CN1706198A - Drift-free video encoding and decoding method, and corresponding devices - Google Patents
Drift-free video encoding and decoding method, and corresponding devices Download PDFInfo
- Publication number
- CN1706198A CN1706198A CNA2003801015409A CN200380101540A CN1706198A CN 1706198 A CN1706198 A CN 1706198A CN A2003801015409 A CNA2003801015409 A CN A2003801015409A CN 200380101540 A CN200380101540 A CN 200380101540A CN 1706198 A CN1706198 A CN 1706198A
- Authority
- CN
- China
- Prior art keywords
- resolution
- sequence
- low
- time
- gof
- 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.)
- Pending
Links
Images
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
- H04N19/60—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
- H04N19/61—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding in combination with predictive coding
- H04N19/615—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding in combination with predictive coding using motion compensated temporal filtering [MCTF]
-
- 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
- H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
- H04N19/503—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal prediction
- H04N19/51—Motion estimation or motion compensation
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
- H04N19/177—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being a group of pictures [GOP]
-
- 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
- H04N19/30—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using hierarchical techniques, e.g. scalability
- H04N19/31—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using hierarchical techniques, e.g. scalability in the temporal domain
-
- 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
- H04N19/30—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using hierarchical techniques, e.g. scalability
- H04N19/33—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using hierarchical techniques, e.g. scalability in the spatial domain
-
- 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
- H04N19/60—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
- H04N19/61—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding in combination with predictive coding
-
- 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
- H04N19/60—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
- H04N19/63—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding using sub-band based transform, e.g. wavelets
-
- 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
- H04N19/60—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
- H04N19/63—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding using sub-band based transform, e.g. wavelets
- H04N19/635—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding using sub-band based transform, e.g. wavelets characterised by filter definition or implementation details
-
- 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
- H04N19/13—Adaptive entropy coding, e.g. adaptive variable length coding [AVLC] or context adaptive binary arithmetic coding [CABAC]
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
Three-dimensional (3D) subband coding schemes use motion compensation in their temporal filtering stage. Unfortunately, this procedure introduces two drawbacks : (a) the MC being applied at the full resolution, a drift appears when decoding at a lower resolution, and (b) all the motion vectors estimated at full resolution are transmitted, which is a waste of bits. According to the invention, a low resolution sequence is first obtained by generating from the original input sequence of frames - by means of a wavelet decomposition - a sequence of low resolution frames and performing on them a motion compensated spatio-temporal analysis. Then, a motion compensated spatio-temporal analysis of each full resolution group of frames is performed, and the low frequency subbands of the decomposition are finally replaced, at each temporal decomposition level, by the corresponding spatio-temporal subbands of the generated low resolution sequence. The modified sequence thus obtained is finally coded. Thanks to this approach, a good behavior at low resolution is maintained (no more drift) while getting closer to the performance of a classic 3D subband codec at full resolution.
Description
Technical field
The present invention relates to a kind of coding method that the original video sequence that is divided into successive frame group (GOF) is compressed, and relate to a kind of corresponding coding/decoding method.The invention still further relates to respective coding and decoding device.
Background technology
The expansion of the Internet and the progress of multimedia technology are that video compression has been brought new application and service.In order to adapt to variety of network environments and terminal capabilities: scalability satisfies these demands, and many so new application and service not only have requirement to the efficient of coding, also require stronger function and flexibility.Present has often comprised some scalability features based on the video compression standard that mixes DCT (discrete cosine transform, discrete cosine transform) predict.These mixed structures are based on a prediction scheme, in this prediction scheme, each frame of time prediction from a given reference frame (optionally prediction is the bi-directional predicted of the forward prediction of P frame or B frame), and then thus obtained predicated error is carried out space conversion (adopting the two-dimensional dct conversion in standard scheme) so that utilize spatial redundancies.Scalability is to be obtained by additional enhancement layer.
Instead, three-dimensional (3D) subband video coding technology produces the single telescopic embedded bit stream fully that has.They depend on space-time filtering, and this space-time filtering allows to rebuild with any required spatial resolution or frame rate.Such method for example is the IEEE Transactions on ImageProcessing that people such as Podilchuk are published in February nineteen ninety-five, the disclosed method of article " Three-dimensionalsubband coding of video " of the 4th volume the 2nd phase 125-139 page or leaf, wherein frame group (GOF) is used as three-dimensional (2D+t, or 3D) structure is handled and by space-time filtering, so that the energy in the compression low frequency (further research comprises the motion compensation of this scheme, to improve whole code efficiency).
The 3D sub band structure of method acquisition has thus been described in Fig. 1, Fig. 1 shows the 3D wavelet decomposition with motion compensation that is applied to a frame group (GOF), and at first current GOF is carried out motion compensation (MC) has big motion with processing sequence, and then utilize Ha Er small echo (Haarwavelet) that the 3D small echo is carried out time filtering (TF) (dotted arrow is corresponding to high-pass time filtering, and another is corresponding to the low pass time filtering).After operation of motion compensation and time filtering operation, each time subband spatially is broken down into a space-time subband, the last 3D Wavelet representation for transient that produces original GOF of this space-time subband has illustrated three phases (L and the H=phase I of decomposing in the example of Fig. 1; LL and LH=second stage; LLL and LLH=phase III).Then, in order with respect to the space-time decomposition texture to be encoded in last coefficient bits plane effectively, select the known spiht algorithm that extends to 3D from 2D.
When implementing, in coder side, this 3D sub band structure is used the analysis of motion compensation (MC) space-time with full original resolution.Spatial scalability is to obtain by the high spatial subbands of removing in the decomposition.Yet, when motion compensation is used to the 3D analytical plan, this method can not realize the reconstruction fully of the video sequence of low resolution, even under the situation of very high bit rate, can not realize, this phenomenon is called as drift (drift) hereinafter, with compare with the final display size direct coding of target, this phenomenon has reduced the visual quality of scalable scheme.As being published in people such as P.Y.Cheng as described in the document " Multiscale video compression using wavelet transform and motioncompensation " on Proceedings of the InternationalConference on Image Processing (ICIP95) nineteen ninety-five the 1st the 606th to 609 page of volume, described drift is produced by the order of wavelet transformation and motion compensation, and this is immutable in proper order.When allow spatial scalability at decoder-side, the high spatial subbands of the decomposition of carrying out in coder side is omitted so, and this allows the low-definition version a of primitive frame A
dReconstruction or synthetic.In order to carry out such synthesizing, use following operation:
a＝DWT
L(L)+MC[DWT
L(H)]
＝DWT
L(A)+[MC[DWT
L(H)]-DWT
L(MC[H])] (1)
DWT wherein
L(discrete cosine transform of carrying out in the spatial domain) expression utilizes the resolution reduced sampling of analyzing identical wavelet filter with 3D.In preferred scalable scheme, wish to have:
a＝DWT
L(A) (2)
Thereby the remainder of expression formula (1) is equivalent to drift.It may be noted that if do not use MC drift is eliminated so.If frame is used single motion vector, so also can produce identical phenomenon (except in the image border).Yet hereinafter as can be known, in order to obtain good code efficiency, MC is inevitably, and the possibility of single all motions is very little, is enough to eliminate these special circumstances.
Some authors for example J.W.Wood etc. have openly eliminated the technical scheme of the suggestion of this drift in the article " A resolution and frame-rate scalablesubband/wavelet video coder " of IEEETransactions on Circuits and Systems for Video Technology the 1st volume the 9th phase 1035-1044 page or leaf that is published in September calendar year 2001.Yet in this article, above-mentioned scheme also means the extra information (the correctly synthetic required drift bearing of high-resolution) of transmission in bit stream except unusual complexity, caused the waste of some bits.The scheme of describing in the article of " the Multiscale videocompression...... " that quotes has in front been avoided this bottleneck, but this scheme is effective to prediction scheme, and the 3D subband codec of can not migrating.
In the european patent application the 02290155.7th (PHFR020002) of application on January 22nd, 2002, advised a scheme of avoiding these shortcomings, and, be used for the method for video coding that the original video sequence that is divided into successive frame group (GOF) compresses is comprised the steps: according to this scheme
(1) utilize wavelet-decomposing method from original video sequence, to produce a low-resolution sequence that comprises continuous low resolution GOF;
(2) by each low resolution GOF is carried out the analysis of motion compensation space-time, above-mentioned low-resolution sequence is carried out a low resolution decomposition;
(3) be fixed to described low resolution decomposition by the high frequency spatial subbands that this wavelet decomposition is produced, from described low resolution decomposition, produce a full resolution sequence.
(4) the described full resolution sequence and the motion vector that produce during the analysis of described motion compensation space-time are encoded, so that produce the coded bit stream of an output.
In described method, the global structure of the decomposition tree during 3DS analyzes is saved, and do not send the extraneous information that is used for the tuning drift effect (only changing the mechanism of decompositions/reconstruction), below to recall described method with reference to the more detailed mode of the encoding scheme of figure 2 and the reference motion-compensated time analysis when the lowest resolution shown in Figure 3.
Provide two key steps: (a) motion compensation step during lowest resolution; (b) coding step of high spatial subbands.At first, the drift during for fear of low resolution is used motion compensation (MC) in this level.Therefore, the GOF (21 among Fig. 2) during full resolution is at first reduced (represented by Fig. 3 Reference numeral d by this step; Corresponding to the step 22 among Fig. 2 and 23), and then as shown in Figure 3 common 3D subband MC decomposing scheme being applied to this has dwindled GOF rather than full-scale GOF, as described in the step 24 of Fig. 2.In Fig. 3, determine time subband (L according to known method for improving (at first limit H, limit L by A and H then) by A and B
0, d, H
0, d) and (L
1, d, H
1, d), dotted arrow is corresponding to high-pass time filtering, and solid arrow is corresponding to the low pass time filtering, and curve arrow (is positioned at low frequency space subband A (the Reference numeral A of the frame of this sequence
0, d, A
1, d, A
2, dAnd A
3, d) between, perhaps low frequency time subband L (Reference numeral L
0, dAnd L
1, d) between) corresponding to motion compensation (side effect can noticing this method is that the quantity of the motion vector that will send in bit stream descends, and this has saved some bits for texture coding).(for example shown in Figure 2 before the entropy coder that subband is transferred to based on tree structure, transfer to 3D-SPIHT encoder 27, described in the article " Low bit-ratescalable video coding with 3D set partitioning in hierarchical trees (3D-SPIHT) " of IEEE Transactions on Circuits and Systems forVideo Technology the 10th volume the 8th phase 1374-1387 page or leaf that is published on December 8th, 2000 people such as B.J.Kim), one is moved the high spatial subbands (26 among Fig. 2) that (step 25) allows the reconstruction of full resolution.Last tree structure looks and resembles very much for example at IEEEConference on Image Processing (ICIP2001), vol.2, pp.1017-1020, Thessaloniki, Greece, October 7-10, the tree structure of the 3D subband codec that the article " A fully scalable3D subband codec " on 2001 is put down in writing, and therefore, the entropy coder based on tree structure can be applied on it under the situation without any restriction.In the encoding scheme of Fig. 2, Reference numeral following (at a frame of full resolution sequence 21)
FRS: full resolution sequence 21
WD: wavelet decomposition 22
LRS: low-resolution sequence 23
MC-3DSA: motion compensation 3D Substrip analysis 24
LRD: low resolution decomposition (251)
HS: high subband 26
U-HFSS: the associating of three high frequency spatial subbands of a frame
FR-3D-SPIHT: full resolution 3D SPIHT 27
OCB: the coded bit stream of output
Corresponding decoding scheme is shown in Figure 4, this decoding scheme and this encoder symmetry (in Fig. 4, the Reference numeral that replenishes is as follows:
FR-3D-SPIHT: decoding step 41
MC-3DSS: motion compensation 3D Substrip analysis 43
HSS: high subband separates 44
FRR: the full resolution of full resolution sequence rebuilds 45).
In order to make spatial scalable, by cutting, the decoding scheme of Fig. 4 showed and obtains low-resolution sequence how naturally high frequency spatial subbands only need resemble in the scheme of common 3D subband codec.
Then, for high spatial subbands is encoded, proposed two major programmes, first does not have MC, and second has MC.
In the first string, high subband is only corresponding to the high frequency spatial subbands of original (full resolution) frame of GOF in the wavelet decomposition.These subbands allow to rebuild with full resolution in the decoding side.In fact, frame can be decoded when low resolution.Yet the low spatial subband during the small echo of these frames and primitive frame is synthetic is corresponding.Therefore, people only must put low-resolution frames and corresponding high subband together, and provide small echo synthetic, thereby obtain the frame of full resolution, and therefore optimize the 3D-SPIHT encoder.In the MC of 3D subband coder scheme, low time subband sees it is one of primitive frame that always resembles GOF.This be because:
Therefore, L looks and resembles A.Therefore, the high spatial subbands of A should be put together with the low resolution decomposition corresponding to L.This method (in the situation of forward direction MC, high spatial subbands reorder) is shown in Figure 5, jt express time decomposition level (full frame speed is 0, and lowest frame speed is jt-max) wherein, and nf is the subband index of time stage jt, DWT
HExpression high frequency wavelet filter, and coefficient c
JtBe multiplication constant, and OF, LRF, TS represent respectively: be labeled as 0-3 primitive frame, be labeled as the low-resolution frames of 00-03 and transmitted subband.
In second kind of scheme, can not get not having the reconstruction of drift because in each subband, use MC, still be possible so partly adopt MC to make up high spatial subbands, and can rebuild each resolution.Carry out wavelet decomposition according to the predicated error that the MC that carries out from the full resolution sequence obtains, and utilize for example low frequency resolution motion vector once more, rather than directly utilize the high frequency spatial subbands of wavelet decomposition.
Summary of the invention
Therefore, an object of the present invention is, simultaneously as far as possible near the performance of traditional 3D subband codec when full resolution by keeping its good performance to improve such scheme when the low resolution.
For this reason, the present invention relates to a kind of method for video coding that the original video sequence that is divided into successive frame group (GOF) is compressed of being used for, described method comprises the steps:
(1), from the full resolution frames of this original video sequence, produces a low resolution frame sequence that is organized among the continuous low resolution GOF by wavelet decomposition;
(2) GOF to each resolution in the described low resolution frame sequence carries out the analysis of a motion compensation space-time, produces a low-resolution sequence;
(3) each the full resolution GOF to this original video sequence carries out the analysis of a motion compensation space-time;
(4) in each time decomposition level, replace the low frequency sub-band of described decomposition with the time subband of corresponding low-resolution sequence;
(5) to thus obtained after changing sequence and the motion vector that during the motion compensation space-time analysis of each full resolution GOF, produces encode so that produce the output of coding bit rate.
The invention still further relates to the paired video encoding/decoding method of a kind of and above-mentioned method for video coding, and relate to corresponding video coding and decoding device.
Description of drawings
To be described in more detail the present invention in conjunction with the accompanying drawings now, wherein:
Fig. 1 shows a 3D sub-band division;
Fig. 2 has described the embodiment according to the encoding scheme of last embodiment;
Motion-compensated time analysis when Fig. 3 shows lowest resolution;
Fig. 4 has described a decoding scheme embodiment corresponding to encoding scheme shown in Figure 2;
Fig. 5 reorders to high spatial subbands and is illustrated (being used for forward motion compensation);
Fig. 6 is illustrated the key step of coding method of the present invention;
Fig. 7 A and 7B are illustrated corresponding motion compensated temporal filter decomposing scheme;
Fig. 8 A and 8B to the decoding side, be illustrated corresponding to the enforcement of the synthetic schemes of coding method shown in Figure 5.
Embodiment
With respect to aforesaid scheme, referring now to basic step of the present invention, the present invention will be described: (strictness of describing in this first step-motion compensation (MC)-in fact and the aforementioned schemes equates: at first utilize the wavelet based space filter to reduce the size of GOF in (a) motion compensation when lowest resolution, and the GOF after then size being reduced uses common 3D subband MC decomposing scheme), (b) high frequency spatial subbands is encoded.
With main different second steps that are of above-mentioned previous scheme: its principle is to go in each decomposition level is inserted into the time subband of low spatial resolution analysis the time subband of full resolution analysis.Therefore, in the time filtering of carrying out a reality, stressing to build primitive frame in decoding is possible (and for high frequency spatial subbands, not resembling the previous scheme only is extra coding and forecasted variances).
Under establish an equation this mechanism carried out more detailed description.As mentioned above, carry out the very first time with low resolution and analyze, it can be represented by equation (4) and (5):
Symbol wherein:
The A=reference frame
The B=present frame
The DWT=wavelet transform
A
dThe low frequency space subband of the DWT of=frame A, i.e. frame A low spatial resolution pattern
B
dThe low frequency space subband of the DWT of=frame B, i.e. frame B low spatial resolution pattern
High frequency time subband during the H=low spatial resolution
Low frequency time subband during the L=low spatial resolution
MC
Down=the motion compensation on low resolution (being sub-sampling) frame, carried out
MC
-1=contrary motion compensation (being used for of being calculated dopes frame B from frame A motion vector is used on the contrary from frame B predictive frame A)
Equation (6)-(9) can be used to limit L
sAnd H
s:
H’＝B-MC
full(A) (6)
H
s＝H’ (8)
Wherein:
X
sThe combination of three high frequency spatial subbands of the DWT of=given frame X (X wherein
s=H
sOr L
s)
MC
Full=the motion compensation on full resolution frames, carried out
L ' and H '=be respectively low frequency and high frequency time subband in traditional 3D subband scheme
H＝DWT
-1[H
d∪H
s]
L＝DWT
-1[L
d∪L
s]
In case all low frequencies and great number time subband all produce on given time stage jt, when low and full spatial resolution, low frequency time subband L is further decomposed to obtain next time stage jt+1 so.
Each step of decomposing in the time all repeats this operation, produces a time decomposition texture at last, and the time decomposition texture of itself and traditional 3D subband coder is very similar.Then, the low frequency time subband of afterbody and the high frequency time subband of all grades are carried out spatial decomposition, and it is encoded to form bit stream by wavelet filter.
Above-mentioned the present invention makes aforementioned schemes keep good performance when low resolution, and when full resolution the performance of more approaching traditional 3D subband codec (overall structure of decomposition tree is saved and does not send extraneous information and comes tuning drift influence in the 3D Substrip analysis; Only change decomposition/reconstruction mechanism).Main raising is owing to the new method that produces high frequency spatial subbands, and it is more coherence for decomposition tree brings, and therefore improves the code efficiency of system.
At decoder, aforesaid all equations can be reversed, to obtain good reconstruction.Only a^ is added in each subband and goes, so as the expression relate to now be the decoding and some information may lose.At first, the traditional 3D Substrip analysis during low resolution allows from L
dAnd H
dThe middle low spatial resolution subband A that recovers
dAnd B
d:
Can also easily obtain A by synthetic H with by reversing equation (7)
sThis process is illustrated by equation (12) to (15):
(15)
Then, according to
With
Rebuild simply
Thereby can obtain B
sAnd finally synthetic B.This can be summarized by serial equation (16)-(19):
(16)
(17)
Repeat these operations, up to very first time level, promptly up to the GOF complete decoding.Can know and find out, this scheme is because as long as L and H end of transmission in bit stream just can obtain to rebuild fully, thereby do not produce drift and (it can also be seen that on each time stage, full spatial resolution is synthetic closely to link to each other with the synthetic of low resolution, does not have this situation in described previous scheme).
Now, with reference to Fig. 6 and Fig. 7 above-mentioned coding principle is described in detail, Fig. 6 shows the key step of coding method, and Fig. 7 (in fact comprising Fig. 7 A and 7B) shows corresponding motion compensated temporal filter scheme in more detailed mode.
In encoding scheme shown in Figure 6, at first,, utilize primitive frame group GOF (this current GOF comprises full resolution frames FRF) to produce low-resolution frames LRF by wavelet decomposition WD, then low-resolution frames is carried out the motion compensation space-time and analyze MCSTA.Thereby obtain a low-resolution sequence.Original full resolution frames (being each full resolution GOF) also is used to carry out motion compensation space-time analysis (corresponding consecutive steps MCSTA and WD correspondence: " MC time series analysis " and " wavelet decomposition "), so that produce high spatial subbands HSS.
Carry out after these two groups of parallel steps on full resolution frames, obtain the low frequency sub-band of this decomposition thus, these low frequency sub-bands are replaced according to following operation by the corresponding space-time subband of low-resolution sequence LRS repeatedly in each time decomposition level:
(a) at first, a storage operation 62 is used to last coding step 69 to store the high frequency spatial-time subband of these decomposition;
(b) then, a small echo synthesizes 63, and its low frequency space from described decomposition-time subband is carried out (test 61 " L or H time subbands " allow to separate described low frequency and high frequency spatial-time subband);
(c) then, test 64 about this time decomposition level, low frequency space-time the subband that is used for storage (65) this decomposition when described level is last is not carried out two groups of opposite parallel steps at next time stage (66) when described level is not last.
The more detailed description of whole decomposing scheme (in the coding side) and corresponding motion compensation synthetic schemes (in the decoding side) can be seen Fig. 7 and 8 (also comprise two figure: Fig. 8 A and Fig. 8 B) respectively.The example that this space-time of the present invention decomposes relates to a GOF (for the sake of simplicity) who has only four frame A0-A3, has forward motion compensation and two decomposition level.Being called as lifting scheme by utilization (is respectively H from primitive frame calculating height and low frequency
0', H
1' with L
0', L
1') the time subband, be published in Bell Laboratoriestechnical report (Bell Laboratory technical report) in 1996 at I.Daubechies and W.Seldens, the existing description in the Lucent Techologies article " Factoring wavelet transforms into lifting steps ".Symbol DWT and DWT
-1Represent that respectively wavelet decomposition and small echo are synthetic.The right side of Fig. 7 shows the first space-time decomposition level continuously, the low frequency space-time subband and the second space-time decomposition level to this decomposition are used anti-synthetic (replaced the back to carry out at the low frequency sub-band that decomposes by the corresponding space-time subband of low-resolution sequence, described replacement is illustrated by the arrow from Fig. 7 left side).
Method for video coding of the present invention and equipment are had been described in detail hereinbefore, but clearly the invention still further relates to the corresponding video decoding method, this coding/decoding method comprises the paired consecutive steps of carrying out when realizing described method for video coding of step, and relating to corresponding decoding device, this decoding device comprises the paired continuous apparatus of device that has with described video encoder.
Claims (4)
1. one kind is used for compressing a method for video coding that is divided into the original video sequence of successive frame group (GOF), and described method comprises the steps:
(1), from the full resolution frames of this original video sequence, produces a low resolution frame sequence that is organized among the continuous low resolution GOF by wavelet decomposition;
(2) each the low resolution GOF to described low resolution frame sequence carries out the analysis of a motion compensation space-time, thereby produces a low-resolution sequence;
(3) each the full resolution GOF to this original video sequence carries out the analysis of a motion compensation space-time;
(4), replace the low frequency sub-band of described decomposition with the corresponding space-time subband of this low-resolution sequence in each time decomposition level;
(5) sequence after changing that is obtained and the motion vector that produces during the motion compensation space-time analysis of each full resolution GOF are encoded, so that produce the coded bit stream of an output.
2. one kind is used for compressing a video encoder that is divided into the original video sequence of successive frame group (GOF), and described equipment comprises:
(1), from the full resolution frames of this original video sequence, produces a device that is organized in the low resolution frame sequence among the continuous low resolution GOF by wavelet decomposition;
(2) each the low resolution GOF to described low resolution frame sequence carries out the analysis of a motion compensation space-time, thereby produces the device of a low-resolution sequence;
(3) each the full resolution GOF to this original video sequence carries out the device that a motion compensation space-time is analyzed;
(4), replace the device of the low frequency sub-band of described decomposition with the corresponding space-time subband of this low-resolution sequence in each time decomposition level;
(5) sequence after changing that is obtained and the motion vector that produces during the motion compensation space-time analysis of each full resolution GOF are encoded, so that produce the device of the coded bit stream of an output.
3. one kind is used for video encoding/decoding method that a coded bit stream is decoded, wherein this coded bit stream correspondence a video sequence of encoding with following method for video coding, and in order to compress described original video sequence, this method for video coding comprises the steps:
(1), from the full resolution frames of this original video sequence, produces a low resolution frame sequence that is organized among the continuous low resolution GOF by wavelet decomposition;
(2) each the low resolution GOF to described low resolution frame sequence carries out the analysis of a motion compensation space-time, thereby produces a low-resolution sequence;
(3) each the full resolution GOF to this original video sequence carries out the analysis of a motion compensation space-time;
(4), replace the low frequency sub-band of described decomposition with the corresponding space-time subband of this low-resolution sequence in each time decomposition level;
(5) sequence after changing that is obtained and the motion vector that produces during the motion compensation space-time analysis of each full resolution GOF are encoded, so that produce the coded bit stream of an output,
Described video encoding/decoding method comprises continuous step, and the performed step of the described method for video coding of these steps and claim 1 is paired.
4. one kind is used for video decoding apparatus that a coded bit stream is decoded, and wherein this coded bit stream correspondence one by the video sequence of following video encoder coding, and in order to compress described original video sequence, this video encoder comprises:
(1), from the full resolution frames of this original video sequence, produces a device that is organized in the low resolution frame sequence among the continuous low resolution GOF by wavelet decomposition;
(2) each the low resolution GOF to described low resolution frame sequence carries out the analysis of a motion compensation space-time, thereby produces the device of a low-resolution sequence;
(3) each the full resolution GOF to this original video sequence carries out the device that a motion compensation space-time is analyzed;
(4), replace the device of the low frequency sub-band of described decomposition with the corresponding space-time subband of this low-resolution sequence in each time decomposition level;
(5) sequence after changing that is obtained and the motion vector that produces during the motion compensation space-time analysis of each full resolution GOF are encoded, so that produce the device of the coded bit stream of an output,
Described video decoding apparatus comprises continuous device, and the device that the described video coding of these devices and claim 2 has is paired.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| EP02292551 | 2002-10-16 | ||
| EP02292551.5 | 2002-10-16 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN1706198A true CN1706198A (en) | 2005-12-07 |
Family
ID=32104016
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNA2003801015409A Pending CN1706198A (en) | 2002-10-16 | 2003-10-01 | Drift-free video encoding and decoding method, and corresponding devices |
Country Status (6)
| Country | Link |
|---|---|
| EP (1) | EP1554886A1 (en) |
| JP (1) | JP2006503475A (en) |
| KR (1) | KR20050057655A (en) |
| CN (1) | CN1706198A (en) |
| AU (1) | AU2003263547A1 (en) |
| WO (1) | WO2004036918A1 (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9448437B2 (en) | 2011-06-27 | 2016-09-20 | Samsung Display Co., Ltd. | Display apparatus for controlling optical transmissivity |
-
2003
- 2003-10-01 EP EP03808817A patent/EP1554886A1/en not_active Withdrawn
- 2003-10-01 JP JP2004544560A patent/JP2006503475A/en active Pending
- 2003-10-01 CN CNA2003801015409A patent/CN1706198A/en active Pending
- 2003-10-01 AU AU2003263547A patent/AU2003263547A1/en not_active Abandoned
- 2003-10-01 KR KR1020057006443A patent/KR20050057655A/en not_active Application Discontinuation
- 2003-10-01 WO PCT/IB2003/004326 patent/WO2004036918A1/en not_active Application Discontinuation
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9448437B2 (en) | 2011-06-27 | 2016-09-20 | Samsung Display Co., Ltd. | Display apparatus for controlling optical transmissivity |
Also Published As
| Publication number | Publication date |
|---|---|
| EP1554886A1 (en) | 2005-07-20 |
| WO2004036918A1 (en) | 2004-04-29 |
| AU2003263547A1 (en) | 2004-05-04 |
| JP2006503475A (en) | 2006-01-26 |
| KR20050057655A (en) | 2005-06-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN1251509C (en) | Method of encoding sequence of frames | |
| CN1266649C (en) | Video coding method | |
| KR101203369B1 (en) | A system and method for image coding employing hybrid directional prediction and lifting wavelet | |
| CN1161708C (en) | Apparatus and method for encoding wavelet trees generated by wavelet-based coding method | |
| CN1722838A (en) | Use the scalable video coding method and apparatus of basal layer | |
| CN1722831A (en) | To comprising basic layer the bit stream pre decoding and the method and apparatus of decoding | |
| CN1383684A (en) | Color video encoding and decoding method | |
| US6965700B2 (en) | Embedded and efficient low-complexity hierarchical image coder and corresponding methods therefor | |
| CN108293138A (en) | Video/image coding in the effective and scalable frame encoded using small echo and AVC, AVC, VPx of modification, the VPx of modification or the HEVC of modification | |
| CN1764280A (en) | Method and apparatus based on multilayer effective compressing motion vector in video encoder | |
| CN1976458A (en) | Method of encoding flags in layer using inter-layer correlation, method and apparatus for decoding | |
| CN1943244A (en) | Inter-frame prediction method in video coding, video encoder, video decoding method, and video decoder | |
| CN1431828A (en) | Optimum scanning method for change coefficient in coding/decoding image and video | |
| CN1322442A (en) | Encoding method for compression of video sequence | |
| CN1620815A (en) | Drift-free video encoding and decoding method, and corresponding devices | |
| CN1678073A (en) | Direction-adaptive scalable motion parameter coding for scalable video coding | |
| CN1234249C (en) | Embedded image processor | |
| CN1212806A (en) | Progressive image coding | |
| CN1665299A (en) | Method for designing architecture of scalable video coder decoder | |
| CN1237817C (en) | Encoding method for the compression of a video sequence | |
| CN1885945A (en) | Hierarchical coding and decoding method | |
| CN1914926A (en) | Moving picture encoding method and device, and moving picture decoding method and device | |
| CN101044764A (en) | Method for hierarchically coding video images | |
| CN1809169A (en) | Method and apparatus for encoding pictures without loss of DC components | |
| CN1310190C (en) | Discrete wavelet transform unit and method for adaptively encoding still image based on energy of each block |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |