2012-09-20 | Fujitsu Limited | Image-encoding method, image-encoding device, and computer-readable recording medium storing image-encoding program |
CN106034235A (en) * | 2015-03-11 | 2016-10-19 | 杭州海康威视数字技术股份有限公司 | Method for calculating coding distortion degree and coding mode control and system thereof |
Also Published As
Publication number | Publication date |
---|---|
CN108347603B (en) | 2020-08-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Tao et al. | Significantly improving lossy compression for scientific data sets based on multidimensional prediction and error-controlled quantization | |
Nystad et al. | Adaptive scalable texture compression | |
US9171383B2 (en) | Apparatus and method of scalable encoding of 3D mesh, and apparatus and method of scalable decoding of 3D mesh | |
US10599935B2 (en) | Processing artificial neural network weights | |
He et al. | Low-power VLSI design for motion estimation using adaptive pixel truncation | |
CN106062816B (en) | Method and apparatus for being coded and decoded to HDR image | |
CN1230003C (en) | Distortion quantizer model for video encoding | |
JP2021520162A (en) | Video compression processing methods and devices, electronic devices and storage media | |
WO2021135715A1 (en) | Image compression method and apparatus | |
CN105103192B (en) | Method and apparatus for vertex error correction | |
CN111050176B (en) | Video encoding method, video decoding method, video encoding apparatus, video decoding apparatus, computer device, and storage medium | |
CN108322749A (en) | The coefficient optimization method of RDOQ, the accelerating method and device of RDOQ | |
MX2007008444A (en) | Method and system for coding/decoding fo a video bit stream for fine granularity scalability. | |
US7283460B1 (en) | Bus switch encoding for reducing crosstalk effects in buses | |
CN108347603A (en) | Coded dynamic image device and coded dynamic image method | |
US10559093B2 (en) | Selecting encoding options | |
TW202044129A (en) | Computing system and compressing method thereof for neural network parameters | |
TWI635742B (en) | Dynamic image encoding apparatus and dynamic image encoding method | |
WO2013011355A1 (en) | Method and apparatus for encoding an image | |
CN1216497C (en) | Video codeconversion method based on motion vector systemsis | |
CN101268623B (en) | Method and device for creating shape variable blocks | |
CN109547781B (en) | Compression method and device based on image prediction | |
CN116325741A (en) | Point cloud encoding method, point cloud decoding method and related devices | |
CN100429946C (en) | Method and apparatus for processing digital motion picture | |
CN103794219A (en) | Vector quantization codebook generating method based on M codon splitting |
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 | ||
TA01 | Transfer of patent application right |
Effective date of registration: 20200413 Address after: No.1, Duhang 1st Road, Hsinchu City, Hsinchu Science Park, Taiwan, China Applicant after: MEDIATEK Inc. Address before: 1/2, 4th floor, 26 Taiyuan Street, Zhubei City, Hsinchu County, Taiwan, China Applicant before: MSTAR SEMICONDUCTOR Inc. |
|
TA01 | Transfer of patent application right | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20200804 Termination date: 20220125 |
|
CF01 | Termination of patent right due to non-payment of annual fee |