CN111510722A - High-quality transcoding method for video image with excellent error code resistance - Google Patents
High-quality transcoding method for video image with excellent error code resistance Download PDFInfo
- Publication number
- CN111510722A CN111510722A CN202010345944.8A CN202010345944A CN111510722A CN 111510722 A CN111510722 A CN 111510722A CN 202010345944 A CN202010345944 A CN 202010345944A CN 111510722 A CN111510722 A CN 111510722A
- Authority
- CN
- China
- Prior art keywords
- distortion
- coding
- video
- error
- transcoding
- 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
- H04N19/40—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using video transcoding, i.e. partial or full decoding of a coded input stream followed by re-encoding of the decoded output stream
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/134—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or criterion affecting or controlling the adaptive coding
- H04N19/146—Data rate or code amount at the encoder output
- H04N19/147—Data rate or code amount at the encoder output according to rate distortion criteria
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
The invention provides a high-quality transcoding method of a video image with excellent error code resistance, which provides a fragment self-adaptive transcoding algorithm for optimizing distortion-limited information source coding at a video image transcoding end, wherein the algorithm is based on the MPEG-4 advanced video coding fragmentation technology and combines the distortion-limited information source coding theory to establish a uniform fragment number-limited distortion information source coding model, adaptively adjusts the number strategy of each frame of image fragments to take the code rate and the fault tolerance into consideration, and provides a space-time related fast stepping error masking correction algorithm at a video image decoding end, wherein the algorithm corrects a video residual error on the basis of the fast stepping image error recovery method, and then recovers the whole macro block by combining time domain information of the video image. The experimental result shows that the two fault-tolerant algorithms can effectively enhance the fault-tolerant performance of the video image.
Description
Technical Field
The invention relates to a high-quality transcoding method for video images, in particular to a high-quality transcoding method for video images with excellent error code resistance, and belongs to the technical field of video image transcoding.
Background
Since the nineties of the twentieth century, the rapid development of mobile communication technology, computer technology and internet technology has led to the development of a large number of media industries under new technology support systems, and the data volume of multimedia information has increased explosively. With such a background, people are increasingly engaged in their life-work and learning with the multimedia industry. Video image information has a large amount of information and much redundant information, and the purpose of video image encoding is to compress the video image information as much as possible. With the development of video image encoding technology, various video encoding standards have been proposed, such as MPEG-4, MPEG-4 advanced video encoding, and the like. Once the video code stream is compressed, the characteristics of high delay, low bandwidth and high error code of a channel inevitably cause error code problem in the video transmission process, and error code propagation is caused. Especially for wireless networks, the error rate is high, and different error rates are possessed by different wireless channels. In order to overcome error propagation, fault-tolerant information needs to be added into a video code stream to improve the robustness of the code stream. The fault-tolerant error-correcting video transcoding is to enhance the robustness of a compressed code stream by adding a fault-tolerant error-correcting algorithm in the video transcoding communication process and improve the error recovery performance. Therefore, the method for transcoding the video image with fault tolerance and error correction is researched and developed, and has important practical value and application value.
In the prior art, the research and development of the fault-tolerant error-correcting transcoding of the video image mainly focuses on two aspects of a fault-tolerant algorithm and a distortion-limited source coding model. The research and development of the fault-tolerant algorithm are the basic problem of the fault-tolerant error-correcting video transcoding, various fault-tolerant algorithms are mainly embedded into the video image transcoding communication process, and the fault-tolerant algorithm can be divided into three aspects according to the different embedding positions of the fault-tolerant algorithm: encoder-based fault-tolerant algorithms, decoder-based fault-tolerant algorithms and feedback channel-based fault-tolerant algorithms. The encoder-based fault-tolerant algorithm mainly comprises mode updating, resynchronization mark insertion and fault-tolerant entropy coding, the decoder-based fault-tolerant algorithm mainly comprises error masking correction, and the feedback channel-based fault-tolerant algorithm mainly comprises automatic repeat request, error tracking and the like. The research and development of a limited distortion source coding model is a key problem of video fault-tolerant error correction transcoding, and mainly solves the balance problem between code rate and distortion caused by embedding of a fault-tolerant error correction algorithm.
In the aspect of a fault-tolerant algorithm in the prior art, early work is a fault-tolerant entropy coding technology provided in a fault-tolerant transcoding framework based on MPEG-2, the technology takes a frame with a fixed length consisting of V L C codes as a resynchronization unit, only one frame needs to be dropped when a transmission error occurs, error diffusion is avoided, the signal-to-noise ratio can be improved by 2 to 5dB compared with the insertion of a resynchronization mark, but EREC recombines V L C codes, so that source syntax needs to be changed, and the use of the algorithm is limited.
In the aspect of a limited distortion source coding optimization model in the prior art, an idea is provided at the beginning of this century, and the limited distortion source coding of fault-tolerant insertion is decomposed into two sub-problems: optimization of macroblock mode selection and optimization of resynchronization marker insertion enable a 1.3 to 2.3dB increase in signal-to-noise ratio, but this approach does not take into account the error diffusion characteristics of Inter frames. Almost simultaneously, a distortion-limited source coding model based on statistical models has emerged, which expresses the error diffusion properties of video bit streams and can be used to calculate the optimal bit allocation between source coding and error-tolerant coding, but this approach does not take into account the effects of error concealment corrections. And then redundant motion vectors are inserted into the video stream in video transcoding, so that a decoder can recover damaged or lost macro block motion vectors by using the redundant motion vectors, the performance of error concealment correction at a decoding end is improved, and the code rate is increased due to the insertion of the redundant motion vectors, but the algorithm needs to require the capability of decoding the redundant motion vectors at the decoding end.
In summary, the drawbacks of the video image transcoding method in the prior art are mainly manifested in the following aspects: firstly, once a video code stream is compressed in the prior art, the video code stream is inevitably subjected to the characteristics of high delay, low bandwidth and high error code of a channel, so that the problem of error code in the video transmission process is caused, error code propagation is caused, particularly for a wireless network, the error code rate is higher, and the prior art lacks an effective solution; secondly, fault-tolerant information is not added into the video code stream to improve the robustness of the code stream in the prior art, error recovery performance is not available, error propagation cannot be overcome, and the robustness of the compressed code stream is poor; thirdly, the fault-tolerant methods in the prior art are not optimized for distortion-limited source coding, and when the embedding of a fault-tolerant algorithm causes a problem of code rate increase, better balance between code rate and distortion cannot be obtained; fourthly, in the prior art, the limited distortion source coding optimization model does not consider the error diffusion characteristic of Inter frames and the influence of error masking correction, and a decoding end is required to have the capability of decoding redundant motion vectors, so that various large restriction factors exist, a plurality of difficulties exist in the using process, and the quality of image transcoding is not high.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides a high-quality transcoding method of video images with excellent error code resistance, which is a fault-tolerant error-correcting video transcoding algorithm based on image error recovery on the basic framework of video transcoding of MPEG-4 advanced video coding. A self-adaptive transcoding algorithm for optimizing the slicing of distortion-limited information source coding is provided at the video image transcoding end, the algorithm is based on the slicing technology of MPEG-4 advanced video coding and combines the distortion-limited information source coding theory to establish a uniform slicing number distortion information source coding model, the strategy of adaptively adjusting the quantity of image slices of each frame is used for considering both the code rate and the fault tolerance, a space-time related fast stepping error masking correction algorithm is provided at the video image decoding end, the video residual is corrected based on the algorithm fast stepping image error recovery method, and then the whole macro block is recovered by combining the time domain information of the video image. Experimental results show that the two fault-tolerant algorithms can effectively enhance the fault-tolerant performance of the video image, the objective quality of the video image is improved by 0.38dB to 1.49dB under the condition that the code rate is properly increased by the fragmentation self-adaptive transcoding algorithm, and the objective quality of the space-time related fast stepping error concealment correction algorithm is improved by 6.6% to 9.5% compared with the traditional frame copy error concealment correction algorithm in MPEG-4 advanced video coding.
In order to achieve the technical effects, the technical scheme adopted by the invention is as follows:
a video image high quality transcoding method with excellent error code resistance performance is based on the video transcoding basic framework of MPEG-4 advanced video coding, the error code resistance performance of the video image transcoding method is improved from two angles, firstly, the error code resistance performance of a compressed code stream is improved through fault-tolerant error correction transcoding, and secondly, a video image decoding end completes error shielding correction after error code packet loss occurs;
when the MPEG-4 high-level video coding video code stream is transmitted to a video image transcoding end, the strategy of adaptively adjusting the fragment quantity of each frame image is adopted by the fragment adaptive transcoding algorithm for optimizing distortion-limited source coding provided by the invention at the video image transcoding end according to the packet loss rate fed back by a video channel, so as to achieve the balance between the code rate and the error code resistance; at a video image decoding end, a space-time related fast stepping error masking correction algorithm is provided by the invention, the residual error of the video is corrected on the basis of a fast stepping image error recovery method, and then lost blocks are corrected in a matched manner by combining the time domain characteristics of the video image;
the optimal slicing self-adaptive transcoding algorithm for optimizing the distortion-limited source coding is based on the distortion-limited source coding, an optimal slicing number is selected for each frame of image by establishing a slicing number distortion-limited source coding model, the code rate is not excessively increased, and meanwhile the error code resistance of the transcoded video stream is enhanced by slicing, and the method comprises the steps of limiting the distortion source coding and optimizing the slicing number of the distortion source coding.
The invention relates to a high-quality transcoding method of a video image with excellent error code resistance, and further, limited distortion information source coding is obtained by an information rate-distortion functionm＝{f1，f2，f3...fnAfter encoding, the sink is Gm＝{g1，g2，g3...gnM and n are coefficients, and probability spaces of an information source and an information sink are defined as A (f) respectivelyj)、B(gk) The average distortion function e (b) is defined as follows:
wherein, e (f)j，gk) Expressing the distortion degree, and selecting absolute difference Sum (SAD) as a measurement criterion;
in the presence of a (f)j) I.e. the source probability distribution, the mean value of the distortion is only subject to the conditional probability B (g)k|fj) Thus a conditional probability B (g) must be foundk|fj) So that E (B) is less than or equal to E, and recording as follows:
BE(E) E ≦ B) formula 2
I.e. BEIs a conditional probability set that ensures that the average distortion E (b) is within the allowable range E;
further defining the average information content C (X, Y) received by the receiving end:
similarly, when condition A (f)j) At a given time, only B (g) is presentk|fj) The size of C (X, Y) is determined so that H (E), the distortion-limited source coding function, becomes at BEMinimizing the information quantity in the range, as shown in formula 4;
formula 4 shows that under the condition of distortion degree E tolerable for subjective perception of people, the lower limit of average mutual information obtained by source coding, namely the limit code rate of data compression, if the code rate H is lower than the limit distortion source coding function H (E), the average distortion of the coding strategy is certainly larger than E no matter which coding strategy is adopted;
the distortion-limited source coding optimization problem of video coding is under the condition that the compression bit rate does not exceed H, namely, under the condition that the bit rate H is less than or equal to the transmission rate H of a channelcThen, finding the most suitable coding mode to obtain the minimum video image distortion E as the following formula;
min { E }, when H is less than or equal to HcHour type 5
Satisfying equation 5 requires choosing a balance point between the distortion E and the compression bit rate H;
when coding a video sequence, different coding parameters can be selected to obtain the image distortion E after decoding and the bit rate H during coding of the corresponding coding mode, each H corresponds to one E under different modes, so that each (H, E) point is found, the parameter coding in the candidate parameter set is used in sequence to obtain a value pair of (H, E), the value pairs are fitted by a curve to obtain a distortion-limiting source coding curve, namely an H-E curve, if the code rate condition H is satisfiedcIf the minimum distortion E is fixed, the position of the minimum distortion E is positioned on a limited distortion source coding curve, the video coding limited distortion source coding optimization aims to calculate an optimal coding parameter set, and H-E points in the coding parameter set are close to or even on the limited distortion source coding curve as much as possible;
the invention adopts a Lagrange multiplier method to solve the problem of solving the extreme value under the constraint condition, obtains a plurality of optimal encoding parameter sets, and performs encoding under the optimal encoding parameter sets, so that the encoding performance of the video can reach the best.
A video image high-quality transcoding method with excellent error code resistance performance, further, the optimization of the number of distortion-limited information source coding fragments comprises an information source code rate model, a transmission distortion model, a Lagrange multiplier and the selection of the number of fragments;
the selection problem in coding mode when coding can be classified as: finding the best mode among the candidate set of coding modes such that the code rate condition HcIn the case, the video quality distortion E is minimal;
assuming that X represents the frame to be encoded and M represents the set of alternative coding modes, the distortion-limited source coding optimization problem during encoding can be expressed as:
where E (X, M) is the distortion level of the current coding frame in coding mode M, H (X, M) is the number of bits of the current coding frame in the selected coding mode M, and HcTo define the code rate.
The invention selects the best number of fragmentation modes N under the condition that the packet loss rate of a channel is L through a distortion limiting source coding function, wherein E is the distortion E (X, N | P) under the condition of packet loss and error code, and the extreme value under the condition of solving the constraint adopts a Lagrange multiplier method to obtain the following equation:
min{J(X，N，λ，L)}
j (X, N, λ, L) ═ D (X, N, L) + λ × H (X, N) formula 7
Lambda represents Lagrange operator, if lambda parameter can be obtained, the optimum fragment number N can be found through extremum, and the constrained code rate H is obtainedcMinimum distortion E (X, N) under the conditions;
the fragmentation technology does not affect the distortion caused by quantization in the source coding, so that only the influence caused by the code rate change caused by the fragmentation quantity and the distortion caused by the error code packet loss in the transmission is considered. The proper number of fragments is selected for transcoding through distortion-limited source coding optimization, and balance between code rate and distortion is achieved, and a distortion-limited source coding model selected based on the number of fragments is established:
j (n) ═ e (n) + λ h (n) formula 8
Wherein n represents the number of slices per frame, e (n) represents the transmission distortion when the number of slices is n, and h (n) represents the code rate when the number of slices is n. Then the optimal number of slices:
a high-quality transcoding method of video image with excellent error code resistance performance is further characterized in that in an information source code rate model, the increase of transcoding code rate caused by fragmentation can be divided into two parts, namely H (n) in a distortion-limited information source coding model
H(n)＝H0+ Δ H (n) formula 10
Wherein H0The code rate before transcoding is represented, and Δ h (n) is composed of two parts, the first part is the increased bit number inserted into the slice header due to the slicing, the second part is the increased bit number caused by all predictions performed in slices due to the slicing transcoding function, so that the inter-predicted macroblock before transcoding originally becomes the intra-slice prediction after transcoding, and the increased bit number is represented by a formula:
ΔH(n)＝n·Hhead+m·ΔHMBformula 11
Wherein HheadThe bit number of slice header is represented, m represents the number of macroblocks of inter-prediction of transcoding front in current frame which become intra-prediction due to slice transcoding, and Δ HMBThe average value of the number of bits increased by the macro block which becomes the intra-slice prediction after the transcoding of the previous frame of slice is represented, and the calculation formula of m is as follows:
m＝∑mMBformula 12
In the transmission distortion model, aiming at the estimation problem of transmission distortion E (n) caused by fragmentation, a transmission distortion model of the transmission distortion E about the fragmentation number n and the packet loss rate L is established, and the following relation is obtained by simulating the transmission distortion result of each frame under different fragmentation numbers and different packet loss rates, wherein SAD (C)i) The maximum distortion representing the ith frame, i.e. the sum of all pixel points, is defined as follows:
the transmission distortion model reflects the relationship between the transmission distortion E and the number n of the fragments and the packet loss rate L, the distortion gradually increases with the increase of the packet loss rate, and the increase of the number of the fragments can inhibit the transmission distortion under the condition that the packet loss rate of the current frame is constant, so that the transmission distortion model is used for estimating the distortion degree under different fragment codes when the packet loss rate of the current channel is known.
In the Lagrange multiplier, because the position derivative of the minimum value of J (n) is zero, the derivative of n is obtained by taking the distortion limiting source coding function value J (n) as E + lambda H:
therefore, λ can be obtained by the transmission distortion model E and the source rate model H, but since the number of slices n is a discrete integer, different code rates H and distortions E can be calculated only when the number of slices n is different positive integers, so the value of λ is approximately calculated by the difference between the code rate H and the distortion E, that is, equation 16:
a video image high-quality transcoding method with excellent error code resistance performance is further provided, and in the selection of the number of fragments, the detailed steps of optimizing a fragment self-adaptive transcoding algorithm of distortion-limited source coding are as follows:
firstly, feeding back the current average packet loss rate L by a video channel;
selecting different fragment quantities, and respectively calculating code rate H (n), transmission distortion E (n) and Lagrange multiplier lambda under different fragment quantities;
thirdly, respectively calculating corresponding J values by using a limited distortion source coding function formula J (n) E + lambda H, and selecting n with the minimum J as the last fragment number to perform fault-tolerant error correction fragment transcoding;
fourthly, repeating the second step and the third step for the next frame;
the algorithm improves the inhibition effect on error code diffusion by adaptively adjusting the quantity strategy of each frame of image fragments through a distortion-limited source coding optimization method, and balances the code rate and the error code resistance.
A video image high quality transcoding method with excellent error code resistance performance, further, a space-time related fast stepping error masking correction algorithm acts on a video image decoding end, fast image error recovery of video residual domain diffusion is firstly carried out, and then lost blocks are corrected in a matched mode by combining time correlation information of a video;
the fast stepping error masking correction algorithm of the time-space correlation is based on the fast stepping algorithm, combines the time correlation characteristic of the video, not only utilizes the correlation of the image in the space domain, but also utilizes the information of the reference frame in the time domain to correct the lost macro block; when an error code stream generated by channel packet loss of a transcoded video file is received by a decoder, correcting an error macro block area, if a reference macro block of a reference frame can be correctly received, performing error concealment correction by adopting a space-time related fast stepping error concealment correction algorithm, if a motion vector of an error macro block is not lost, directly finding the best reference macro block of the reference frame through the motion vector, and if the motion vector of the error macro block is lost, matching the most suitable macro block in a frame before the outer boundary of the lost macro block for error concealment correction.
Compared with the prior art, the invention has the advantages and innovation points that:
the invention provides a high-quality transcoding method of video images with excellent error code resistance, which is a fault-tolerant video transcoding method based on image restoration on the basis of a basic framework of video image transcoding based on MPEG-4 advanced video coding, and the method is elaborated in detail from the aspects of fault-tolerant error correction based on an encoder, fault-tolerant error correction based on a decoder and fault-tolerant error correction based on a feedback channel respectively due to different positions of the action of an error-correcting algorithm. A self-adaptive transcoding algorithm for optimizing the slicing of distortion-limited information source coding is provided at the video image transcoding end, the algorithm is based on the slicing technology of MPEG-4 advanced video coding and combines the distortion-limited information source coding theory to establish a uniform slicing number distortion information source coding model, the strategy of adaptively adjusting the quantity of image slices of each frame is used for considering both the code rate and the fault tolerance, a space-time related fast stepping error masking correction algorithm is provided at the video image decoding end, the video residual is corrected based on the algorithm fast stepping image error recovery method, and then the whole macro block is recovered by combining the time domain information of the video image. The two algorithms are mutually matched and coordinated, and cooperate to correct errors in video transcoding, so that a video image high-quality transcoding method with excellent error code resistance is formed.
The invention provides a high-quality transcoding method of video images with excellent error code resistance, which is characterized in that in the transcoding process of the video images, a fragment self-adaptive transcoding algorithm for optimizing distortion-limited information source coding is provided, the algorithm establishes a uniform fragment number distortion-limited information source coding model on the basis of the distortion-limited information source coding theory, gives consideration to the error code resistance and the code rate, and improves the inhibition effect on error code diffusion by adaptively adjusting the fragment number strategy of each frame of image on the premise of not excessively increasing the code rate. The experimental result shows that the objective quality of the video is improved by 0.38dB to 1.49dB under the condition of properly increasing the code rate.
Thirdly, the invention provides a high-quality transcoding method of video images with excellent error code resistance, and in the process of decoding the video images, a space-time related rapid stepping error concealment correction algorithm is provided, the algorithm corrects the residual error of the video on the basis of a rapid stepping image error recovery method, and then combines the time domain characteristics of the video to cooperatively repair the lost block so as to obtain better video error concealment effect. The experimental result shows that the objective quality of the algorithm is improved by 6.6-9.5% compared with the traditional frame copy error concealment algorithm in MPEG-4 high-level video coding, and meanwhile, the subjective effect is obviously improved.
Drawings
Fig. 1 is a block diagram of a video image high-quality transcoding method with excellent error-code-resistance performance according to the present invention.
Fig. 2 is a diagram of the distortion-limited source coding H-E curve of the present invention.
FIG. 3 is a flow chart of the spatiotemporal correlation fast stepping error concealment correction algorithm of the present invention.
Fig. 4 is a schematic diagram of the repair principle of the fast stepping algorithm of the present invention.
FIG. 5 is a schematic diagram of the fast stepping error concealment method of the invention.
Detailed Description
The following describes a technical solution of a video image high-quality transcoding method with excellent error-code-error-resistance performance, which is provided by the present invention, with reference to the accompanying drawings, so that those skilled in the art can better understand the present invention and can implement the present invention.
The invention provides a high-quality transcoding method of video images with excellent error code resistance, which is based on a video transcoding basic framework of MPEG-4 high-level video coding and provides an MPEG-4 high-level video coding fault-tolerant transcoding algorithm for improving the image error recovery performance.
As shown in fig. 1, when an MPEG-4 advanced video coding video bitstream is transmitted to a video image transcoding end, at the video image transcoding end, according to a packet loss rate fed back by a video channel, a strategy for adaptively adjusting the number of segments of each frame of image is provided by a segment adaptive transcoding algorithm for optimizing distortion-limited source coding according to the present invention, so as to achieve a balance between a code rate and an error-resistant performance; at the video image decoding end, the invention provides a space-time related fast stepping error masking correction algorithm, corrects the residual error of the video on the basis of a fast stepping image error recovery method, and then corrects the lost block by combining the time domain characteristics of the video image in a matched manner. These two algorithms encompassed by the present invention are described in detail below.
First, optimizing slicing self-adaptive transcoding algorithm of limited distortion source coding
The slicing technique of MPEG-4 high-grade video coding can achieve the purpose of error code resistance through the slicing technique due to the independent coding in slices. The more the number of slices, the more the code rate increases and the lower the coding efficiency. Therefore, the slicing technique is a double-edged sword, and it is very important to balance the code rate and the distortion while adopting the slicing technique.
The invention provides a fragmentation self-adaptive transcoding algorithm for optimizing the limited distortion information source coding by establishing a fragmentation number limited distortion information source coding model from limited distortion information source coding, selects an optimal fragmentation number for each frame of image, and enhances the error code resistance of the transcoded video stream by fragmentation while ensuring that the code rate is not excessively increased.
Distortion limited source coding
The limited distortion source coding is obtained by an information rate-distortion function, and under the real condition, a small amount of distortion exists in a signal and can be tolerated, but once the distortion exceeds a certain degree, the information quality is sharply reduced. Therefore, the distortion degree is required to be limited, meanwhile, the distortion measure gives a quantitative expression, and the invention introduces a distortion-limiting source coding function.
The video coding inner-limit distortion source coding is the correlation between the coding rate and the distortion degree of a video image, wherein the coding rate comprises the bit number of the total coding data of reference frame numbers, motion vectors and prediction residual values which are finally required to be transmitted when different coding parameters, prediction modes and quantization parameters are selected.
The purpose of the distortion-limited source coding optimization is to obtain the image distortion degree as small as possible under the condition that the coding code rate is as small as possible, and help the coder to achieve the highest coding efficiency. Meanwhile, the distortion-limited source coding optimization can also have some other special purposes, such as minimum distortion degree under the condition that the limited code rate does not exceed a certain value, and minimum code rate under the condition that the distortion degree is ensured not to exceed a certain lower limit.
In the MPEG-4 high-grade video coding encoder, the Lagrange optimization algorithm is used for selecting the optimal prediction mode or coding parameters, so that the purposes of coding control and distortion-limiting source coding optimization are achieved. The complexity of the distortion-limited source coding optimization algorithm and the degree of optimization realization influence the operation efficiency of the encoder and the requirement on configuration to a great extent, and the process is very important.
Let the information source be Fm＝{f1，f2，f3...fnAfter encoding, the sink is Gm＝{g1，g2，g3...gnM and n are coefficients, and probability spaces of an information source and an information sink are defined as A (f) respectivelyj)、B(gk) The average distortion function e (b) is defined as follows:
wherein, e (f)j，gk) The distortion is represented, and the measurement criterion is absolute difference Sum (SAD).
In the presence of a (f)j) I.e. the source probability distribution, the mean value of the distortion is only subject to the conditional probability B (g)k|fj) Thus a conditional probability B (g) must be foundk|fj) So that E (B) is less than or equal to E, and recording as follows:
BE(E) E ≦ B) formula 2
I.e. BEIs a conditional probability set that ensures that the average distortion E (b) is within the allowable range E.
Further defining the average information content C (X, Y) received by the receiving end:
similarly, when condition A (f)j) At a given time, only B (g) is presentk|fj) The size of C (X, Y) is determined so that H (E), the distortion-limited source coding function, becomes at BEThe minimization of the amount of information in the range is shown in equation 4:
equation 4 shows that under the condition of distortion E tolerable for human subjective perception, the lower limit of the average mutual information obtained by source coding, i.e. the limit code rate of data compression, if the code rate H is lower than the distortion-limited source coding function H (E), the average distortion of the coding strategy is certainly greater than E no matter which coding strategy is adopted.
The problem of the limited distortion source coding optimization of video coding is thatUnder the condition that the compressed bit rate does not exceed H, i.e. at a bit rate H less than or equal to the transmission rate H of the channelcThen, find the best coding mode, so as to obtain the minimum video image distortion E, as follows:
min { E }, when H is less than or equal to HcHour type 5
Satisfying equation 5 requires choosing a balance point between the distortion E and the compression bit rate H.
In practical cases, when encoding a video sequence, different encoding parameters, such as slice number and quantization step size, may be selected to obtain the decoded image distortion E of a corresponding encoding mode and the bit rate H during encoding, where each H corresponds to one E in different modes, so that each (H, E) point is found, and parameter encoding in the candidate parameter set is sequentially used to obtain a value pair of (H, E), as shown in fig. 2, and the value pairs are fitted with a curve to obtain a distortion-limited encoded source curve, i.e., an H-E curve. If the code rate condition is HcAnd fixing, wherein the position of the minimum distortion E is positioned on a distortion-limited source coding curve, the video coding distortion-limited source coding optimization aims to calculate an optimal coding parameter set, and H-E points in the coding parameter set are as close as possible even on the distortion-limited source coding curve.
The invention adopts a Lagrange multiplier method to solve the problem of solving the extreme value under the constraint condition, obtains a plurality of optimal encoding parameter sets, and performs encoding under the optimal encoding parameter sets, so that the encoding performance of the video can reach the best.
(II) Limited distortion source code division chip number optimization
According to the limited distortion source coding method, the selection problem in the coding mode during coding can be classified as follows: finding the best mode among the candidate set of coding modes such that the code rate condition HcIn this case, the video quality distortion E is minimal.
Assuming that X represents the frame to be encoded and M represents the set of alternative coding modes, then the distortion-limited source coding optimization problem during encoding can be expressed as:
where E (X, M) is the distortion level of the current coding frame in coding mode M, H (X, M) is the number of bits of the current coding frame in the selected coding mode M, and HcTo define the code rate.
The invention selects the best number of the fragmentation modes N under the condition that the packet loss rate of a channel is L through a distortion limiting source coding function, wherein E is the distortion E (X, N | P) under the condition of packet loss and error code, and the extreme value under the condition of solving the constraint adopts a Lagrange multiplier method, so that the above problem becomes the following equation:
min{J(X，N，λ，L)}
j (X, N, λ, L) ═ D (X, N, L) + λ × H (X, N) formula 7
Lambda represents Lagrange operator, if lambda parameter can be obtained, the optimum fragment number N can be found through extremum, and the constrained code rate H is obtainedcMinimum distortion E (X, N) under the conditions.
The fragmentation technology does not affect the distortion caused by quantization in the source coding, so that only the influence caused by the code rate change caused by the fragmentation quantity and the distortion caused by the error code packet loss in the transmission is considered. The proper number of fragments is selected for transcoding through distortion-limited source coding optimization, and balance between code rate and distortion is achieved, and a distortion-limited source coding model selected based on the number of fragments is established:
j (n) ═ e (n) + λ h (n) formula 8
Wherein n represents the number of slices per frame, e (n) represents the transmission distortion when the number of slices is n, and h (n) represents the code rate when the number of slices is n. Then the optimal number of slices:
(1) source code rate model
As known from H (n) in the distortion-limited source coding model, the transcoding code rate increase caused by the fragmentation can be divided into two parts, namely
H(n)＝H0+ Δ H (n) formula 10
Wherein H0Representation before transcodingThe code rate, Δ h (n), is composed of two parts, the first part is the increased number of bits inserted into slice header due to slicing, the second part is due to slicing transcoding, all predictions are made in slices, so that inter-predicted macroblocks before transcoding originally become the increased number of bits due to intra-slice prediction after transcoding, and can be expressed by formula:
ΔH(n)＝n·Hhead+m·ΔHMBformula 11
Wherein HheadThe bit number of slice header is represented, m represents the number of macroblocks of inter-prediction of transcoding front in current frame which become intra-prediction due to slice transcoding, and Δ HMBThe average value of the number of bits increased by the macro block which becomes the intra-slice prediction after the transcoding of the previous frame of slice is represented, and the calculation formula of m is as follows:
m＝ΣmMBformula 12
(2) Transmission distortion model
Aiming at the estimation problem of transmission distortion E (n) caused by fragmentation, a transmission distortion model of the transmission distortion E about the fragmentation number n and the packet loss rate L is established, and the following relation is obtained by simulating the transmission distortion result of each frame under different fragmentation numbers and different packet loss rates, wherein SAD (C)i) The maximum distortion representing the ith frame, i.e. the sum of all pixel points, is defined as follows:
the transmission distortion model reflects the relationship between the transmission distortion E and the number of slices n and the packet loss rate L, the distortion gradually increases with the increase of the packet loss rate, and the increase of the number of slices can suppress the transmission distortion under the condition that the packet loss rate of the current frame is constant, therefore, when the packet loss rate of the current channel is known, the distortion degree under different slice codes can be estimated by using the transmission distortion model, and because the number of slices is too large, the influence of the code rate is larger, at this time, the significance of the slice transmission relative to the video is not large, and the subjective quality of the video with the packet loss rate exceeding 50% is not acceptable or considered, so n ═ 0, 1, 2, … 10 represents the number of slices, L∈ [0, 0.5] represents the packet loss rate range, and μ is a constant, and in the embodiment, 0.5 is taken.
(3) Lagrange multiplier
Since the position derivative where the minimum value j (n) is located is zero, the derivative of n is derived from the distortion-limited source coding function value j (n) ═ E + λ H:
therefore, λ can be obtained by the transmission distortion model E and the source rate model H, but since the number of slices n is a discrete integer, different code rates H and distortions E can be calculated only when the number of slices n is different positive integers, so the value of λ is approximately calculated by the difference between the code rate H and the distortion E, that is, formula 3.16:
(4) slice number selection
The detailed steps of the slicing self-adaptive transcoding algorithm for optimizing the distortion-limited source coding are as follows:
firstly, feeding back the current average packet loss rate L by a video channel;
selecting different fragment quantities, and respectively calculating code rate H (n), transmission distortion E (n) and Lagrange multiplier lambda under different fragment quantities;
thirdly, respectively calculating corresponding J values by using a limited distortion source coding function formula J (n) E + lambda H, and selecting n with the minimum J as the last fragment number to perform fault-tolerant error correction fragment transcoding;
fourthly, repeating the second step and the third step for the next frame;
the algorithm improves the inhibition effect on error code diffusion by adaptively adjusting the quantity strategy of each frame of image fragments through a distortion-limited source coding optimization method, and ensures that the balance between code rate and error code resistance is achieved.
Quick step error shading correction algorithm of two, space-time correlation
As shown in fig. 1, the fast stepping error concealment correction algorithm of the spatio-temporal correlation is applied to the video image decoding end. The invention describes an algorithm flow chart based on the basic thought of a space-time related fast stepping algorithm, and corrects the lost block by combining the time domain characteristic of a video image on the basis of the error masking correction of the fast stepping algorithm.
Algorithm (I) concept
If other characteristics of the video are not considered, the video is only used as an image for correcting one frame, and a method of sample texture or a method of sparse prior block splicing patches is selected, so that a good error repairing effect can be obtained. However, these correction methods are complicated, and when we use the image error recovery technique for video error concealment correction, one of the importance information of video different from image, i.e. temporal correlation, cannot be ignored. This results in video with more temporal reference information than images, i.e. there is a strong correlation in the temporal domain between the image frames in the video. For video sequences with relatively smooth motion, the stronger the temporal correlation, the higher the compression rate when encoding. By using the time correlation, more reference information can be provided in the image error recovery process of the video, however, this also puts a certain limit on the time requirement, because unlike the image error recovery, the natural correction speed is better for the video on the premise of ensuring the correction effect. On one hand, temporal correlation provides more available information for video fault tolerance; on the other hand, the time correlation also puts higher demands on the fault-tolerant effect.
Therefore, the maximum utilization of the temporal correlation of the video to better correct the lost block information in the video spatial domain is an important means for applying the image error recovery method to the video error concealment correction. However, the data to be processed is huge in video compared to images, and therefore, a fast image error recovery algorithm is selected to overcome this problem.
The invention provides a space-time related fast stepping error masking correction algorithm, which is not an image error recovery method directly used for carrying out video error masking correction, but fast image error recovery of video residual error domain diffusion is firstly carried out, and then lost blocks are corrected in a matched mode by combining time correlation information of videos.
The flow chart of the space-time related fast stepping error concealment correction algorithm is shown in fig. 3, and the space-time related fast stepping error concealment correction algorithm is based on the fast stepping algorithm and combines the time-related characteristics of the video, and not only utilizes the correlation of the image in the space domain, but also utilizes the information of the reference frame in the time domain to correct the lost macro block. When an error code stream generated by channel packet loss of a transcoded video file is received by a decoder, correcting an error macro block area, if a reference macro block of a reference frame can be correctly received, performing error concealment correction by adopting a space-time related fast stepping error concealment correction algorithm, if a motion vector of an error macro block is not lost, directly finding the best reference macro block of the reference frame through the motion vector, and if the motion vector of the error macro block is lost, matching the most suitable macro block in a frame before the outer boundary of the lost macro block for error concealment correction.
(II) error concealment correction algorithm
The fast stepping algorithm is a fast image error recovery method, and the correction principle is shown in FIG. 4, where I is the region to be corrected, and OIReferring to the boundary of the region I, the pixel points in the region to be corrected can be divided into 3 sets: firstly, all known pixels in the region outside the correction region I, secondly, all unknown pixels in the region to be corrected, and thirdly, pixels on the boundary of I.
Correcting the pixel in I, new pixel value is calculated to replace the original value, if the point m is the pixel to be corrected, a small neighborhood S (j) taking m as the center is selected, the value of each pixel in the neighborhood is known, and n is Bj(m), the formula for calculating m pixels from n pixel values is as follows:
where K (n) is the gradient of the brightness of n pointsValue, by neighborhood SjAnd (m) calculating new gray values of m pixels by all pixels, and introducing a weight function to evaluate the contribution of the values of different pixels in the neighborhood to the new pixel values due to different functions of all pixel points.
In equation 18, w (m, n) is a weight function, and the following equation is defined:
w (m, n) ═ dir (m, n) · dst (m, n) · lev (m, n) formula 19
Where dir (m, n) is defined as a direction factor, dst (m, n) is defined as a geometric distance factor, the pixel points closer to the normal direction have a larger influence on the m point, lev (m, n) is defined as a level set distance factor, and the pixel values closer to the edge of the correction region of the pixel m have a larger influence on the calculation result of the pixel m.
A quick stepping algorithm corrects lost areas sequentially, and is a diffusion-based image error recovery method.
Since the fast stepping algorithm is a fast correction method, which is a great advantage for correcting false occlusion in video, but since the fast stepping algorithm calculates the synergistic effect of corresponding geometric distance factor, direction factor and level set factor when introducing the weight function, more information which has little or no relation with correction points can be introduced. This results in a weak edge retention performance of the fast stepping algorithm, and if the fast stepping algorithm is only selected to correct the image in the video, the correction effect is not ideal.
In the error masking correction algorithm, time domain information is introduced by utilizing the time correlation of a video, and a quick stepping error masking correction method introducing the time domain information, namely a space-time correlation quick stepping error masking correction algorithm, is provided by combining with a quick stepping algorithm. The specific details of the algorithm are shown in fig. 5.
Wherein M isnFor the current error frame, its missing macroblock is the black part SnFrom reference frame M, assuming that reference frame information is not lostn-1The known reference block Sn-1，cnFor the current missing macroblock SnInformation of surrounding correctly received macroblocks, cn-1Is a reference block Sn-1Information of surrounding correctly received macroblocks. The spatio-temporal correlated fast stepping error concealment correction algorithm can be expressed as the formula:
in formula 20
Indicating a missing block SnCorrection result of cn-cn-1If the current frame M is the difference between the correctly received macroblock information around the lost macroblock and the correctly received information around the reference blocknS ofnIf the macroblock is not lost, Sn-Sn-1I.e. the residual between the current block and the reference block contains a certain amount of information even if it is small. However, due to the loss of information of the lost block, if the reference block in the reference frame is directly copied to the current lost block, a blocking effect is generated at the boundary, and in consideration of the correlation and smooth change characteristics on the image space domain, the residual between the current lost block and the reference block is corrected by using a fast stepping algorithm under the condition of known time domain information, so as to make up for the lost residual information. The method makes the area composed of the lost position and the surrounding macro blocks smooth, thereby eliminating the block effect. Therefore, under the condition of fully utilizing time domain and space domain information, the fast stepping algorithm is combined, and the error masking correction algorithm at the video image decoding end provided by the invention has good effect recovery capability on subjective quality.
The invention provides a high-quality transcoding method of video images with excellent error code resistance performance based on a fault-tolerant error-correcting video transcoding algorithm of image error recovery on the basis of a video transcoding basic framework of MPEG-4 advanced video coding. A self-adaptive transcoding algorithm for optimizing the slicing of distortion-limited information source coding is provided at the video image transcoding end, the algorithm is based on the slicing technology of MPEG-4 advanced video coding and combines the distortion-limited information source coding theory to establish a uniform slicing number distortion information source coding model, the strategy of adaptively adjusting the quantity of image slices of each frame is used for considering both the code rate and the fault tolerance, a space-time related fast stepping error masking correction algorithm is provided at the video image decoding end, the video residual is corrected based on the algorithm fast stepping image error recovery method, and then the whole macro block is recovered by combining the time domain information of the video image.
The invention respectively performs experimental test and analysis on the provided fragmentation self-adaptive transcoding algorithm for optimizing distortion-limited source coding and the space-time related fast stepping error masking correction algorithm. Experimental results show that the two fault-tolerant algorithms can effectively enhance the fault-tolerant performance of the video image, the objective quality of the video image is improved by 0.38dB to 1.49dB under the condition that the code rate is properly increased by the fragmentation self-adaptive transcoding algorithm, and the objective quality of the space-time related fast stepping error concealment correction algorithm is improved by 6.6% to 9.5% compared with the traditional frame copy error concealment correction algorithm in MPEG-4 advanced video coding.
Claims (8)
1. A video image high quality transcoding method with excellent error code resistance performance is characterized in that based on a video transcoding basic framework of MPEG-4 advanced video coding, the error code resistance performance of the video image transcoding method is improved from two angles, firstly, the error code resistance performance of a compressed code stream is improved through fault-tolerant error correction transcoding, and secondly, a video image decoding end completes error shielding correction after error code packet loss occurs;
when the MPEG-4 high-level video coding video code stream is transmitted to a video image transcoding end, the strategy of adaptively adjusting the fragment quantity of each frame image is adopted by the fragment adaptive transcoding algorithm for optimizing distortion-limited source coding provided by the invention at the video image transcoding end according to the packet loss rate fed back by a video channel, so as to achieve the balance between the code rate and the error code resistance; at a video image decoding end, a space-time related fast stepping error masking correction algorithm is provided by the invention, the residual error of the video is corrected on the basis of a fast stepping image error recovery method, and then lost blocks are corrected in a matched manner by combining the time domain characteristics of the video image;
the optimal slicing self-adaptive transcoding algorithm for optimizing the distortion-limited source coding is based on the distortion-limited source coding, an optimal slicing number is selected for each frame of image by establishing a slicing number distortion-limited source coding model, the code rate is not excessively increased, and meanwhile the error code resistance of the transcoded video stream is enhanced by slicing, and the method comprises the steps of limiting the distortion source coding and optimizing the slicing number of the distortion source coding.
2. The method as claimed in claim 1, wherein the limited distortion source coding is obtained by information rate-distortion function, and the invention introduces the limited distortion source coding function, and the source is set as Fm＝{f1，f2，f3...fnAfter encoding, the sink is Gm＝{g1，g2，g3...gnM and n are coefficients, and probability spaces of an information source and an information sink are defined as A (f) respectivelyj)、B(gk) The average distortion function e (b) is defined as follows:
wherein, e (f)j，gk) Expressing the distortion degree, and selecting absolute difference Sum (SAD) as a measurement criterion;
in the presence of a (f)j) I.e. the source probability distribution, the mean value of the distortion is only subject to the conditional probability B (g)k|fj) Thus a conditional probability B (g) must be foundk|fj) So that E (B) is less than or equal to E, and recording as follows:
BE(E) E ≦ B) formula 2
I.e. BEIs a conditional probability set that ensures that the average distortion E (B) is within the allowable range E；
Further defining the average information content C (X, Y) received by the receiving end:
similarly, when condition A (f)j) At a given time, only B (g) is presentk|fj) The size of C (X, Y) is determined so that H (E), the distortion-limited source coding function, becomes at BEMinimizing the information quantity in the range, as shown in formula 4;
formula 4 shows that under the condition of distortion degree E tolerable for subjective perception of people, the lower limit of average mutual information obtained by source coding, namely the limit code rate of data compression, if the code rate H is lower than the limit distortion source coding function H (E), the average distortion of the coding strategy is certainly larger than E no matter which coding strategy is adopted;
the distortion-limited source coding optimization problem of video coding is under the condition that the compression bit rate does not exceed H, namely, under the condition that the bit rate H is less than or equal to the transmission rate H of a channelcThen, finding the most suitable coding mode to obtain the minimum video image distortion E as the following formula;
min { E }, when H is less than or equal to HcHour type 5
Satisfying equation 5 requires choosing a balance point between the distortion E and the compression bit rate H;
when coding a video sequence, different coding parameters can be selected to obtain the image distortion E after decoding and the bit rate H during coding of the corresponding coding mode, each H corresponds to one E under different modes, so that each (H, E) point is found, the parameter coding in the candidate parameter set is used in sequence to obtain a value pair of (H, E), the value pairs are fitted by a curve to obtain a distortion-limiting source coding curve, namely an H-E curve, if the code rate condition H is satisfiedcFixing, the position of minimum distortion E is positioned on the distortion limiting source coding curve, and optimizing the video coding distortion limiting source codingThe method aims to calculate an optimal encoding parameter set, wherein H-E points in the encoding parameter set are as close as possible even on a limited distortion source encoding curve;
the invention adopts a Lagrange multiplier method to solve the problem of solving the extreme value under the constraint condition, obtains a plurality of optimal encoding parameter sets, and performs encoding under the optimal encoding parameter sets, so that the encoding performance of the video can reach the best.
3. The method of claim 1, wherein the optimization of the number of distortion-limited source coding slices comprises selection of a source code rate model, a transmission distortion model, a Lagrange multiplier and the number of slices;
the selection problem in coding mode when coding can be classified as: finding the best mode among the candidate set of coding modes such that the code rate condition HcIn the case, the video quality distortion E is minimal;
assuming that X represents the frame to be encoded and M represents the set of alternative coding modes, the distortion-limited source coding optimization problem during encoding can be expressed as:
where E (X, M) is the distortion level of the current coding frame in coding mode M, H (X, M) is the number of bits of the current coding frame in the selected coding mode M, and HcTo define the code rate.
The invention selects the best number of fragmentation modes N under the condition that the packet loss rate of a channel is L through a distortion limiting source coding function, wherein E is the distortion E (X, N | P) under the condition of packet loss and error code, and the extreme value under the condition of solving the constraint adopts a Lagrange multiplier method to obtain the following equation:
min{J(X，N，λ，L)}
j (X, N, λ, L) ═ D (X, N, L) + λ × H (X, N) formula 7
Lambda represents Lagrange operator, if lambda parameter can be obtained, the optimum fragment number N can be found through extremum, and the constrained code rate H is obtainedcMinimum distortion E (X, N) under the conditions;
the fragmentation technology does not affect the distortion caused by quantization in the source coding, so that only the influence caused by the code rate change caused by the fragmentation quantity and the distortion caused by the error code packet loss in the transmission is considered. The proper number of fragments is selected for transcoding through distortion-limited source coding optimization, and balance between code rate and distortion is achieved, and a distortion-limited source coding model selected based on the number of fragments is established:
j (n) ═ e (n) + λ h (n) formula 8
Wherein n represents the number of slices per frame, e (n) represents the transmission distortion when the number of slices is n, and h (n) represents the code rate when the number of slices is n. Then the optimal number of slices:
4. the method as claimed in claim 3, wherein in the source code rate model, as shown by H (n) in the distortion-limited source coding model, the increase of transcoding code rate due to fragmentation can be divided into two parts, that is, the increase of transcoding code rate can be divided into two parts
H(n)＝H0+ Δ H (n) formula 10
Wherein H0The code rate before transcoding is represented, and Δ h (n) is composed of two parts, the first part is the increased bit number inserted into the slice header due to the slicing, the second part is the increased bit number caused by all predictions performed in slices due to the slicing transcoding function, so that the inter-predicted macroblock before transcoding originally becomes the intra-slice prediction after transcoding, and the increased bit number is represented by a formula:
ΔH(n)＝n·Hhead+m·ΔHMBformula 11
Wherein HheadThe bit number of slice header is represented, m represents the number of macroblocks of inter-prediction of transcoding front in current frame which become intra-prediction due to slice transcoding, and Δ HMBRepresenting the average value of the number of bits added to macroblocks which become predicted in a slice after transcoding a slice of a previous frame, mThe calculation formula is as follows:
m＝∑mMBformula l2
5. The method of claim 3, wherein in the transmission distortion model, for the estimation problem of the transmission distortion E (n) caused by fragmentation, a transmission distortion model of the transmission distortion E with respect to the fragmentation number n and the packet loss rate L is established, and the following relationship is obtained by simulating the transmission distortion result of each frame under different fragmentation numbers and different packet loss rates, wherein SAD (C)i) The maximum distortion representing the ith frame, i.e. the sum of all pixel points, is defined as follows:
the transmission distortion model reflects the relationship between the transmission distortion E and the number n of the fragments and the packet loss rate L, the distortion gradually increases with the increase of the packet loss rate, and the increase of the number of the fragments can inhibit the transmission distortion under the condition that the packet loss rate of the current frame is constant, so that the transmission distortion model is used for estimating the distortion degree under different fragment codes when the packet loss rate of the current channel is known.
6. The method as claimed in claim 3, wherein in the Lagrangian multiplier, since the position derivative of the minimum value of J (n) is zero, the distortion-limited source coding function value J (n) is derived from E + λ H by:
therefore, λ can be obtained by the transmission distortion model E and the source rate model H, but since the number of slices n is a discrete integer, different code rates H and distortions E can be calculated only when the number of slices n is different positive integers, so the value of λ is approximately calculated by the difference between the code rate H and the distortion E, that is, equation 16:
7. the method of claim 3, wherein in the selection of the number of slices, the detailed steps of optimizing the slice adaptive transcoding algorithm for distortion-limited source coding are as follows:
firstly, feeding back the current average packet loss rate L by a video channel;
selecting different fragment quantities, and respectively calculating code rate H (n), transmission distortion E (n) and Lagrange multiplier lambda under different fragment quantities;
thirdly, respectively calculating corresponding J values by using a limited distortion source coding function formula J (n) E + lambda H, and selecting n with the minimum J as the last fragment number to perform fault-tolerant error correction fragment transcoding;
fourthly, repeating the second step and the third step for the next frame;
the algorithm improves the inhibition effect on error code diffusion by adaptively adjusting the quantity strategy of each frame of image fragments through a distortion-limited source coding optimization method, and balances the code rate and the error code resistance.
8. The method of claim 1, wherein a fast stepping error masking correction algorithm of spatiotemporal correlation acts on a video image decoding end, fast image error recovery of video residual domain diffusion is performed first, and then lost blocks are corrected in cooperation with time correlation information of a video;
the fast stepping error masking correction algorithm of the time-space correlation is based on the fast stepping algorithm, combines the time correlation characteristic of the video, not only utilizes the correlation of the image in the space domain, but also utilizes the information of the reference frame in the time domain to correct the lost macro block; when an error code stream generated by channel packet loss of a transcoded video file is received by a decoder, correcting an error macro block area, if a reference macro block of a reference frame can be correctly received, performing error concealment correction by adopting a space-time related fast stepping error concealment correction algorithm, if a motion vector of an error macro block is not lost, directly finding the best reference macro block of the reference frame through the motion vector, and if the motion vector of the error macro block is lost, matching the most suitable macro block in a frame before the outer boundary of the lost macro block for error concealment correction.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010345944.8A CN111510722A (en) | 2020-04-27 | 2020-04-27 | High-quality transcoding method for video image with excellent error code resistance |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010345944.8A CN111510722A (en) | 2020-04-27 | 2020-04-27 | High-quality transcoding method for video image with excellent error code resistance |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111510722A true CN111510722A (en) | 2020-08-07 |
Family
ID=71876448
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010345944.8A Pending CN111510722A (en) | 2020-04-27 | 2020-04-27 | High-quality transcoding method for video image with excellent error code resistance |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111510722A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113422950A (en) * | 2021-05-31 | 2021-09-21 | 北京达佳互联信息技术有限公司 | Training method and training device for image data processing model |
| CN114499765A (en) * | 2022-04-14 | 2022-05-13 | 航天宏图信息技术股份有限公司 | Data transmission method and system based on Beidou short message |
| CN117979006A (en) * | 2023-12-29 | 2024-05-03 | 吴水勇 | Low-illumination live video real-time transcoding enhancement bright structure quantization method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070177667A1 (en) * | 2006-01-20 | 2007-08-02 | Qualcomm Incorporated | Method and apparatus for error resilience algorithms in wireless video communication |
| CN102438150A (en) * | 2011-09-29 | 2012-05-02 | 华中科技大学 | Anti-channel-error-code lossless compression error-detection/correction coding/decoding method for image |
| CN107027031A (en) * | 2016-01-31 | 2017-08-08 | 西安电子科技大学 | A kind of coding method and device for video image |
| US20190281302A1 (en) * | 2018-03-12 | 2019-09-12 | Nvidia Corporation | Ssim-based rate distortion optimization for improved video perceptual quality |
-
2020
- 2020-04-27 CN CN202010345944.8A patent/CN111510722A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070177667A1 (en) * | 2006-01-20 | 2007-08-02 | Qualcomm Incorporated | Method and apparatus for error resilience algorithms in wireless video communication |
| CN102438150A (en) * | 2011-09-29 | 2012-05-02 | 华中科技大学 | Anti-channel-error-code lossless compression error-detection/correction coding/decoding method for image |
| CN107027031A (en) * | 2016-01-31 | 2017-08-08 | 西安电子科技大学 | A kind of coding method and device for video image |
| US20190281302A1 (en) * | 2018-03-12 | 2019-09-12 | Nvidia Corporation | Ssim-based rate distortion optimization for improved video perceptual quality |
Non-Patent Citations (2)
| Title |
|---|
| 周琴: ""无线视频通信中的错误恢复技术研究"", 《中国优秀硕士学位论文全文数据库 信息科技辑》, no. 03, pages 3 * |
| 韩镇: ""一种基于分层率失真优化的容错性视频转码算法"", 《计算机学报》, vol. 32, no. 6, pages 1152 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113422950A (en) * | 2021-05-31 | 2021-09-21 | 北京达佳互联信息技术有限公司 | Training method and training device for image data processing model |
| CN114499765A (en) * | 2022-04-14 | 2022-05-13 | 航天宏图信息技术股份有限公司 | Data transmission method and system based on Beidou short message |
| CN114499765B (en) * | 2022-04-14 | 2022-08-16 | 航天宏图信息技术股份有限公司 | Data transmission method and system based on Beidou short message |
| CN117979006A (en) * | 2023-12-29 | 2024-05-03 | 吴水勇 | Low-illumination live video real-time transcoding enhancement bright structure quantization method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111510722A (en) | High-quality transcoding method for video image with excellent error code resistance | |
| US20080025390A1 (en) | Adaptive video frame interpolation | |
| US20060188025A1 (en) | Error concealment | |
| US8982947B2 (en) | Rate control and video denoising for noisy video data | |
| CN105379268A (en) | Adapting robustness in video coding | |
| CN108012149B (en) | Code rate control method in video coding | |
| Heng et al. | End-to-end rate-distortion optimized MD mode selection for multiple description video coding | |
| CN110493603B (en) | Multi-view video transmission error control method based on rate distortion optimization of joint information source channel | |
| CN107343202B (en) | Feedback-free distributed video coding and decoding method based on additional code rate | |
| KR100712532B1 (en) | Apparatus and method for transcoding video error-resiliently using single description and multiple description switching | |
| Ye et al. | Improved side information generation with iterative decoding and frame interpolation for distributed video coding | |
| CN116647704B (en) | Video coding and decoding joint error control method for HEVC data option | |
| CN108737839B (en) | Lagrange factor calculation method for intra-frame code rate control | |
| EP2888877A1 (en) | Method and apparatus for estimating content complexity for video quality assessment | |
| Zhu et al. | Error concealment for video communication | |
| KR100601846B1 (en) | Apparatus and Method for Distortion Optimization of Moving Picture Compaction Encoder | |
| KR100513014B1 (en) | Video communication system and video coding method | |
| CN115002482A (en) | End-to-end video compression method and system using structural preservation motion estimation | |
| US20050175109A1 (en) | Optimal bit allocation for error resilient video transcoding | |
| KR101307682B1 (en) | method for error detection using the data hiding of motion vector based on the RDO for H.264/AVC baseline profile | |
| GB2488334A (en) | Decoding a sequence of encoded digital frames | |
| Kodavalla et al. | Distributed video coding: feedback-free architecture and implementation | |
| Kung et al. | Packet video transmission over wireless channels with adaptive channel rate allocation | |
| KR100801155B1 (en) | A spatial error concealment method with low complexity in h. 264/avc video coding | |
| Jang et al. | An efficient error resilient technique for applications of one-way video using transcoding and analysis by synthesis |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |