CN1484452A - enhanced type fineness extensible video coding structure - Google Patents
enhanced type fineness extensible video coding structure Download PDFInfo
- Publication number
- CN1484452A CN1484452A CNA031344747A CN03134474A CN1484452A CN 1484452 A CN1484452 A CN 1484452A CN A031344747 A CNA031344747 A CN A031344747A CN 03134474 A CN03134474 A CN 03134474A CN 1484452 A CN1484452 A CN 1484452A
- Authority
- CN
- China
- Prior art keywords
- frame
- intermediate layer
- layer
- coding
- carry out
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Landscapes
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
This invention relates to an enhanced fine grain flexible video code structure composed of streams of a basic, an interlayer and an enhanced layer, among which, the design of the basic layer is the same with that of the existing second layer of stream, the interlayer has two ways: the oven frame of frame p code applies the last basic layer for motion compensation to generate frame memory F1 then to form a stream by F1 motion compensation coding, the odd frame of frame p code applies the last interlayer to generate frame memory FO then to form a stream by FO motion, the interlayer, basic layer and enhanced layer of frame B use F1 to make motion compensation, frame p and the enhanced layer of frame B are generated utilizing F1 motion compensation.
Description
Technical field:
The present invention relates to the coding techniques field, a kind of specifically enhancement mode fine granularity scalable video coding structure is used for the international standard of multi-medium data compressed encoding.
Background technology
Along with development of internet technology, different international standards is all stipulated in the various transmission on the relevant the Internet.For example, MPEG-4 just is based on the international standard of multi-medium data compressed encoding, and this standard is an open standard, and many high performance new technologies and new algorithm can be constantly included.Fine granularity scalable video coding FGS is a very important core technology in the MPEG-4 international standard, and it is mainly used in problems such as solving the network bandwidth fluctuation that runs into when video flowing transmits on the internet and data-bag lost.
The fine granularity scalable video coding FGS that adopts at present in the MPEG-4 international standard is based on the Bit-Plane Encoding scheme of discrete cosine transform DCT coefficient, it is in the coded system that all is better than other aspect algorithm complex and the coding efficiency two, for example is better than just being better than based on the fine granularity scalable video coding FGS of small echo other coded system.This scheme is the basic scheme BFGS of fine granularity scalable video coding, as shown in Figure 4.The structure of this basic scheme is made of basic layer and two kinds of code streams of enhancement layer.Wherein, basic layer bit stream is to transmit, and code check is lower, and enhancement layer bitstream can block arbitrarily according to bandwidth situation.Though the video flowing that this basic scheme generates can adapt to complicated network bandwidth fluctuation, has robustness preferably, but because the motion compensation in encoding scheme is the reconstruction of layer with reference to a minimum quality, thereby there is the too low deadly defect of code efficiency.Under equal code check, quality is than the low 2dB～3dB of the non-scalable coding among the MPEG-4 (3dB means that code check doubles), and this is beyond affordability in actual applications.Need to study new high performance fine granularity scalable video coding scheme for this reason.In present main research, the most representative have following two kinds of schemes:
(1) motion-compensation fine scalable coding scheme MC-FGS, this scheme proposes (M.vander Schaar by Philip, H.Radha, " Motion compensation based fine-granular scalability forwireless multimedia ", in Proc.IEEE Workshop Multimedia Signal Processing, vol.10, Oct.2001, pp.453-458.).On the basic scheme basis of this scheme in the MPEG-4 standard, two kinds of Collapsible structures have been proposed, be called two shuttling movement compensation fine granularity scalable coding two-loopMC-FGS and single cycle motion-compensation fine scalable coding one-loop MC-FGS, these two kinds of structures are applied to the situation of high bit-error and low implementation complexity respectively;
(2) progressive meticulous scalable video scheme PFGS, this scheme proposes (F.Wu by Microsoft Research, S.Li and Y.-Q.Zhang " A framework for efficient progressive finegranularity scalable video coding.; " IEEE Trans.Circuit and systems for videotechnology, vol.11, Mar.2001, pp.332-344.).The thought of this encoding scheme is alternately to use low-quality reference and high-quality reference to prevent the propagation and the accumulation of error, and the while is improved the code efficiency of enhancement layer to a certain extent.
Though above-mentioned two kinds of schemes (MC-FGS and PFGS) are compared on code efficiency with basic scheme BFGS and are greatly increased, and still have many deficiencies.For example, two shuttling movement compensation fine granularity scalable coding two-loop MC-FGS have seriously limited its code efficiency because an enhancement layer to the B frame carries out high-quality motion compensation; And single cycle motion-compensation fine scalable coding one-loopMC-FGS has been owing to adopted high-quality motion compensation to basic layer, thus be subject to the influence of channel width fluctuation and channel error code, and the generation accumulation of error.The subject matter that progressive meticulous scalable video scheme PFGS exists is: the first, use low-quality basic layer as with reference to the time image code efficiency can be low, thereby reduced the code efficiency of whole system.The second, alternately use low-quality reference and high-quality reference, can make the consecutive frame decoding quality differ bigger, and human eye is very responsive to the fluctuation of consecutive frame picture quality.Because the existence of these problems makes motion-compensation fine scalable coding scheme MC-FGS and progressive meticulous these two kinds of schemes of scalable video scheme PFGS all be far from reaching the standard optimal compression performance towards the compression method of transmission.
Summary of the invention::
The objective of the invention is to overcome the deficiency of above-mentioned prior art, propose a kind of code efficiency height, network bandwidth adaptive changes and eliminates the enhancement mode fine granularity scalable video coding structure of the accumulation of error.
The key problem in technology of realizing the object of the invention is to set up the intermediate layer between the basic layer of existing fine granularity scalable video coding device and enhancement layer, constitutes the three-decker of every frame for basic layer, intermediate layer and enhancement layer, the formation code stream of encoding.This intermediate layer is provided with according to different frame codings, the even frame intermediate layer of P frame coding adopts the basic layer of former frame to carry out motion compensation delta frame memory F1 earlier, carry out motion compensation encoding by frame memory F1 again, form even frame intermediate layer code stream, the odd-numbered frame intermediate layer of P frame coding adopts the intermediate layer of former frame to carry out motion compensation delta frame memory F0 earlier, carry out motion compensation encoding by frame memory F0 again, form odd-numbered frame intermediate layer code stream, promptly, form the intermediate layer code stream of present frame by alternately using frame memory F0 and frame memory F1 to carry out the intermediate layer motion compensation encoding.The intermediate layer of B frame coding mode is identical with enhancement layer with basic layer, and the frame memory F1 that two frames generate before and after promptly using carries out motion compensation, has so both improved code efficiency, can not cause the propagation and the accumulation problem of error again because of the fluctuation of bandwidth.
This enhancement layer is to utilize frame memory F1 to carry out motion compensation and generate, and the P coded system is identical with the B frame coding mode.
The basic layer information of this structure all can obtain, if the intermediate layer information of the 1st frame makes a mistake owing to reasons such as the fluctuation of channel width or channel error codes or loses, then the enhancement layer of the 2nd frame will be subjected to the influence of error, cause the reduction of the 2nd frame quality.But because the intermediate layer of the 2nd frame is to use the basic layer of the 1st frame to carry out motion compensation, so the intermediate layer of the 2nd frame can correctly rebuild in decoding end, the image of such the 3rd frame just can not be subjected to the influence in the intermediate layer of the 1st frame.
Be provided with two frame memory F0 and F1 in the above-mentioned P frame coding structure, wherein, frame memory F0 is used to store the basic layer of reconstruction of former frame, with the basic layer coding to subsequent frame; Frame memory F1 is used to store the reconstruction intermediate layer of former frame, with intermediate layer and the enhancement layer coding to subsequent frame.
Intermediate layer in the above-mentioned P frame coding structure is provided with switch S 1, S2 and S3, be used to control the coded system in intermediate layer, three switches are divided into 1 and 2 two position, position 1 is connected with the intermediate layer, position 2 is connected with basic layer, when three switch S 1, S2 and S3 push position 2, is the intermediate layer coding of even frame, when three switch S 1, S2 and S3 push position 1, be the intermediate layer coding of odd-numbered frame.
Above-mentioned P frame coding structure, wherein utilize frame memory F1 carry out motion compensation encoding be enhanced the layer code stream, be by after the residual error of the high-quality moving compensating data HQCD in original video data Vi and intermediate layer is carried out discrete cosine transform DCT, deduct basic layer and intermediate layer coded data, again this residual error ED is carried out Bit-Plane Encoding, form enhancement layer bitstream.
Above-mentioned P frame coding structure, wherein utilize frame memory F1 to carry out the code stream that motion compensation encoding obtains the intermediate layer, be by after original video Vi is carried out discrete cosine transform DCT with the residual error of basic layer coded data BD, carry out Bit-Plane Encoding, preceding m bit plane of coding forms the intermediate layer code stream of even frame; Preceding m bit plane of enhancement layer forms the intermediate layer code stream of odd-numbered frame.
Above-mentioned P frame coding structure, the vision signal in its intermediate layer is obtained, be after the bit plane decoding, to sue for peace with the dequantized coefficients of basic layer by the intermediate layer code stream, carry out discrete cosine DCT inverse transformation again, the data that obtain are sued for peace with basic layer video data BD, obtain by the amplitude limit operation at last.
The present invention has solved error propagation and accumulation problem effectively owing to adopted the intermediate layer code flow structure; Because the coding structure that enhancement layer of the present invention adopts the intermediate layer to carry out motion compensation has improved the precision of motion compensation, the code efficiency of enhancement layer is improved significantly simultaneously.
Description of drawings:
Fig. 1 is a principle assumption diagram of the present invention
Fig. 2 is a P frame coding instance graph of the present invention
Fig. 3 is a P frame decoding instance graph of the present invention
Fig. 4 is the principle assumption diagram of prior art
Embodiment:
With reference to Fig. 1, each frame of the present invention is made of basic layer, intermediate layer and three layers of coding structure of enhancement layer.Wherein:
In the P frame coding structure shown in Fig. 1 a, F0 is the frame memory that is used for basic layer prediction, and F1 is the frame memory that is used for intermediate layer and enhancement layer prediction.M is the intermediate layer information of using when being used to rebuild F1, motion compensation information derived from the intermediate layer of former frame when the solid line hollow arrow was represented to rebuild F1, motion compensation information derived from the basic layer of former frame when the dotted line hollow arrow was represented to rebuild F1, and the solid line filled arrows is represented the flow direction of data.The generation and the existing basic scheme BFGS of basic layer are in full accord, that is: by after original video Vi is carried out discrete cosine transform DCT with the residual error of basic layer moving compensating data BMCD, quantize again, form basic layer bit stream by variable-length encoding at last; The generating mode in intermediate layer has two kinds: carry out motion compensation for the first basic layer by former frame of even frame and obtain frame memory F1, carry out motion compensation encoding by frame memory F1 again, form even frame intermediate layer code stream, i.e. the 2nd frame and the 4th frame code stream, carry out motion compensation and obtain frame memory F0 by the intermediate layer of former frame for odd-numbered frame, carry out motion compensation encoding by frame memory F0 again, form odd-numbered frame intermediate layer code stream, i.e. the 3rd frame and the 5th frame code stream.The generation of enhancement layer is to utilize frame memory F1 to carry out motion compensation and generate.
In the B frame coding structure shown in Fig. 1 b, because the B frame is a bi-directional predicted frames, not as the prediction reference of other frame, so when in the present invention the B frame being encoded, the frame memory F1 that uses front and back two frames to generate to each coding layer carries out motion compensation, so promptly improved code efficiency, can not cause the propagation and the accumulation problem of error again because of the fluctuation of bandwidth.
With reference to Fig. 2, the present invention is provided with two frame memories in the P frame coding example according to Fig. 1 coding structure principles of construction, and wherein frame memory F0 is used for storing the basic layer of reconstruction of former frame, is used for the basic layer coding of subsequent frame; Frame memory F1 is used for storing the reconstruction intermediate layer of former frame, and it is used for the intermediate layer and the enhancement layer coding of subsequent frame.Switch S 1, S2 and S3, be used to control the coded system in intermediate layer, three switches are divided into 1 and 2 two position, position 1 is connected with the intermediate layer, position 2 is connected with basic layer, when encode in the intermediate layer of carrying out even frame, three switch S 1, S2 and S3 are pushed position 2, three switch S 1, S2 and S3 push position 1 when encode in the intermediate layer of carrying out odd-numbered frame.Bit-Plane Encoding mode in this encoder is identical with coded system among the existing fine granularity scalable video coding basic scheme BFGS, all is the variable-length encoding of adopting based on bit plane.When the video data of input is sent to when encoding in basic layer, intermediate layer and three coding layers of enhancement layer, the coded system of basic layer with H.263 wait the coded system of standard identical, Q limits basic layer bit rate by the control quantization step, in enhancement layer coding, at first the residual error of the high-quality moving compensating data HQCD in original video data Vi and intermediate layer is carried out the discrete cosine dct transform, deduct basic layer and intermediate layer coded data then, at last this residual error ED is carried out Bit-Plane Encoding.When encode in the intermediate layer of dual numbers frame, at first original video is looked data Vi and carry out the discrete cosine dct transform with the residual error of basic layer coded data BD, carry out Bit-Plane Encoding then, preceding m bit plane of coding forms the intermediate layer code stream of even frame; Preceding m bit plane of enhancement layer forms the intermediate layer code stream of odd-numbered frame.
With reference to Fig. 3, the present invention also is provided with two frame memories in the P frame decoding example according to Fig. 1 coding structure principles of construction, and this frame memory F0 and frame memory F1 are used for storing the basic layer and the intermediate layer of former frame respectively.Switch S is used for selecting frame memory, to form the enhancement layer reference of next frame, promptly determines the generating mode in intermediate layer.The decoding end of this decoder is imported basic layer bit stream, intermediate layer code stream and enhancement layer bitstream, and wherein, basic layer bit stream is decoded by standard H.263, produces the low-quality decoded video of basic layer.The intermediate layer code stream is sued for peace with the dequantized coefficients of basic layer then at first through the bit plane decoding, carries out discrete cosine DCT inverse transformation again, after at last the data that obtain being sued for peace with basic layer video data BD, carries out the video that the amplitude limit operation obtains the intermediate layer again.This intermediate layer video also is stored among the frame storage F1 motion compensation reference as the next frame enhancement layer.Enhancement layer bitstream at first carries out the bit plane decoding, data LD with low layer sues for peace then, and summed data carried out discrete cosine DCT inverse transformation, after at last the data that obtain after the DCT inverse transformation and high-quality moving compensating data HQCD being sued for peace, carry out the be enhanced video of layer of amplitude limit operation again.
Claims (6)
1. an enhancement mode fine granularity scalable coding and decoding video structure comprises by basic layer and enhancement layer, it is characterized in that:
(1) set up the intermediate layer between basic layer and enhancement layer, constitute every frame and be basic layer, intermediate layer and enhancement layer three-decker, the coding back forms three layer bit streams;
(2) intermediate layer is according to different frame encoding settings, the even frame intermediate layer of P frame coding adopts the basic layer of former frame to carry out motion compensation delta frame memory (F1) earlier, carry out motion compensation encoding by frame memory (F1) again, form even frame intermediate layer code stream, the odd-numbered frame intermediate layer of P frame coding adopts the intermediate layer of former frame to carry out motion compensation delta frame memory (F0) earlier, carry out motion compensation encoding by frame memory (F0) again, form odd-numbered frame intermediate layer code stream, promptly, form the intermediate layer code stream of present frame by alternately using frame memory (F0) and frame memory (F1) to carry out the intermediate layer motion compensation encoding;
(3) enhancement layer bitstream is to utilize frame memory (F1) to carry out motion compensation and generate.
2. coding structure according to claim 1 is characterized in that being provided with frame memory (F0) and frame memory (F1) in the P frame coding, and frame memory (F0) is used to store the basic layer of reconstruction of former frame, with the basic layer coding to subsequent frame; Frame memory (F1) is used to store the reconstruction intermediate layer of former frame, encodes with intermediate layer and enhancement layer to subsequent frame.
3. coding structure according to claim 1, it is characterized in that the intermediate layer of P frame coding is provided with three switches (S1) of control intermediate layer coded system, (S2) and (S3), three switches are divided into two positions (1) and (2), position (1) is connected with the intermediate layer, position (2) is connected with basic layer, three switches (S1), (S2) and (S3) when being connected to position (2) simultaneously, intermediate layer coding for even frame, three switches (S1), (S2) and (S3) when being connected to position (1) simultaneously are the intermediate layer coding of odd-numbered frame.
4. coding and decoding video structure according to claim 1, it is characterized in that utilizing frame memory (F1) to carry out the motion compensation layer bit stream that is enhanced, be by after the original video data (Vi) and the residual error of the moving compensating data (HQCD) in intermediate layer are carried out discrete cosine transform (DCT) conversion, deduct basic layer and intermediate layer coded data, again this residual error (ED) is carried out Bit-Plane Encoding and obtain.
5. coding and decoding video structure according to claim 1, it is characterized in that utilizing frame memory (F1) to carry out motion compensation and obtain the intermediate layer code stream, be by after the original video (Vi) and the residual error of basic layer coded data (BD) are carried out discrete cosine transform (DCT), carry out Bit-Plane Encoding, preceding m bit plane of coding forms the intermediate layer code stream of even frame; Preceding m bit plane of enhancement layer forms the intermediate layer code stream of odd-numbered frame.
6. coding structure according to claim 1, the vision signal that it is characterized in that the intermediate layer is obtained, be with the intermediate layer code stream after bit plane decoding, dequantized coefficients summation with basic layer, carry out discrete cosine (DCT) inverse transformation again, after the data that obtain and basic layer video data (BD) are sued for peace, carry out the amplitude limit operation again and obtain.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 03134474 CN1202673C (en) | 2003-08-07 | 2003-08-07 | Enhanced type fineness extensible video coding structure |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 03134474 CN1202673C (en) | 2003-08-07 | 2003-08-07 | Enhanced type fineness extensible video coding structure |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1484452A true CN1484452A (en) | 2004-03-24 |
| CN1202673C CN1202673C (en) | 2005-05-18 |
Family
ID=34154500
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 03134474 Expired - Fee Related CN1202673C (en) | 2003-08-07 | 2003-08-07 | Enhanced type fineness extensible video coding structure |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN1202673C (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101931812B (en) * | 2004-10-15 | 2011-12-21 | 弗劳恩霍夫应用研究促进协会 | Device and method for forecasting, generating and decoding video sequence by intermediate layer residue value |
| CN101411197B (en) * | 2006-03-27 | 2012-09-05 | 高通股份有限公司 | Methods and systems for refinement coefficient coding in video compression |
| CN105657541A (en) * | 2015-12-29 | 2016-06-08 | 华为技术有限公司 | Frame processing method and device |
-
2003
- 2003-08-07 CN CN 03134474 patent/CN1202673C/en not_active Expired - Fee Related
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101931812B (en) * | 2004-10-15 | 2011-12-21 | 弗劳恩霍夫应用研究促进协会 | Device and method for forecasting, generating and decoding video sequence by intermediate layer residue value |
| CN101959066B (en) * | 2004-10-15 | 2012-10-10 | 弗劳恩霍夫应用研究促进协会 | Device and method for generating a coded video sequence and for decoding a coded video sequence while using an inter-layer residual value prediction |
| CN101411197B (en) * | 2006-03-27 | 2012-09-05 | 高通股份有限公司 | Methods and systems for refinement coefficient coding in video compression |
| TWI393446B (en) * | 2006-03-27 | 2013-04-11 | Qualcomm Inc | Methods and systems for refinement coefficient coding in video compression |
| CN105657541A (en) * | 2015-12-29 | 2016-06-08 | 华为技术有限公司 | Frame processing method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN1202673C (en) | 2005-05-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| KR100781525B1 (en) | Method and apparatus for encoding and decoding FGS layers using weighting factor | |
| KR100703724B1 (en) | Apparatus and method for adjusting bit-rate of scalable bit-stream coded on multi-layer base | |
| JP5014989B2 (en) | Frame compression method, video coding method, frame restoration method, video decoding method, video encoder, video decoder, and recording medium using base layer | |
| KR100791299B1 (en) | Multi-layer based video encoding method and apparatus thereof | |
| US20060104354A1 (en) | Multi-layered intra-prediction method and video coding method and apparatus using the same | |
| CN100512446C (en) | A multi-description video encoding and decoding method based on self-adapted time domain sub-sampling | |
| JP2007520149A (en) | Scalable video coding apparatus and method for providing scalability from an encoder unit | |
| CN1611077A (en) | Spatial scalable compression | |
| CN1633812A (en) | Global motion compensation for video pictures | |
| CN101288308A (en) | Intra-base-layer prediction method satisfying single loop decoding condition, and video coding method and apparatus using the prediction method | |
| JP2005531258A (en) | Scalable and robust video compression | |
| CN101180882A (en) | Method and apparatus for encoding video data, and method and apparatus for decoding video data | |
| KR100654431B1 (en) | Method for scalable video coding with variable GOP size, and scalable video coding encoder for the same | |
| CN1575606A (en) | Spatial scalable compression | |
| CN101002476B (en) | Coding and decoding method and coding and decoding device for video coding | |
| Wu et al. | DCT-prediction based progressive fine granularity scalable coding | |
| KR100382516B1 (en) | Object-based image signal coding/decoding apparatus and method | |
| CN1672421A (en) | Method and apparatus for performing multiple description motion compensation using hybrid predictive codes | |
| EP1841235A1 (en) | Video compression by adaptive 2D transformation in spatial and temporal direction | |
| Yan et al. | A survey on the techniques for the transport of MPEG-4 video over wireless networks | |
| CN1202673C (en) | Enhanced type fineness extensible video coding structure | |
| CN1656816A (en) | Improved efficiency fgst framework employing higher quality reference frames | |
| Huchet et al. | Distributed video coding without channel codes | |
| KR20060063619A (en) | Method for encoding and decoding video signal | |
| CN100499815C (en) | Video frequency coding and de-coding method for supporting video frequency frame random reading |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| C17 | Cessation of patent right | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20050518 Termination date: 20100807 |