CN100508608C - Non-predicted circulation anti-code error video frequency coding method - Google Patents

Non-predicted circulation anti-code error video frequency coding method Download PDF

Info

Publication number
CN100508608C
CN100508608C CN 200610012092 CN200610012092A CN100508608C CN 100508608 C CN100508608 C CN 100508608C CN 200610012092 CN200610012092 CN 200610012092 CN 200610012092 A CN200610012092 A CN 200610012092A CN 100508608 C CN100508608 C CN 100508608C
Authority
CN
China
Prior art keywords
frame
image
coding
quantization
obtains
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.)
Expired - Fee Related
Application number
CN 200610012092
Other languages
Chinese (zh)
Other versions
CN1889687A (en
Inventor
丁贵广
戴琼海
于志涛
杨峰
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tsinghua University
Original Assignee
Tsinghua University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Tsinghua University filed Critical Tsinghua University
Priority to CN 200610012092 priority Critical patent/CN100508608C/en
Publication of CN1889687A publication Critical patent/CN1889687A/en
Application granted granted Critical
Publication of CN100508608C publication Critical patent/CN100508608C/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Landscapes

  • Compression Or Coding Systems Of Tv Signals (AREA)

Abstract

A video coding method includes carrying out in frame coding on the first frame of image to obtain movement vector by making movement estimation on the first frame of coded image, obtaining bit stream of movement vector by using variable length coder to make entropy coding, obtaining calibration bit stream by making channel coding after step size quantization and obtaining residual error image by utilizing current frame of image and antiquantization result to derive difference, utilizing transform quantization coefficient to obtain residual error bit stream, forming code stream by said bit stream and movement vector and forming output code stream by said code stream and calibration bit stream.

Description

A kind of anti-error code video coding-decoding method that does not have prediction loop
Technical field
The invention belongs to technical field of computer multimedia, particularly the video encoding design technology.
Background technology
Along with the develop rapidly of Internet and wireless communication technology, people are the real-time acquiring multi-media data on network, and particularly informative video data has become possibility.Yet because the existence of the factors such as error code of the fluctuation of the isomerism of network, channel width and channel makes to be difficult to satisfy the requirement that real-time video is communicated by letter towards the compression algorithm of storing originally.So in current cybertimes, video communication applications to the requirement of video coding is: the code stream of generation should have the ability of anti-channel error code and channel fluctuation, should have stronger error code recovery capacity.
Up-to-date international video encoding standard H.264/MPEG-4 AVC and previous standard (as H.261, H.263, MPEG-1, MPEG-2, MPEG-4 etc.) compare, existing significant raising the on code efficiency, it still adopts traditional prediction loop hybrid video coding method, and the relativity of time domain that makes full use of video sequence by prediction loop reaches the purpose of efficient compression.But the existence of prediction loop makes the code stream that generates be easy to be subjected to the influence of channel error code or fluctuation in transmission course, produces the propagation accumulation problem that does not match with error of predicting.Prediction does not match and is meant that the prediction reference frame of rebuilding in decoding end is different with the prediction reference frame that coding side uses, the decoding error of caused present frame.Mistake can be propagated along time domain, causes the decoding error of all subsequent prediction frames.
As the part of MPEG-4 video encoding standard, (finc granularityscalability FGS) is a kind of scalable video scheme that is widely known by the people to fine granular scalable.It is encoded the residual error after the motion compensation based on the block-based hybrid encoding frame of tradition on continuous bit-planes, thereby produces the scalability code stream.But a problem of its maximum is exactly that its code efficiency is too low, so it is difficult to and non-scalable encoder competition in actual applications.Field of video encoding is badly in need of solving the anti-error code problem the video transmission breakthroughly from theory and technology at present.
Summary of the invention
The object of the invention is to propose a kind of anti-error code video coding-decoding method that does not have prediction loop.
The invention is characterized in: this method contains two parts of Code And Decode, and wherein coded portion contains following steps successively when FPGA realizes:
Step (a): to the first two field picture F K-1Press H.264/MPEG-4 AVC intraframe coding method coding;
Step (b): subsequent frame is encoded in the following manner
Step (b1): with an exercise estimator to current frame image F to be encoded kThe former frame image F that has encoded K-1Carry out estimation by the motion estimation algorithm that adopts in the international video encoding standard of AVC H.264/MPEG-4, obtain corresponding motion vector;
Step (b2): the motion vector that obtains in the step (b1) is carried out entropy coding with variable length encoder by the entropy coder H.264/MPEG-4AVC, obtain motion vector bitstream (B 1);
Step (c): to described current frame image F to be encoded kSet by step described in (b1) H.264/MPEG-4AVC in the quantization method quantizer Q that adopts 1Quantize, quantization step is
Figure C200610012092D0005135621QIETU
, span is 4～128, the result after the quantification adopts the low-density checksum coding device to carry out chnnel coding, and output verification bit stream (B 2);
Step (d): to quantizer Q 1Output with an inverse DCT Q 1 -1Carry out inverse quantization, and with current frame image F to be encoded kWith this inverse DCT Q 1 -1The output result ask difference to obtain residual image;
Step (e): the residual image that step (d) is obtained by the integer transform mode that adopts in the international video encoding standard of AVC H.264/MPEG-4 with an integer transformer conversion, again to conversion coefficient set by step (c) described quantization method advance to use quantizer Q 2Quantize, quantization step is
Figure C200610012092D0005135643QIETU
, span is 8～32, the change quantization coefficient that the obtains entropy coding algorithm in (b2) set by step carries out entropy coding, obtain the residual error bit stream (
Figure C200610012092D0005142249QIETU
), and with the described motion vector bitstream (B of step (b2) 1) constitute code stream (1) together;
Step (f): described code stream 1 of step (e) and the described check bit stream of step (c) (B 2) constitute final output code flow jointly.
Decoded portion contains following steps successively when realizing in FPGA:
Step (g): to the first two field picture F K-1Decode decoded image by the intraframe decoder mode among the AVC H.264/MPEG-4
Figure C200610012092D00061
Deliver to a frame memory, the reference during as the next frame motion compensation, wherein motion compensation is meant by coding side and passes the motion vector information of coming, and finds corresponding prediction piece in reference frame, reconstructs whole predictive frame at last;
Step (h): all the other each frames are decoded as follows;
Step (i): the length variable decoder of described code stream (1) being delivered to the described algorithm of execution in step (b2) is decoded, and obtains motion vector information and change quantization coefficient;
Step (j): the change quantization coefficient that step (i) obtains is used inverse DCT Q successively 2 -1With inverse transformer T -1, obtain residual error coefficient;
Step (k): the motion vector information that obtains with step (i) is to the former frame decoded picture in the described frame memory
Figure C200610012092D00062
Carry out motion compensation, obtain the estimated image of present frame
Figure C200610012092D0006142323QIETU
, reuse described quantizer Q 1To this image
Figure C200610012092D0006142331QIETU
With quantization step be
Figure C200610012092D0006142336QIETU
Quantize, quantized result is the side information that low-density parity-check decoder is used;
Step (1): step (c) is obtained the input as low-density parity-check decoder of side information that check bit stream (B.) and step (k) obtain, the output result inverse DCT Q of low-density parity-check decoder then 1 -1Carry out inverse quantization, obtain the inverse quantization value ( ):
Step (m): the inverse quantization value that residual error coefficient that step (j) is obtained and step (1) obtain Sue for peace, obtain the decoded picture of present frame
Figure C200610012092D0006082751QIETU
, and this decoded picture
Figure C200610012092D0006082751QIETU
Store the reference of frame memory into as the next frame motion compensation.
Use proof:
1. coding side does not have prediction loop to avoid encoding and decoding end prediction mismatch problem;
2. adopt the Wyner-Ziv coding thinking, the code stream of generation has very strong error recovery characteristics;
3. clear in structure;
4. has higher code efficiency
Description of drawings
Fig. 1 encoder theory diagram of the present invention
Fig. 2 decoder theory diagram of the present invention
Fig. 3 encoder instance graph of the present invention
Fig. 4 decoder instance graph of the present invention
Embodiment
The present invention to the anti-error code requirement that video coding technique proposes, according to the Wyner-Ziv coding principle, provides a kind of no prediction loop method for video coding with anti-error code capacity according to current video communication.The codec theory structure block diagram of this method as depicted in figs. 1 and 2, F wherein kAnd F K-1Represent present frame to be encoded and encoded former frame image respectively,
Figure C200610012092D00071
With
Figure C200610012092D00072
Represent the decoded picture of present frame and former frame respectively, B 1, B 2And B 3Represent the check bit stream of motion vector encoder bit stream, channel encoder output and the coded bit stream of residual coding device output respectively.
A kind of anti-error code video coding-decoding method that does not have prediction loop that the present invention proposes may further comprise the steps:
(1) as shown in Figure 1, current frame image F to be encoded kThe former frame image F that has encoded K-1Deliver to exercise estimator and carry out estimation, the motion vector information that obtains is delivered to motion vector coder and is encoded, output movement vector bit stream B 1
(2) current frame image F to be encoded kDeliver to the channel encoder coding after quantizer quantizes, channel encoder generates check bit stream B 2
(3) inverse DCT, F are delivered in the output of quantizer simultaneously kDeliver to residual coding device coding generation residual error bit stream with the residual image that the inverse quantization result asks difference to obtain
Figure C200610012092D0007142521QIETU
(annotate: above is the encoder workflow)
(4) as shown in Figure 2, motion vector bitstream B 1Deliver to motion vector decoder decoding and obtain motion vector information, and with the decoded picture of former frame
Figure C200610012092D00073
Deliver to together and carry out motion compensation in the motion compensator, obtain the estimated image of present frame
Figure C200610012092D0007142542QIETU
, and
Figure C200610012092D0007142542QIETU
Deliver in the quantizer and quantize, quantized result is as the side information of channel decoder;
(5) check bit stream B 2Obtain side information with previous step and deliver in the channel decoder and decode, the information that decoding obtains is delivered to and is carried out inverse quantization in the inverse DCT;
(6) residual error bit stream B 3Deliver in the residual error decoding device and decode, the numerical value summation that residual error that decoding obtains and previous step inverse quantization obtain obtains the decoded picture of present frame
The employing that the present invention proposes the H.264/MPEG-4 anti-code error video frequency coding method for designing embodiment of the no prediction loop of AVC standard is described with reference to the accompanying drawings as follows:
The encoder workflow is as follows:
(1) first two field picture is carried out intraframe coding method and encode, subsequent frame is carried out following mode encode;
(2) as shown in Figure 3, to current frame image F to be encoded kThe former frame image F that has encoded K-1The motion estimation algorithm that carries out adopting in the standard H.264 carries out estimation, and the motion vector that obtains carries out entropy coding with variable length encoder (VLC, Vary Length Code);
(3) to current frame image F to be encoded kAdopt the H.264 quantization method of standard quantizer Q 1Quantize, the result after the quantification adopts LDPC (Low Density Parity Check) encoder to carry out chnnel coding, and output verification code stream (Parity Bitstrcam);
(4) to Q 1Output inverse DCT Q 1 -1Carry out inverse quantization, and use F kAsk poor with the inverse quantization result, obtain residual image, residual image is carried out integer transform, conversion coefficient is adopted and identical quantization method quantizer of the 3rd step
Figure C200610012092D0008142643QIETU
Quantize, the change quantization coefficient that obtains carries out entropy coding with variable length encoder, and constitutes code stream 1 with motion vector;
(5) code stream 1 constitutes final output code flow with the verification code stream.
The decoder functions flow process is as follows:
(6) to first two field picture frame by frame in decoding process decode, decoded image is delivered in as shown in Figure 4 the frame memory (Framc Mcmory) as the reference of next frame motion compensation, and all the other each frames are decoded as follows:
(7) as shown in Figure 4, from code stream 1, decode motion vector and change quantization coefficient with length variable decoder;
(8) to change quantization coefficient inverse DCT Q 2 -1Carry out inverse quantization and inverse transformer T -1Carry out inverse transformation, obtain residual error coefficient, as the part of the 11st step reconstructed image;
(9) with the 7th motion vector that obtains of step to the former frame decoded picture in the frame memory
Figure C200610012092D00082
Carry out motion compensation, obtain the estimated image of present frame , and right Use quantizer Q 1Quantize, quantized result is as the side information of next step LDPC decoding;
(10) side information that obtains of verification code stream and previous step carries out inverse quantization to the decoding output of LDPC again as the input of LDPC decoder, obtains the inverse quantization value , as the part of next step reconstructed image;
Residual error coefficient that (11) the 8th steps obtained and previous step obtain
Figure C200610012092D0009142727QIETU
Summation obtains the decoded picture of present frame
Figure C200610012092D0006082751QIETU
, and this decoded picture stored in the frame memory into reference as the next frame motion compensation.

Claims (1)

1, a kind of anti-error code video coding-decoding method that does not have prediction loop is characterized in that, this method contains two parts of Code And Decode, contains following steps when wherein coded portion is realized in FPGA successively:
Step (a): to the first two field picture F K-1Press H.264/MPEG-4 AVC intraframe coding method coding;
Step (b): subsequent frame is encoded in the following manner
Step (b1): with an exercise estimator to current frame image F to be encoded kThe former frame image F that has encoded K1Carry out estimation by the motion estimation algorithm that adopts in the international video encoding standard of AVC H.264/MPEG-4, obtain corresponding motion vector;
Step (b2): the motion vector that obtains in the step (b1) is carried out entropy coding with variable length encoder by the entropy coder H.264/MPEG-4AVC, obtain motion vector bitstream B 1
Step (c): to described current frame image F to be encoded kSet by step described in (b1) H.264/MPEG-4AVC in the quantization method quantizer Q that adopts 1Quantize, quantization step is , span is 4～128, the result after the quantification adopts the low-density checksum coding device to carry out chnnel coding, and output verification bit stream B 2
Step (d): to quantizer
Figure C200610012092C0002095552QIETU
Output with an inverse DCT Q 1 1Carry out inverse quantization, and with current frame image F to be encoded kWith this inverse DCT Q 1 -1The output result ask difference to obtain residual image;
Step (c): the residual image that step (d) is obtained by the integer transform mode that adopts in the international video encoding standard of AVC H.264/MPEG-4 with an integer transformer conversion, again to conversion coefficient (c) described quantization method quantizer Q set by step 2Quantize, quantization step is
Figure C200610012092C0002095619QIETU
, span is 8～32, the change quantization coefficient that the obtains entropy coding algorithm in (b2) set by step carries out entropy coding, obtains the residual error bit stream
Figure C200610012092C0002095629QIETU
, and with the described motion vector bitstream B of step (b2) 1Constitute code stream 1 together;
Step (f): described code stream 1 of step (e) and the described check bit stream of step (c) B 2The final output code flow of common formation;
Decoded portion contains following steps successively when realizing in FPGA:
Step (g): to the first two field picture F KlDecode decoded image by the intraframe decoder mode among the AVC H.264/MPEG-4 Deliver to a frame memory, the reference during as the next frame motion compensation, wherein motion compensation is meant by coding side and passes the motion vector information of coming, and finds corresponding prediction piece in reference frame, reconstructs whole predictive frame at last;
Step (h): all the other each frames are decoded as follows;
Step (i): the length variable decoder of described code stream 1 being delivered to execution in step (b2) algorithm is decoded, and obtains motion vector information and change quantization coefficient;
Step (j): the change quantization coefficient that step (i) obtains is used inverse DCT Q successively 2 -1With inverse transformer T -1, obtain residual error coefficient;
Step (k): the motion vector information that obtains with step (i) is to the former frame decoded picture in the described frame memory
Figure C200610012092C0003095725QIETU
Carry out motion compensation, obtain the estimated image of present frame
Figure C200610012092C0003095730QIETU
, reuse described quantizer Q 1To this image
Figure C200610012092C0003095742QIETU
With quantization step be
Figure C200610012092C0003095747QIETU
Quantize, quantized result is the side information that low-density parity-check decoder is used;
Step (l): step (c) is obtained check bit stream B 2And the side information that obtains of step (k) is as the input of low-density parity-check decoder, the output result inverse DCT Q of low-density parity-check decoder then 1 1Carry out inverse quantization, obtain the inverse quantization value
Figure C200610012092C0003095809QIETU
:
Step (m): the inverse quantization value that residual error coefficient that step (j) is obtained and step (1) obtain
Figure C200610012092C0003095809QIETU
Sue for peace, obtain the decoded picture of present frame And this decoded picture Store the reference of frame memory into as the next frame motion compensation.
CN 200610012092 2006-06-02 2006-06-02 Non-predicted circulation anti-code error video frequency coding method Expired - Fee Related CN100508608C (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 200610012092 CN100508608C (en) 2006-06-02 2006-06-02 Non-predicted circulation anti-code error video frequency coding method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 200610012092 CN100508608C (en) 2006-06-02 2006-06-02 Non-predicted circulation anti-code error video frequency coding method

Publications (2)

Publication Number Publication Date
CN1889687A CN1889687A (en) 2007-01-03
CN100508608C true CN100508608C (en) 2009-07-01

Family

ID=37578945

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 200610012092 Expired - Fee Related CN100508608C (en) 2006-06-02 2006-06-02 Non-predicted circulation anti-code error video frequency coding method

Country Status (1)

Country Link
CN (1) CN100508608C (en)

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP5368631B2 (en) 2010-04-08 2013-12-18 株式会社東芝 Image encoding method, apparatus, and program
CN103813163B (en) * 2010-04-08 2017-03-01 株式会社东芝 Picture decoding method and picture decoding apparatus
CN101945286B (en) * 2010-09-29 2013-06-05 上海华为技术有限公司 Compression coding method, decoding decompression method, device and communication system
CN102055486A (en) * 2010-12-29 2011-05-11 启东金凤凰科技发展有限公司 Application of low-density check code in wireless data communication
CN102447898B (en) * 2011-08-26 2013-09-25 西安电子科技大学 Method for realizing KLT (Karhunen-Loeve Transform) by means of FPGA (Field Program Gate Array)
JP6391694B2 (en) * 2015-07-31 2018-09-19 エスゼット ディージェイアイ テクノロジー カンパニー リミテッドＳｚ Ｄｊｉ Ｔｅｃｈｎｏｌｏｇｙ Ｃｏ．，Ｌｔｄ Video encoding method and system
WO2017020184A1 (en) 2015-07-31 2017-02-09 SZ DJI Technology Co., Ltd. Methods of modifying search areas

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
基于H.26L的精细度可伸缩视频编码. 丁贵广，郭宝龙，计文平.通信学报，第25卷第9期. 2004
基于H.26L的精细度可伸缩视频编码. 丁贵广，郭宝龙，计文平.通信学报，第25卷第9期. 2004 *
精细可伸缩视频编码中的增强层编码方法研究. 丁贵广，郭宝龙.计算机工程与应用，第39卷第21期. 2003
精细可伸缩视频编码中的增强层编码方法研究. 丁贵广，郭宝龙.计算机工程与应用，第39卷第21期. 2003 *

Also Published As

Publication number Publication date
CN1889687A (en) 2007-01-03

Similar Documents

Publication Publication Date Title
CN101854548B (en) Wireless multimedia sensor network-oriented video compression method
CN103329522B (en) For the method using dictionary encoding video
CN100508608C (en) Non-predicted circulation anti-code error video frequency coding method
US9014499B2 (en) Distributed source coding using prediction modes obtained from side information
CN102572435B (en) Compressive sampling-based (CS-based) video coding/decoding system and method thereof
CN101601296A (en) Use the system and method that is used for gradable video encoding of telescopic mode flags
CN103002283A (en) Multi-view distributed video compression side information generation method
CN102769747A (en) Parallel iteration-based grading and distributed video coding/decoding method and system
TW201301900A (en) Method for decoding video encoded as bit stream in video decoder
CN101026758B (en) Video transcoding method and apparatus
CN102572428B (en) Side information estimating method oriented to distributed coding and decoding of multimedia sensor network
WO2019179096A1 (en) Compressive sensing-based quality-scalable fast coding method
CN102833536A (en) Distributed video encoding and decoding method facing to wireless sensor network
CN100551060C (en) A kind of video coding-decoding method
CN105611301A (en) Distributed video coding and decoding method based on wavelet domain residual errors
CN103826122B (en) Complexity-balanced video coding method and decoding method thereof
KR20120014947A (en) Apparatus and method for intra prediction encoding/decoding based on all direction
CN103634608A (en) High-performance video coding lossless mode residual error transform method
US9307257B2 (en) Method for improving compression efficiency of distributed source coding using intra-band information
CN103533351B (en) A kind of method for compressing image quantifying table more
Milani et al. Distributed video coding based on lossy syndromes generated in hybrid pixel/transform domain
CN101150721B (en) Coding method with adaptable bit element plane coding mode
CN107749993B (en) Distributed video coding information source based on MMSE reconstruct is distorted evaluation method
KR101554541B1 (en) Method for distributed vidio encoding and decoding of foreground block utilizing video analytics information
Maugey et al. Analysis of error propagation due to frame losses in a distributed video coding system

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
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20090701

Termination date: 20150602

EXPY Termination of patent right or utility model