CN103210649B - The method and apparatus using high-precision filter coding/decoding video - Google Patents
The method and apparatus using high-precision filter coding/decoding video Download PDFInfo
- Publication number
- CN103210649B CN103210649B CN201180054874.XA CN201180054874A CN103210649B CN 103210649 B CN103210649 B CN 103210649B CN 201180054874 A CN201180054874 A CN 201180054874A CN 103210649 B CN103210649 B CN 103210649B
- Authority
- CN
- China
- Prior art keywords
- block
- sub
- value
- video
- sampling
- 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
- 238000005070 sampling Methods 0.000 claims abstract description 163
- 238000000034 method Methods 0.000 claims description 87
- 238000006243 chemical reaction Methods 0.000 claims description 23
- 230000001131 transforming Effects 0.000 claims description 21
- 238000011002 quantification Methods 0.000 claims description 12
- 241001442055 Vipera berus Species 0.000 claims description 9
- 238000000605 extraction Methods 0.000 claims description 6
- 238000010586 diagram Methods 0.000 description 6
- 238000004891 communication Methods 0.000 description 5
- 238000007906 compression Methods 0.000 description 5
- 239000000284 extract Substances 0.000 description 4
- 238000001914 filtration Methods 0.000 description 4
- 230000004044 response Effects 0.000 description 3
- 239000000969 carrier Substances 0.000 description 2
- 238000004590 computer program Methods 0.000 description 2
- 230000000875 corresponding Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 239000000203 mixture Substances 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 1
- 238000002592 echocardiography Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 230000003287 optical Effects 0.000 description 1
- 230000003252 repetitive Effects 0.000 description 1
Abstract
An embodiment of the invention relates to a kind of method and apparatus for using high-precision filter coding/decoding video.An embodiment of the invention provides a kind of method and apparatus for coding/decoding video, and in coding/decoding video method, by utilizing FIR filter and linear interpolation to predict that block encodes and decode the value of the sub sampling element of the reference block by the motion vector reference of current block from the value generation of interpolation, with interpolation current block more accurately, thus increase code efficiency by reducing the difference between actual block and prediction block, and therefore improve relative efficiency.
Description
Technical field
The disclosure relates to a kind of method by using high-precision filter coding/decoding video in some embodiments
And equipment.More specifically, it relates to wherein utilize high-precision filter generate prediction signal and use this prediction signal
Generate video coding/decoding method and the equipment of high accuracy video signal.
Background technology
Description in this part provide only the background information relevant to the disclosure, and can not constitute prior art.
Motion characteristics planning (MPEG) and Video Coding Experts group (VCEG) super generating are superior to existing MPEG-
The new standard of the video compression technology of 4 part 2s and H.263 standard.New standard be referred to as the most H.264/AVC (advanced video coding) and
And advise H.264 issuing simultaneously as MPEG-4 the 10th part and ITU-T.H.264/AVC (below, it is simply referred as H.264)
By using various coded methods to substantially improving picture quality and performance.It addition, about Video coding associating group (MPEG
With VCEG combine group) discussing for the new standard than high definition (HD) higher picture quality.
As existing moving picture coding method, it is used for intra-frame predictive encoding method and inter prediction encoding side
Method.Intra-frame predictive encoding method predicts block by using the predictive value of the block of the intraframe coding currently encoded.Interframe is pre-
Survey coded method and predict the block of present frame by estimating to move according to the frame reconstructed before.
In the intra-frame prediction method of luminance signal, according to being coded of size and the Forecasting Methodology of block,
Employ 16 × 16 predictions and 8 × 8 predictions in frame in intra-frame 4 * 4 prediction, frame.
Fig. 1 is the figure illustrating typical nine kind of 4 × 4 intra prediction mode.
With reference to Fig. 1, intra-frame 4 * 4 prediction includes nine kinds of predictive modes: vertical mode, horizontal pattern, direct current (DC) pattern,
Point to the diagonal model of lower-left, point to the diagonal model of bottom right, vertical pattern to the right, level pattern on the lower side, the most to the left
Pattern and level pattern on the upper side.
Fig. 2 is the figure illustrating typical four kind of 16 × 16 intra prediction mode.
With reference to Fig. 2, in frame, 16 × 16 infra-frame predictions include four kinds of predictive modes: vertical mode, horizontal pattern, DC pattern and
Plane mode.With 16 × 16 predictions in frame similarly, in frame, 8 × 8 predictions also include four kinds of predictive modes.
In the inter-frame prediction method (inter prediction encoding) for the video with 4:2:0 video format, use
Motion compensation.Specifically, frame of video is divided, and by estimating that current block is predicted in motion from the frame encoded before.As
Fruit reduces the block size of motion compensation so that using, then can predict current block with higher accuracy.But, coding is used
Requirement in the motion vector information of each piece causes the increase of encoding amount.It addition, when performing motion compensation, by not only closing
Note has the motion vector in the integer samples of integer pixel and also concern has 1/4 sampling relevant to luminance component and differentiates
The sub sampling of rate and 1/8 sampling resolution relevant to chromatic component obtains motion vector more accurately.But, owing to son is adopted
The brightness of sample position and chroma samples are not present in reference base picture, it is therefore desirable to by adjacent whole of interpolation in reference base picture
Number sampling generates these values.
Summary of the invention
Technical problem
Embodiment of the present disclosure relates to by using during view interpolation higher more high-precision than linear interpolation accuracy
Degree wave filter improves video compression efficiency and improves subjective picture quality by reconstructing video efficiently.
The technological means of solution problem
Embodiments of the present invention provide a kind of video encoding/decoding apparatus, comprising: video encoder, by basis
The sub sampling component value of the reference block of the motion vector institute reference of current block utilizes FIR filter or/and linear interpolation is obtained
Interpolation generate prediction block, by from the chromatic component of current block deduct prediction block generate residual block, by conversion and amount
Change residual block and generate sampling frequency transform block, and sampling frequency transform block is encoded in bit stream；And video decoding
Device, for from bit stream generating quantification frequency transformed block, bringing weight by sampling frequency transform block carries out re-quantization and inversion
Structure residual block, the sub sampling component value of the reference block of the motion vector institute reference according to current block is utilized FIR filter or/and
The interpolation that linear interpolation is obtained is to generate prediction block, and the prediction Kuai Xiang Calais by the residual block that will reconstruct and generation is heavy
The chromatic component of structure current block.
Another embodiment of the disclosure provides a kind of video encoder, comprising: predicting unit, for by basis
The sub sampling component value of the reference block of the motion vector institute reference of current block utilizes FIR filter or/and linear interpolation is obtained
Interpolation generate prediction block；Subtrator, for by deducting prediction block generation residual block from the chromatic component of current block；Become
Change unit, for generating frequency transformed block by residual block is carried out conversion；Quantifying unit, for by frequency transformed block
Carry out quantization to generate sampling frequency transform block；And coding unit, for sampling frequency transform block is encoded to bit stream.
The another embodiment of the disclosure provides a kind of video encoder, comprising: predicting unit, in order to obtain
Obtain 1/2 also sub-sample of the reference block of the motion vector institute reference of current block, be suitable for filtering system by adjacent integer pixel values
Number obtains 1/2 amplifier value, and utilizes the value next life of all sub samplings of the described 1/2 amplifier value described reference block of acquisition
Become prediction block；Subtrator, for by from current block deduct prediction block generate residual block；Converter unit, for by right
Residual block carries out conversion to generate frequency transformed block；Quantifying unit, for carrying out growing amount by frequency transformed block is carried out quantization
Change frequency transformed block；And coding unit, for sampling frequency transform block is encoded in bit stream.
The another embodiment of the disclosure provides a kind of video decoding apparatus, comprising: decoding unit, for from bit
Stream extracts and quantifies frequency transformed block；Inverse quantization unit, for carrying out reconfiguration frequency by sampling frequency transform block is carried out re-quantization
Transform block；Inverse transformation block, for bringing reconstructed residual block by frequency transformed block is carried out inversion；Predicting unit, being used for will
The sub sampling component value of the reference block of the motion vector institute reference according to current block utilizes FIR filter and linear interpolation to be obtained
Interpolation generate prediction block；And adder unit, for by by reconstructed residual block and prediction Kuai Xiang Calais reconstruct current block.
The another embodiment of the disclosure provides a kind of video decoding apparatus, comprising: decoding unit, for from bit
Stream extracts and quantifies frequency transformed block；Inverse quantization unit, for carrying out reconfiguration frequency by sampling frequency transform block is carried out re-quantization
Transform block；Inverse transformation block, for bringing reconstructed residual block by frequency transformed block is carried out inversion；Predicting unit, be used for be
Obtain 1/2 also sub-sample of the reference block of the motion vector institute reference of current block, be suitable for filter by adjacent integer pixel values
Wave system number obtains 1/2 amplifier value, and utilizes the value of all sub samplings of the described 1/2 amplifier value described reference block of acquisition
Generate prediction block；And adder unit, for by the residual block of reconstruct and prediction Kuai Xiang Calais are reconstructed current block.
The another embodiment of the disclosure provides a kind of video coding/decoding method, including: perform at Video coding
Reason, this Video coding processes for being utilized by the sub sampling component value of the reference block of the motion vector institute reference according to current block
FIR filter is or/and the interpolation that obtained of linear interpolation is to generate prediction block, by deducting prediction from the chromatic component of current block
Block generates residual block, generates sampling frequency transform block by conversion quantized residual block, and by sampling frequency transform block
It is encoded in bit stream；And performing video decoding process, this video decoding process is for becoming from bit stream generating quantification frequency
Change block, by sampling frequency transform block being carried out re-quantization and inversion brings reconstructed residual block, by the motion according to current block to
The sub sampling component value of the reference block of amount institute reference utilizes interpolation that FIR filter and linear interpolation obtained to generate prediction
Block, and by the prediction Kuai Xiang Calais reconstruct current block by the residual block of reconstruct with generation.
The another embodiment of the disclosure provides a kind of method for video coding, including: perform prediction and process, at this prediction
Reason, and will be according to working as obtaining the motion vector of chromatic component by the chromatic component of current block carries out motion compensation
The interpolation that the sub sampling component value of the reference block of the motion vector institute reference of front piece utilizes FIR filter and linear interpolation to be obtained
Generate prediction block；Performing subtraction process, this subtraction process is for by deducting prediction block generation from the chromatic component of current block
Residual block；Performing conversion process, this conversion process is for generating frequency transformed block by residual block is carried out conversion；Execution amount
Change processes, and this quantification treatment is for generating sampling frequency transform block by frequency transformed block is carried out quantization；And perform volume
Code processes, and this coded treatment is for being encoded to sampling frequency transform block in bit stream.
The another embodiment of the disclosure provides a kind of method for video coding, including: perform prediction and process, at this prediction
Reason is for 1/2 also sub-sample of the reference block of the motion vector institute reference in order to obtain current block, by adjacent integer pixel
Value is suitable for filter factor and obtains 1/2 amplifier value, and utilizes described 1/2 amplifier value to obtain all sons of described reference block
The value of sampling generates prediction block；Perform subtraction process, this subtraction process for by from current block deduct prediction block generate
Residual block；Performing conversion process, this conversion process is for generating frequency transformed block by residual block is carried out conversion；Execution amount
Change processes, and this quantification treatment is for generating sampling frequency transform block by frequency transformed block is carried out quantization；And perform volume
Code processes, and this coded treatment is for being encoded to sampling frequency transform block in bit stream.
The another embodiment of the disclosure provides a kind of video encoding/decoding method, including: perform decoding process, at this decoding
Reason is for from bit stream generating quantification frequency transformed block；Execution re-quantization processes, and this re-quantization processes for by quantifying frequency
Rate transform block carries out re-quantization and carrys out reconfiguration frequency transform block；Performing inversion process, this inversion process is for by frequency
Transform block carries out inversion and brings reconstructed residual block；Execution prediction process, this prediction process for by the motion according to current block to
The sub sampling component value of the reference block of amount institute reference utilizes interpolation that FIR filter and linear interpolation obtained to generate prediction
Block；And execution addition process, this addition process is for by reconstructing current block by the residual block of reconstruct and prediction Kuai Xiang Calais
Chromatic component.
The another embodiment of the disclosure provides a kind of video encoding/decoding method, comprising: perform decoding process, this decoding
Process for from bitstream extraction sampling frequency transform block；Execution re-quantization processes, and this re-quantization processes for by quantization
Frequency transformed block carries out re-quantization and carrys out reconfiguration frequency transform block；Performing inversion process, this inversion process is for by frequency
Rate transform block carries out inversion and brings reconstructed residual block；Execution prediction processes, and this prediction processes for for the fortune obtaining current block
1/2 also sub-sample of the reference block of moving vector institute reference, is suitable for filter factor by adjacent integer pixel values and obtains 1/2 sampling
Value of magnification, and utilize described 1/2 amplifier value to obtain the value of all sub samplings of described reference block to generate prediction block；And
Performing addition process, this addition process is for by by the residual block of reconstruct and prediction Kuai Xiang Calais reconstruct current block.
The technique effect of the present invention
According to the disclosure as above, current block will be coded of by interpolation exactly and decrease actual block and prediction
Difference between block, thus improve code efficiency.Therefore, it is possible to by improving the compression efficiency of current block and considering coding staff
Method is carried out decoded transform and is carried out reconstructing video efficiently to the block of bit stream.
Accompanying drawing explanation
Fig. 1 is the figure illustrating typical nine kind of 4 × 4 intra prediction mode；
Fig. 2 is the figure illustrating typical four kind of 16 × 16 intra prediction mode；
Fig. 3 is the figure of the motion prediction of the typical sub sampling being shown with in luminance component；
Fig. 4 shows the example of typical 6 tap finite impulse response (FIR) wave filter；
Fig. 5 shows the example of the linear interpolation of typical chroma samples；
Fig. 6 is the block diagram schematically showing the video encoder according to embodiment of the present disclosure；
Fig. 7 is for describing the place utilizing also sub-sample in predicting unit interpolation block according to embodiment of the present disclosure
The exemplary diagram of reason；
Fig. 8 is the block diagram of the structure schematically showing the video decoding apparatus according to embodiment of the present disclosure；
Fig. 9 is the flow chart for describing the method for video coding according to embodiment of the present disclosure；And
Figure 10 is the flow chart for describing the video encoding/decoding method according to embodiment of the present disclosure.
Detailed description of the invention
Video encoder according to one or more embodiments as described below and video decoding apparatus can be
Such as personal computer (PC), notebook, personal digital assistant (PDA), portable media player (PMP), just
Take formula game machine (PSP) or the user terminal of wireless communication terminal, smart phone etc. or such as apps server and
The server terminal of service server, and represent equipped with such as various devices or wired/wireless communication network it
Between perform communication such as modem communicator, be used for encoding or decode video for storage or perform for compiling
Interframe or the various programs of infra-frame prediction and the memorizer of related data and being used for of code or decoding perform these programs with
Realize the various equipment of the microprocessor of operation and control.
Can be via including the Internet, short-distance wireless it addition, be encoded to the video of bit stream by video encoder
Or personal area network, wireless lan network, WiBro (also referred to as WiMax) network, mobile communications network or such as cable or logical
It is sent in real time or in non real-time for it is decoded after sending by the communication interface of universal serial bus (USB)
Video decoding apparatus, is reconstructed there and is reproduced as video.
In general, motion picture can be made up of a series of picture, and each picture is divided into the pre-of such as frame or block
Determine region.When picture is divided into block, the block after division can be classified as intra block (intra block) according to coded method
Or interframe block (inter block).Intra block refers to carry out the block that encodes, intraframe predictive coding by intra-frame predictive encoding method
Referring to following method, it is by utilizing the block through previous coding and decoding and rebuilding in the current pixel currently encoded
Pixel predicts the pixel of current block, thus generates prediction block, and carries out prediction block relative to the difference of the pixel of current block
Coding.Interframe block refers to carry out, by inter prediction encoding, the block that encodes, and this inter prediction encoding refers to following method, and it passes through
With reference to one or more passing pictures or in the future the current block in picture prediction photo current generate prediction block, and to prediction block
Encode with the difference of current block.For photo current being encoded and decodes and the frame of reference being referred to as reference frame.
Fig. 3 is the figure of the motion prediction of the typical sub sampling being shown with in luminance component.
With reference to Fig. 3, motion vector can obtained be the sub sampling position at most arriving 1/4 sampling in integer samples unit
Put.
Fig. 4 shows the example of 6 tap finite impulse response (FIR) wave filter.
With reference to Fig. 4, by use have six values of integer pixel values and six filter coefficients 1 ,-5,20,20 ,-
5,1} carry out the pixel value at interpolation 1/2 sampling location, and use 1/2 sampled pixel values after interpolation and integer pixel values or
Two 1/2 sampled pixel values utilize linear interpolation that 1/4 sampled components is carried out interpolation.As shown in following exemplary equation
Ground carries out interpolation to the pixel value of 1/4 sampling location.
Equation) b'=(1 × E+ (-5) × F+20 × G+20 × H+ (-5) × I+1 × J)
B=b'/32
A=((32 × G)+b')/64
Except, in addition to generating the typical method of sub sampling, there is also a lot for interpolation also sub-sample more accurately
Method and the method using different filter coefficients.For interpolation also sub-sample more accurately, when interpolation 1/2 is sampled, pass through
Integer pixel values is multiplied by 32 (rather than divided by 32) and then divided by 64,1/4 sampled value is carried out linear interpolation.Therefore,
The error amount caused due to the division arithmetic of midway can be reduced.
Fig. 5 shows the example of the linear interpolation of typical chroma samples.
With reference to Fig. 5, as shown in following exemplary equation, perform the bilinear interpolation using four integer pixel values with
By considering that the position of each sub sampling is multiplied by weighted value and the pixel value of 1/8 sub sampling is carried out interpolation.
(equation): a=[(8-dx) × (8-dy) × A+dx × (8-dy) × B+ (8-dx) × dy × C+dx × dy × D)]/
64
I.e., in the example of hgure 5,
{ a=[(6 × 5 × A)+(2 × 5 × B)+(6 × 3 × C)+(2 × 3 × D)]/64}
Owing to, in the case of 4:2:0 video format, carrier chrominance signal is 1/4 times of (level: 1/ of the resolution of luminance signal
2, vertical: 1/2), therefore perform 1/8 sub sampling interpolation.
In like fashion, when only using linear interpolation, with most 1/8 sampling resolutions, chroma samples is carried out interpolation, with 6
Tap FIR filters is compared, and it is inaccurate.Therefore, when chroma samples is encoded, compression efficiency is reduced.
Fig. 6 is the block diagram schematically showing the video encoder according to embodiment of the present disclosure.
The video encoder 600 of the one or more embodiments according to the disclosure is by utilizing the motion of luminance component
Vector value generates the sub sampling predictive value of chromatic component and encodes the current block of video.As shown in Figure 6, video is compiled
Decoding apparatus 600 can include predicting unit 610, subtrator 620, converter unit 630, quantifying unit 640 and coding unit
650。
To carry out, for unit input, the input video encoded with macro block.In this embodiment, macro block can have M × N
Form, wherein, M and N is respectively provided with 2nSize and can be equal to each other or different from each other.
The sub sampling component value of the reference block of the motion vector institute reference according to current block is utilized FIR by predicting unit 610
Wave filter is or/and the interpolation that obtained of linear interpolation is to generate prediction block.
Predicting unit 610 can generate prediction block to predict current macro by using other frame.I.e., it was predicted that unit
610 can generate motion vector by having been carried out the estimation in the front frame of the reconstruct of coded treatment, and use
This motion vector generates prediction block in motion compensation process.In this case, it was predicted that unit 610 can divide by using brightness
The motion vector value of amount uses the identical motion vector value in chromatic component, and can by the motion according to current block to
The sub sampling component value of the reference block of amount institute reference utilizes FIR filter or/and the interpolation that obtained of linear interpolation is pre-to generate
Survey block.
Subtrator 620 is pre-with generated by predicting unit 610 by the original pixel value calculating each pixel of current block
Difference between measured value generates residue signal.
The residue signal generated by subtrator 620 is transformed in frequency domain by converter unit 630.Converter unit 630 is permissible
By using for the such as discrete cosine transform (DCT) of time shaft video signal conversion to frequency axis being converted or small echo change
Residue signal is transformed in frequency domain by the various converter techniques changed.
Quantifying unit 640 is to including that the frequency transformed block being transformed to the residue signal in frequency domain by converter unit 630 is carried out
Quantify.As quantization method, it is possible to use various quantization methods.The example of quantization method includes that dead band uniform threshold quantifies
And quantization weight matrix (DZUTQ).
The sampling frequency transform block coding that coding unit 650 will be made up of the coefficient of frequency utilizing quantifying unit 640 to quantify
For bit stream.Entropy coding can be used to be used for encoding, but the disclosure is not limited to this, and can also use various
Coding techniques.
It addition, coding unit 650 can include by by the various information needed for decoding coded bit stream and sampling frequency
Coefficient coding in coded data and the bit stream obtained.That is, coded data can have and includes by fgs encoder block pattern
(CBP), delta quantization parameter and sampling frequency coefficient and the field of bit stream that obtains and include for being predicted required
Another field (such as, intra prediction mode in the case of infra-frame prediction, or the situation of inter prediction of bit of information
Under motion vector).
Inverse quantization unit 660 carries out re-quantization to the residual block (that is, sampling frequency transform block) after transform and quantization, and
Inverse transformation block 670 carries out inverse transformation to the conversion residual block after re-quantization.In like fashion, residual block is reconstructed.Re-quantization and
Inverse transformation can process by the transform and quantization performing in turn to be performed by converter unit 630 and quantifying unit 640 respectively
Perform.That is, inverse quantization unit 660 and inverse transformation block 670 can be raw by using from converter unit 630 and quantifying unit 640
The information about transform/quantization (such as, about the information of transform/quantization type) become and send performs re-quantization and inversion
Change.
Adder unit 680 is by by the prediction block generated by predicting unit 610 and the remnants generated by inverse transformation block 670
Block is added and generates reconstructed blocks.
Frame memory 690 stores the block reconstructed by adder unit 680, and use the block of this reconstruct as reference block with
Prediction block is generated in frame or in inter predication process.
Fig. 7 is for describing the also sub-sample utilized in predicting unit 610 interpolation block according to embodiment of the present disclosure
The exemplary diagram of process.
The interpolation of the sub sampling shown in Fig. 7 can be performed by the method using Equation 1 below to 4.
D'=(1 × A+ (-5) × B+20 × C+20 × D+ (-5) × E+1 × F)
D=d'/32 equation 1
Can obtain by adjacent integer pixel values being multiplied by predetermined value and multiplied value be added by using
1/2 amplifier value obtains 1/2 also sub-sample.
1/2 can be obtained by three integer pixel values A, B and C and right side integer pixel values D, E and F on the left of using
Also sub-sample.
B '=(32 × C)+d '
B=((32 × C)+d ')/64 equatioies 2
Can be by using nearest integer pixel values C and 1/2 amplifier value d ' obtain 1/4 also sub-sample.Pass through
To nearest integer pixel values C and 1/2 amplifier value d ' carry out linear interpolation " b " after carry out division arithmetic (rather than
Directly 1/2 sub sampling is carried out interpolation) obtain " b ".Therefore, it is possible to the error occurred when removing acquisition " d ".
A '=(64 × C)+b '
A=((64 × C)+b ')/128 equatioies 3
It also is able to by using nearest integer pixel values C and 1/2 amplifier value d ' obtain b '.Thus, it is also possible to
By using nearest integer pixel values C and 1/2 amplifier value d ' obtain 1/8 also sub-sample.
A=((w1 × (C × 32)+w2 × d ')/((w1+w2) × 32) equation 4
In equation 1,2,3 and 4 and Fig. 7, A, B, C, D, E and F represent the integer pixel values of chromatic component, and d, b
1/2 sub sampling of chromatic component, 1/4 sub sampling of chromatic component and 1/8 sub sampling of chromatic component is represented respectively with a.
As illustrated in equation 1, can by use high-precision filter (here, high-precision filter can use such as
The various wave filter of FIR filter) generate 1/2 also sub-sample of chromatic component.As shown in Equation 2,1/4 also sub-sample can
To use high-precision filter and linear interpolation.Especially, in the case of 1/8 sub sampling, as shown in Equation 3, Ke Yitong
Cross use integer pixel and 1/4 sub sampling to perform high-precision linear interpolation.It addition, as shown in equation 4, can be by using
Integer pixel and 1/2 sub sampling perform high-precision linear interpolation.In this case, represent will be with integer pixel and 1/ for w1 and w2
The weighted value that 2 sub samplings are multiplied.1/2 also sub-sample of equation 1 is used (such as, by using high accuracy FIR filter to generate
" d ") and 1/4 also sub-sample (such as, " b ") of use equation 2.Use etc. is generated by the linear interpolation between two pixel values
1/8 also sub-sample (such as, " a ") of formula 3.Therefore, with the situation carrying out all sub samplings of interpolation by being used alone linear interpolation
Compare, it is possible to generate and be worth more accurately.
The all division arithmetics used in equation 1 to 4 can improve place by using shift operation (＞ ＞, ＜ ＜)
Reason speed.
It addition, in order to be rounded to nearest integer in equation 1 to 4, can be in advance by the half of divisor and dividend phase
Add.Equation below 5 is that the equation of equation 3 is added in computing of wherein rounding off to.
A=((64 × C)+b '+64)/128 equatioies 5
FIR filter is a kind of digital filter and the predetermined value merely with input signal performs filtering.Therefore, if
Calculating is the impulse response of the characteristic function of wave filter, then FIR filter has finite length.It addition, FIR filter etc.
In formula, FIR does not have any feedback component.Therefore, when implementing identical characteristics, the order of magnitude increases and the time that performs increases.
But, FIR filter can be used when phase offset (that is, keeping the waveform between input and output) is important.
It addition, in the high accuracy FIR filter being used herein, by using FIR filter to generate 1/2 sub sampling, and
And as shown in Equation 2, not over using by the value (d ') by using FIR filter divided by 32 1/2 sub samplings obtained
Value (d in equation 1), but by divided by value (that is, the FIR filter obtained when obtaining 1/2 also sub-sample before 32
End value d ') with carry out linear interpolation by the value (32 × C) that integer pixel values is multiplied by 32 acquisitions and generate 1/4 sub sampling.
Therefore, it is possible to prevent the information dropout caused due to the division arithmetic in pilot process, thus more accurately value is carried out interpolation.
It addition, all also sub-sample (such as, the d ' of equation 1, the b ' of equation 2 and a ' of equation 3) and with addition of
Bit can serve as obtaining in the next step current block with the integer pixel values with the bit number identical with also sub-sample
The sampled value of prediction block.After obtaining prediction block, pixel value is divided by weighted value, thus predicts the bit often sampled of block
Number becomes equal to the bit number often sampled of current block.Therefore, it is possible to realize higher performance.
It addition, equation 4 to be described as the equation of the value after the interpolation for only calculating a (1/8 sub sampling).But,
Even if in the case of obtaining b (1/4 sub sampling), it is possible to come by being arranged differently than weighted value w1 and w2 with the situation obtaining a
Obtain b.In a similar fashion, if using f ' to replace b ' in equation 2, use f replace b and use D to replace C,
Then it is obtained in that f, and if use g ' to replace a ' in equation 3, use g to replace a, use f ' to replace b ', use f
Replace b and use D to replace C, being then obtained in that g.Meanwhile, it is capable to be 1/8 son by b and d is carried out interpolation obtaining
The c of sampling, and e can be obtained by d and f is carried out interpolation.Therefore, with a or g analogously it is possible to connect most by using
Near integer pixel values C and 1/2 amplifier value d ' obtain c and e.
Therefore, 1/2 sub sampling is obtained by the adjacent integer pixel values of acquisition.Can be by using more offending phase
Adjacent also sub-sample or close integer pixel values (that is, can use 1/2 sub sampling and integer when calculating 1/4 sampled value
Pixel) finer also sub-sample (1/4 sub sampling, 1/8 sub sampling etc.) is carried out interpolation.Those skilled in the art even exist
The sub sampling using also sub-sample or integer pixel values to be pointed to each position is derived in the case of not describing all sub samplings
Carry out the technical scheme of interpolation.
Further, it is possible to use except the high accuracy FIR filter being used herein and the interpolation method of linear interpolation.So
And, it is important that for 1/4 samples, use has 6 or more multitap FIR filter for 1/2 sampling and is held
Row interpolation is to have high accuracy function (as shown in Equation 2).
In the case of 4:2:0 video format, it is possible to by using equation 1 to 3 that the reference block in reference frame is carried out
Interpolation.In the case of 4:4:4 video format, luminance component and the resolution of chromatic component in reference block are equal.Therefore, with
Similarly, chromatic component needs only to be interpolated into most 1/4 sampling location luminance component.Therefore, if eliminate in equation 3 1/
The interpolation of 8 sampling locations, then it can be used in 4:4:4 video format.
It addition, in the case of 4:2:2 video format, the chromatic component of horizontal direction is the 1/ of the resolution of luminance component
2 times.Therefore, it is possible to by the chromatic component of horizontal direction being interpolated into 1/8 sub sampling and such as 4:4:4 as equation 1 to 3
Video format uses equation 1 and 2 that the chromatic component of vertical direction is interpolated into 1/4 sub sampling to generate sub sampling like that.
It addition, in the consideration test model (TMuC) of existing HD or higher image graphic, luminance component energy
Enough it is interpolated into 1/8 sub sampling.Therefore, in 4:2:0 video format, chromatic component can be interpolated into 1/16 sub sampling.Cause
This, be at best able to proceed to 1/8 sub-sample locations in the above-mentioned embodiment of the disclosure, and can be by using linear inserting
Value carries out interpolation to 1/16 sub-sample locations again.
Meanwhile, in the foregoing embodiment, using chromatic component as in the case of example to filtering and interpolation side
Method is described, but these methods can also apply to various pieces, such as, and luminance component and the block of R, G and B color, with
And chromatic component.In the coding/decoding method that will describe below, it is also possible to be equally applicable to chromatic component, luminance component and
Various pieces of the such as block of R, G and B color.
Fig. 8 is the block diagram of the structure schematically showing the video decoding apparatus according to embodiment of the present disclosure.
As shown in Figure 8, the video decoding apparatus 800 according to the one or more embodiments of the disclosure is bright by utilizing
The motion vector value of degree component generates the sub sampling predictive value of chromatic component and is decoded the current block of video.Video decodes
Equipment 800 includes decoding unit 810, inverse quantization unit 820, inverse transformation block 830, adder unit 840 and predicting unit 850.
Decoding unit 810 extracts quantization frequency transformed block by bit stream is carried out decoding.
Decoding unit 810 can decode or extract the information needed for decoding and sampling frequency by decoding coded data
Block.Information needed for decoding refers to decode the information needed for the coded bit stream in coded data.Such as, the information needed for decoding
Can be the information about block type, the information about motion vector, the information about transform/quantization type and other are various
Information.
That is, decoding unit 810 is extracted by the bit stream that decoding is the data encoded by video encoder 600 and includes
The sampling frequency transform block of the Pixel Information of the current block of video, and it is single that the information needed for the prediction of extraction is transferred to prediction
Unit 850.
Predicting unit 850 can be by using from solution in the way of identical with the predicting unit 610 of video encoder 600
Information needed for the prediction of code unit 810 transmission predicts current block.
The sub sampling component value of the reference block of the motion vector institute reference according to current block is utilized FIR by predicting unit 850
Wave filter is or/and the interpolation that obtained of linear interpolation is to generate prediction block.When by using the luminance component from bit stream reconstruct
Motion vector when obtaining the motion vector of chromatic component of current block, by using high accuracy FIR filter and linear filtering
Device generates predictive value according to the integer pixel values of the chromatic component of the reference block of motion vector institute reference.
The predicting unit 850 of the video encoder 800 of the one or more embodiments according to the disclosure with reference
The similar mode of predicting unit 610 of Fig. 6 video encoder described above 600 generates sub sampling.Therefore, in order to avoid
Repetitive description, will omit it and describe in detail.
Inverse quantization unit 820 carries out re-quantization to by decoding unit 810 from the sampling frequency transform block of bitstream extraction.Inverse
Converter unit 830 is by by the frequency transformed block inverse transformation of inverse quantization unit 820 re-quantization to time domain.
Adder unit 840 is by by the pixel value generated by predicting unit 850 with by the inverse transformation of inverse transformation block 830
The original pixel value of the residue signal phase Calais reconstruct current block of reconstruct.The current block reconstructed by adder unit 840 can be passed
It is passed to frame memory 860, and may be used for predicting other block in predicting unit 850.
Frame memory 860 stores reconstructing video and makes it possible to generate intra-frame prediction block and interframe prediction block.
Can be by the bit stream outfan of the video encoder 600 of Fig. 6 being connected to the video decoding apparatus of Fig. 8
The bit stream input of 800 constructs the video encoding/decoding apparatus according to embodiment of the present disclosure.
Video encoding/decoding apparatus according to embodiment of the present disclosure includes: video encoder, and being used for will be according to currently
The sub sampling component value of the reference block of the motion vector institute reference of block utilize FIR filter or/and linear interpolation obtained insert
Value generates prediction block, generates residual block by deducting prediction block from the chromatic component of current block, by converting and quantifying residual
Remaining block generates sampling frequency transform block, and sampling frequency transform block is encoded to bit stream；And Video Decoder, it is used for
From bit stream generating quantification frequency transformed block, by sampling frequency transform block being carried out re-quantization and inverse transformation reconstructed residual block,
The sub sampling component value of the reference block of the motion vector institute reference according to current block is utilized FIR filter or/and linear interpolation
The interpolation obtained is to generate prediction block, and reconstructs current by the residual block that will reconstruct with the prediction Kuai Xiang Calais of generation
Block.
Can utilize and implement Video coding according to the video encoder 600 of disclosed one or more embodiments
Device, and the video decoding apparatus 600 according to one or more embodiments can be utilized to implement Video Decoder.
Fig. 9 is the flow chart for describing the method for video coding according to embodiment of the present disclosure.
Video encoder 600 carrys out encoded video by performing following step: prediction steps S910, for being regarded by use
The motion vector value of the luminance component in the current block of frequency generates the sub sampling predictive value of chromatic component；Subtraction process S920,
For generating residue signal by the difference between original pixel value and the predicted pixel values of calculating current block；Shift step
S930, for by using dct transform or wavelet transformation to be transformed in frequency domain by the residue signal of generation；Quantization step
S940, for quantifying the residue signal being converted in frequency domain；And coding step S950, for by sampling frequency conversion remnants
Signal coding is bit stream.
Due to prediction steps S910, subtraction process S920, shift step S930, quantization step S940 and coding step S950
Correspond respectively to predicting unit 610, subtrator 620, converter unit 630, quantifying unit 640 and the function of coding unit 650,
Therefore will omit it to describe in detail.
Figure 10 is the flow chart for describing the video encoding/decoding method according to embodiment of the present disclosure.
Led to by the video decoding apparatus 800 of wired/wireless communication network or cable reception the bit stream storing video
Cross and utilize the motion vector value of luminance component generate the sub sampling predictive value of chromatic component and be decoded reconstructing to video
The current block of video, with the selection according to user or other operation program restructuring video.
To this end, the bit stream that video decoding apparatus 800 receives by performing following step to decode: decoding step S1010,
For decoding bit stream to extract the sampling frequency conversion residue signal of the information of the pixel value of the current block representing video；Inverse amount
Change step S1020, for sampling frequency conversion residue signal is carried out re-quantization；Inverse transformation step S1030, for by re-quantization
After frequency transformation residue signal inverse transformation in time domain；Prediction steps S1040, for according to by by inverse transformation to time domain
And the predictive value of the current block that the residue signal reconstructed represents uses the motion vector value of luminance component to generate chromatic component
Sub sampling predictive value；And adding step S1050, for by by the residue signal of the current block of reconstruct in step S1030
Original pixel value with the predicted pixel values phase Calais reconstruct current block of each pixel of the current block of prediction in step S1040.
Due to decoding step S1010, inverse quantization step S1020, inverse transformation step S1030, prediction steps S1040 and addition
Step S1050 corresponds respectively to decoding unit 810, inverse quantization unit 820, inverse transformation block 830, predicting unit 850 and addition
The operation of unit 840, therefore will omit it and describe in detail.
The video coding/decoding method of the one or more embodiments according to the disclosure can be by according to the disclosure
The method for video coding of one or more embodiments and the video decoding side of the one or more embodiments according to the disclosure
The combination of method realizes.
Video coding/decoding method according to embodiment of the present disclosure includes: Video coding step, and being used for will be according to working as
The sub sampling component value of the reference block of the motion vector institute reference of front piece utilizes FIR filter or/and what linear interpolation was obtained
Interpolation generates prediction block, generates residual block by deducting prediction block from the chromatic component of current block, by converting and quantifying
Residual block generates sampling frequency transform block, and sampling frequency transform block is encoded to bit stream；And video decoding step,
For from bit stream generating quantification frequency transformed block, by sampling frequency transform block is carried out re-quantization and inversion to bring reconstruct residual
Remaining block, utilizes the sub sampling component value of the reference block of the motion vector institute reference according to current block FIR filter and linearly inserts
The interpolation that value is obtained is to generate prediction block, and reconstructs current by the residual block that will reconstruct with the prediction Kuai Xiang Calais of generation
Block.
The Video coding step that can utilize the one or more embodiments according to the disclosure implements Video coding step
And the video decoding step according to one or more embodiments can be utilized to implement video decoding step suddenly,.
In superincumbent description, although all components of multiple embodiments of the present invention have described as be assembled into or
Being operably connected is a unit, but the present invention is not intended to be limited to itself these embodiments.On the contrary, at this
In the target zone of invention, corresponding assembly can optionally and operationally press any number combinations.Each assembly itself is also
Can realize by hardware, simultaneously corresponding assembly partly or optionally can be combined as entirety and uses and have
Realize for performing the computer program of the program module of the function of hardware equivalents.Those skilled in the art can be easily
Infer code or the code segment for constituting this program.This computer program can be stored in computer-readable medium, its
Embodiments of the present invention can be realized when operation.Computer-readable medium can include magnetic recording media, optical recording media,
And carrier media.
Although have been described with the illustrative embodiments of the present invention for exemplary purposes, but those skilled in the art
It should be clear that in the case of without departing from the basic feature of the present invention, various amendment can be carried out, adds and replace.Therefore,
The illustrative embodiments of the present invention is not described for restricted purpose.Therefore, the scope of the present invention non-through
Cross above-mentioned embodiment but limited by claims and equivalent thereof.
Industrial applicability
As it has been described above, the disclosure is very suitable for when generating the sub sampling predictive value of video for minimizing actual component
And the interpolation of the difference between anticipation component and utilize this interpolation to carry out the field of Video coding and decoding.Due to higher essence
Degree prediction will be coded of current block, therefore, it is possible to increase code efficiency by reducing the difference between actual block and prediction block.
Therefore, it is possible to raising compression efficiency.
Cross-Reference to Related Applications
If be suitable for, then this application claims at the patent application No.10-2010-that Korea S submitted on October 6th, 2010
0097547 and on July 21st, 2011 submit to patent application No.10-2011-0072196 under 35U.S.C § 119 (a)
Priority, entire contents is incorporated herein by.It addition, this non-provisional application is based on this korean patent application, with
Same reasons claims the priority in other country in addition to the U.S., entire contents is incorporated herein by.
Claims (30)
1. a video encoder, described video encoder includes:
Predicting unit, for utilizing FIR to filter the sub sampling component value of the reference block of the motion vector institute reference according to current block
Ripple device is or/and the interpolation that obtained of linear interpolation is to generate prediction block；
Subtrator, for generating residual block by deducting described prediction block from described current block；
Converter unit, for generating frequency transformed block by described residual block is carried out conversion；
Quantifying unit, for generating sampling frequency transform block by described frequency transformed block is carried out quantization；And
Coding unit, for described sampling frequency transform block is encoded to bit stream,
Wherein, described predicting unit is configured to:
By using described FIR filter to generate 1/2 also sub-sample, and
Not over using by described 1/2 also sub-sample that FIR filter result value is obtained divided by weight, but by logical
Cross integer pixel values is multiplied by described weight obtain value carry out with the described FIR filter result value not being divided by by described weight
Linear interpolation generates 1/4 sub sampling, and wherein said weight is the summation of filter coefficient.
Video encoder the most according to claim 1, wherein, the motion vector of the chromatic component of current block uses described
The motion vector value of the luminance component of current block.
Video encoder the most according to claim 1, wherein, when described video has 4:2:0 form, described prediction
Unit is by using integer pixel and the linear interpolation of 1/4 sub sampling or by using the linear of integer pixel and 1/2 sub sampling
Interpolation generates 1/8 also sub-sample.
Video encoder the most according to claim 1, wherein, filters for the described FIR of described 1/2 sub sampling component
Device is to have 6 or more multitap FIR filter.
Video encoder the most according to claim 1, wherein, described predicting unit is interpolated into the 1/16 of described reference block
The position of sub sampling, the position of described 1/16 sub sampling is generated by linear interpolation.
Video encoder the most according to claim 1, wherein, when described video has 4:4:4 form, described prediction
Unit is interpolated into position and the position of 1/4 sub sampling of 1/2 sub sampling of chromatic component.
Video encoder the most according to claim 1, wherein, when described video has 4:2:2 form, described prediction
Unit is interpolated into the position of 1/8 sub sampling of the chromatic component of horizontal direction, and be interpolated into the chromatic component of vertical direction
The position of 1/4 sub sampling.
8. a video encoder, described video encoder includes:
Predicting unit, for 1/2 also sub-sample of the reference block of the motion vector institute reference in order to obtain current block, by adjacent
Integer pixel values be suitable for filter coefficient and obtain 1/2 amplifier value, and utilize described 1/2 amplifier value to obtain described base
The value of all sub samplings of quasi-block generates prediction block, and wherein, described 1/2 amplifier value is by by adjacent integer pixel
Value is multiplied by filter coefficient and the value then multiplied value addition obtained；
Subtrator, for generating residual block by deducting described prediction block from described current block；
Converter unit, for generating frequency transformed block by described residual block is carried out conversion；
Quantifying unit, for generating sampling frequency transform block by described frequency transformed block is carried out quantization；And
Coding unit, for being encoded to bit stream by described sampling frequency transform block.
9. a video decoding apparatus, described video decoding apparatus includes:
Decoding unit, for from bitstream extraction sampling frequency transform block；
Inverse quantization unit, for carrying out reconfiguration frequency transform block by described sampling frequency transform block is carried out re-quantization；
Inverse transformation block, for bringing reconstructed residual block by described frequency transformed block is carried out inversion；
Predicting unit, for utilizing FIR to filter the sub sampling component value of the reference block of the motion vector institute reference according to current block
The interpolation that ripple device and linear interpolation are obtained is to generate prediction block；And
Adder unit, reconstructs described current block for the residual block by reconstructing and described prediction Kuai Xiang Calais,
Wherein, described predicting unit is configured to:
By using described FIR filter to generate 1/2 also sub-sample, and
Not over using by described 1/2 also sub-sample that FIR filter result value is obtained divided by weight, but by logical
Cross integer pixel values is multiplied by described weight obtain value carry out with the described FIR filter result value not being divided by by described weight
Linear interpolation generates 1/4 sub sampling, and wherein said weight is the summation of filter coefficient.
Video decoding apparatus the most according to claim 9, wherein, the motion vector of the chromatic component of current block uses institute
State the motion vector value of the luminance component of current block.
11. video decoding apparatus according to claim 9, wherein, when described video has 4:2:0 form, described pre-
Survey unit by using integer pixel and the linear interpolation of 1/4 sub sampling or by using integer pixel and the line of 1/2 sub sampling
Property interpolation generates 1/8 also sub-sample.
12. video decoding apparatus according to claim 9, wherein, filter for the described FIR of described 1/2 sub sampling component
Ripple device is to have 6 or more multitap FIR filter.
13. video decoding apparatus according to claim 9, wherein, described predicting unit is interpolated into the 1/ of described reference block
The position of 16 sub samplings, the position of described 1/16 sub sampling is generated by linear interpolation.
14. video decoding apparatus according to claim 9, wherein, when described video has 4:4:4 form, described pre-
Survey unit is interpolated into position and the position of 1/4 sub sampling of 1/2 sub sampling of chromatic component.
15. video decoding apparatus according to claim 9, wherein, when described video has 4:2:2 form, described pre-
Survey unit is interpolated into the position of 1/8 sub sampling of the chromatic component of horizontal direction, and is interpolated into the chromatic component of vertical direction
The position of 1/4 sub sampling.
16. 1 kinds of video decoding apparatus, described video decoding apparatus includes:
Decoding unit, for from bitstream extraction sampling frequency transform block；
Inverse quantization unit, for carrying out reconfiguration frequency transform block by described sampling frequency transform block is carried out re-quantization；
Inverse transformation block, for bringing reconstructed residual block by described frequency transformed block is carried out inversion；
Predicting unit, for 1/2 also sub-sample of the reference block of the motion vector institute reference in order to obtain current block, by adjacent
Integer pixel values be suitable for filter coefficient and obtain 1/2 amplifier value, and utilize described 1/2 amplifier value to obtain described base
The value of all sub samplings of quasi-block generates prediction block, and wherein, described 1/2 amplifier value is by by adjacent integer pixel
Value is multiplied by filter coefficient and the value then multiplied value addition obtained；And
Adder unit, for by reconstructing described current block by the residual block of reconstruct and described prediction Kuai Xiang Calais.
17. 1 kinds of method for video coding, described method for video coding includes:
Execution prediction processes, and described prediction processes for obtaining motion vector by current block is carried out motion compensation, and
FIR filter and linear interpolation is utilized to be obtained the sub sampling component value of the reference block of the motion vector institute reference according to current block
The interpolation obtained generates prediction block；
Performing subtraction process, described subtraction process is for by deducting described prediction block generation residual block from described current block；
Performing conversion process, described conversion process is for generating frequency transformed block by described residual block is carried out conversion；
Perform quantification treatment, described quantification treatment for by described frequency transformed block is carried out quantization generate sampling frequency become
Change block；And
Performing coded treatment, described coded treatment is used for described sampling frequency transform block is encoded to bit stream,
Wherein, 1/2 also sub-sample is generated by the described FIR filter of use, and
Wherein, 1/4 sub sampling is not over using by being adopted by described 1/2 son that FIR filter result value obtains divided by weight
Sample value, but by by described in integer pixel values being multiplied by value that described weight obtains and not being divided by by described weight
FIR filter result value carries out what linear interpolation generated, and wherein said weight is the summation of filter coefficient.
18. method for video coding according to claim 17, wherein, when described video has 4:2:0 form, described pre-
Survey processes by using integer pixel and the linear interpolation of 1/4 sub sampling or by using integer pixel and the line of 1/2 sub sampling
Property interpolation generates 1/8 also sub-sample.
19. method for video coding according to claim 17, wherein, filter for the described FIR of described 1/2 sub sampling component
Ripple device is to have 6 or more multitap FIR filter.
20. method for video coding according to claim 17, wherein, described prediction processes and is interpolated into the 1/ of described reference block
The position of 16 sub samplings, the position of described 1/16 sub sampling is generated by linear interpolation.
21. method for video coding according to claim 17, wherein, when described video has 4:4:4 form, described pre-
Survey processes position and the position of 1/4 sub sampling of 1/2 sub sampling being interpolated into chromatic component.
22. method for video coding according to claim 17, wherein, when described video has 4:2:2 form, described pre-
Survey processes the position of 1/8 sub sampling of the chromatic component being interpolated into horizontal direction, and is interpolated into the chromatic component of vertical direction
The position of 1/4 sub sampling.
23. 1 kinds of method for video coding, described method for video coding includes:
Execution prediction processes, and described prediction processes 1/2 of the reference block for the motion vector institute reference in order to obtain current block
Also sub-sample, is suitable for filter coefficient by adjacent integer pixel values and obtains 1/2 amplifier value, and utilize described 1/2 to adopt
The value of all sub samplings that sample value of magnification obtains described reference block generates prediction block, and wherein, described 1/2 amplifier value is logical
Cross the value adjacent integer pixel values being multiplied by filter coefficient and then multiplied value addition being obtained；
Performing subtraction process, described subtraction process is for generating residual block by deducting described prediction block from described current block；
Performing conversion process, described conversion process is for generating frequency transformed block by described residual block is carried out conversion；
Perform quantification treatment, described quantification treatment for by described frequency transformed block is carried out quantization generate sampling frequency become
Change block；And
Performing coded treatment, described coded treatment is for being encoded to bit stream by described sampling frequency transform block.
24. 1 kinds of video encoding/decoding methods, described video encoding/decoding method includes:
Performing decoding process, described decoding process is for from bit stream generating quantification frequency transformed block；
Execution re-quantization processes, and described re-quantization processes for reconstructing by described sampling frequency transform block is carried out re-quantization
Frequency transformed block；
Performing inversion process, described inversion process is for bringing reconstructed residual by described frequency transformed block is carried out inversion
Block；
Execution prediction processes, and described prediction processes for the sub sampling by the reference block of the motion vector institute reference according to current block
Component value utilizes interpolation that FIR filter and linear interpolation obtained to generate prediction block；And
Performing addition process, described addition process is for by by described to residual block and the reconstruct of described prediction Kuai Xiang Calais of reconstruct
Current block,
Wherein, 1/2 also sub-sample is generated by the described FIR filter of use, and
Wherein, 1/4 sub sampling is not over using by being adopted by described 1/2 son that FIR filter result value obtains divided by weight
Sample value, but by by described in integer pixel values being multiplied by value that described weight obtains and not being divided by by described weight
FIR filter result value carries out what linear interpolation generated, and wherein said weight is the summation of filter coefficient.
25. video encoding/decoding methods according to claim 24, wherein, when described video has 4:2:0 form, described pre-
Survey processes by using integer pixel and the linear interpolation of 1/4 sub sampling or by using integer pixel and the line of 1/2 sub sampling
Property interpolation generates 1/8 also sub-sample.
26. video encoding/decoding methods according to claim 24, wherein, filter for the described FIR of described 1/2 sub sampling component
Ripple device is to have 6 or more multitap FIR filter.
27. video encoding/decoding methods according to claim 24, wherein, described prediction processes and is interpolated into the 1/ of described reference block
The position of 16 sub samplings, the position of described 1/16 sub sampling is generated by linear interpolation.
28. video encoding/decoding methods according to claim 24, wherein, when described video has 4:4:4 form, described pre-
Survey processes position and the position of 1/4 sub sampling of 1/2 sub sampling being interpolated into chromatic component.
29. video encoding/decoding methods according to claim 24, wherein, when described video has 4:2:2 form, described pre-
Survey processes the position of 1/8 sub sampling of the chromatic component being interpolated into horizontal direction, and is interpolated into the chromatic component of vertical direction
The position of 1/4 sub sampling.
30. 1 kinds of video encoding/decoding methods, described video encoding/decoding method includes:
Performing decoding process, described decoding process is for from bitstream extraction sampling frequency transform block；
Execution re-quantization processes, and described re-quantization processes for reconstructing by described sampling frequency transform block is carried out re-quantization
Frequency transformed block；
Performing inversion process, described inversion process is for bringing reconstructed residual by described frequency transformed block is carried out inversion
Block；
Execution prediction processes, and described prediction processes 1/2 of the reference block for the motion vector institute reference in order to obtain current block
Also sub-sample, is suitable for filter coefficient by adjacent integer pixel values and obtains 1/2 amplifier value, and utilize described 1/2 to adopt
The value of all sub samplings that sample value of magnification obtains described reference block generates prediction block, and wherein, described 1/2 amplifier value is logical
Cross the value adjacent integer pixel values being multiplied by filter coefficient and then multiplied value addition being obtained；And
Performing addition process, described addition process is for by by described to residual block and the reconstruct of described prediction Kuai Xiang Calais of reconstruct
Current block.
Priority Applications (4)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610349136.2A CN105915915B (en) | 2010-10-06 | 2011-10-06 | A kind of video decoding apparatus |
| CN201610350084.0A CN105847808B (en) | 2010-10-06 | 2011-10-06 | Use the method and apparatus of high-precision filter coding/decoding video |
| CN201610348945.1A CN105812800B (en) | 2010-10-06 | 2011-10-06 | A kind of video decoding apparatus |
| CN201610348345.5A CN105812822B (en) | 2010-10-06 | 2011-10-06 | Use the method and apparatus of high-precision filter coding/decoding video |
Applications Claiming Priority (5)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| KR20100097547 | 2010-10-06 | ||
| KR10-2010-0097547 | 2010-10-06 | ||
| KR1020110072196A KR101341993B1 (en) | 2010-10-06 | 2011-07-21 | Method and Apparatus for Image Encoding/Decoding Using High Resolution Filter |
| KR10-2011-0072196 | 2011-07-21 | ||
| PCT/KR2011/007418 WO2012047047A2 (en) | 2010-10-06 | 2011-10-06 | Method and apparatus for encoding/decoding video using high-precision filter |
Related Child Applications (4)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610349136.2A Division CN105915915B (en) | 2010-10-06 | 2011-10-06 | A kind of video decoding apparatus |
| CN201610350084.0A Division CN105847808B (en) | 2010-10-06 | 2011-10-06 | Use the method and apparatus of high-precision filter coding/decoding video |
| CN201610348945.1A Division CN105812800B (en) | 2010-10-06 | 2011-10-06 | A kind of video decoding apparatus |
| CN201610348345.5A Division CN105812822B (en) | 2010-10-06 | 2011-10-06 | Use the method and apparatus of high-precision filter coding/decoding video |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103210649A CN103210649A (en) | 2013-07-17 |
| CN103210649B true CN103210649B (en) | 2016-12-14 |
Family
ID=
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2004015999A1 (en) * | 2002-08-08 | 2004-02-19 | Matsushita Electric Industrial Co., Ltd. | Moving picture encoding method and decoding method |
| CN101641960A (en) * | 2007-01-22 | 2010-02-03 | 三星电子株式会社 | Method and apparatus for encoding and decoding image using adaptive interpolation filter |
| CN101641961A (en) * | 2007-03-28 | 2010-02-03 | 三星电子株式会社 | Image encoding and decoding method and apparatus using motion compensation filtering |
| WO2010044569A3 (en) * | 2008-10-16 | 2010-07-08 | 에스케이텔레콤 주식회사 | Method and apparatus for generating a reference frame, and method and apparatus for image encoding/decoding using same |
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2004015999A1 (en) * | 2002-08-08 | 2004-02-19 | Matsushita Electric Industrial Co., Ltd. | Moving picture encoding method and decoding method |
| CN101641960A (en) * | 2007-01-22 | 2010-02-03 | 三星电子株式会社 | Method and apparatus for encoding and decoding image using adaptive interpolation filter |
| CN101641961A (en) * | 2007-03-28 | 2010-02-03 | 三星电子株式会社 | Image encoding and decoding method and apparatus using motion compensation filtering |
| WO2010044569A3 (en) * | 2008-10-16 | 2010-07-08 | 에스케이텔레콤 주식회사 | Method and apparatus for generating a reference frame, and method and apparatus for image encoding/decoding using same |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105915915B (en) | A kind of video decoding apparatus | |
| CN106375765B (en) | By using the method and apparatus of the inter prediction encoding/decoding image of any block | |
| US9686549B2 (en) | Method and apparatus for generating a reference frame and method and apparatus for encoding/decoding image using the same | |
| CN106331725B (en) | Motion information reconstructing arrangement and motion information generate equipment | |
| US9137545B2 (en) | Image encoding and decoding apparatus and method | |
| CN103210646B (en) | Intra prediction mode group is selected to carry out the method and apparatus to encoding/decoding image by effective | |
| CN104041048A (en) | Method And Apparatus Video Encoding And Decoding Using Skip Mode | |
| KR20130003718A (en) | Method and apparatus for image encoding/decoding by fast coding unit mode decision | |
| KR20120015495A (en) | Method and apparatus for encoding/decoding of video data capable of skipping filtering mode | |
| CN103210649B (en) | The method and apparatus using high-precision filter coding/decoding video | |
| KR20130098465A (en) | Video encoding apparatus and method thereof, and video decoding apparatus and method thereof | |
| KR20110042602A (en) | Video coding method and apparatus by using filtering motion compensation frame | |
| KR20140049613A (en) | Reference frame creating method and apparatus and video encoding/decoding method and apparatus using same |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant |