CN108769700A - Reduce the robustness video steganography method of H.264 interframe drift distortion - Google Patents
Reduce the robustness video steganography method of H.264 interframe drift distortion Download PDFInfo
- Publication number
- CN108769700A CN108769700A CN201810549353.5A CN201810549353A CN108769700A CN 108769700 A CN108769700 A CN 108769700A CN 201810549353 A CN201810549353 A CN 201810549353A CN 108769700 A CN108769700 A CN 108769700A
- Authority
- CN
- China
- Prior art keywords
- embedded
- dct
- formula
- coefficient
- matrix
- 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
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
- H04N19/46—Embedding additional information in the video signal during the compression process
- H04N19/467—Embedding additional information in the video signal during the compression process characterised by the embedded information being invisible, e.g. watermarking
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/12—Selection from among a plurality of transforms or standards, e.g. selection between discrete cosine transform [DCT] and sub-band transform or selection between H.263 and H.264
- H04N19/122—Selection of transform size, e.g. 8x8 or 2x4x8 DCT; Selection of sub-band transforms of varying structure or type
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/13—Adaptive entropy coding, e.g. adaptive variable length coding [AVLC] or context adaptive binary arithmetic coding [CABAC]
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
- H04N19/184—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being bits, e.g. of the compressed video 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
- H04N19/60—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
- H04N19/625—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding using discrete cosine transform [DCT]
Abstract
The invention discloses the robustness video steganography methods for reducing H.264 interframe drift distortion, implement according to the following steps：Step 1, interframe drift distortion is analyzed, obtains drift distortion Producing reason, then drift distortion is classified according to sizes to caused by, the embedded location of DCT luminance factors is selected according to the size of embedded drift distortion；Step 2, convolutional encoding is carried out to information sequence to be embedded, wherein each input symbols correspond to two output symbols and step 3 output sequence is embedded into DCT coefficient, after the completion of embedded, carry out entropy coding formation to H.264/AVC video flowing and carry close video sequence；Step 4, entropy decoding is carried out to H.264/AVC video flowing, extracts original DCT coefficient, obtained output sequence and export information sequence to be embedded to output sequence into row decoding, complete the extraction of information sequence to be embedded.The present invention improves Data Embedding Capacity, video quality and robustness.
Description
Technical field
The invention belongs to Information Hiding Techniques fields, and in particular to a kind of reduction robustness that H.264 interframe drift is distorted
Video steganography method.
Background technology
With the continuous development of current science and technology, we are connected into an entirety by the transmission of Internet resources.In all biographies
In defeated resource, have become the important part in life using video as the resource of medium.It is produced in transmission process various
" prism door " information of information security issue, Snowdon steals secret information event, 58 with city whole nation resume leakage event, McDonald of India
220W user's receipt is revealed etc., and serious country, enterprise, personal economic loss are caused.
The information security issue of itself is paid close attention to based on this more and more people.Therefore video steganography is suggested, video
Steganography refers to being hidden in secret information to be communicated in carrier video by some way, while not influencing regarding for vector properties
Feel effect and use function, to achieve the purpose that secret communication and conceal secret information this behavior itself.
Video information steganography is primarily upon at 3 points：Finding suitable embedded location using the data feature of carrier will believe
Breath is hidden into carrier, while not influencing the vision perception characteristic of carrier；Data medium subjective vision feature and statistics are not influenced
In the case of feature, utmostly can embedding information data hidden capacity；By containing information low error rate the case where
The ability of lower recovery.
Existing achievement in research is put forward according on the basis of information embedding operation position in an encoding process,
It is broadly divided into following three classes：
(1) embedded in uncompressed source code basin, refer to the pixel value of original image modify to
Embedded secret information, then using compression transmission.Embedding operation can be directly applied to original, pixel domain or certain transform domains
In.The algorithm advantage is to realize that water mark method is more, and algorithm is simply ripe.The disadvantage is that：After video compression coding is handled, it can make
It at the loss of part watermark information, makes troubles to the extraction and detection of watermark, increases bit rate, reduce video quality.
(2) the compression domain insertion after encoding refers to before being compressed after coding in such as encoding and decoding of some intermediate variables
The motion vector that will produce in journey, DCT coefficient, prediction mode etc. are embedded in when embedded using these variables as embedded put,
This kind of algorithm is a kind of algorithm of most study in video steganography.Its key is how the insertion of control secret information, makes it
There is stronger robustness and lower algorithm complexity again while meeting vision invisibility.Finally the steganography of video is calculated
Method, which is melted into video coding system, reaches system global optimization.The advantages of steganographic algorithm is in the only embedded DCT coefficient of watermark, no
Data bit-rate can be increased.The disadvantage is that there are error accumulation, embedded data volume is low.
(3) embedded in the code stream domain of compression video, it is a kind of rear-mounted insertion, refers to that secret information is embedded in warp knit
In the compressed video code flow of code.The algorithm advantage is need not to decode and re-encode completely process, is avoided since video is compiled
Embedded information during code and the problem of cause video quality to decline, improve information insertion ability and extraction efficiency.Disadvantage
It is video system to the constraint of compression bit rate and the selection of effective embedded space is very limited will substantially reduce embedding capacity.Its
Secondary, the continuous distortion drift that algorithm is likely to occur can reduce video quality, influence the invisibility of video steganography.
Invention content
The object of the present invention is to provide a kind of reduction robustness video steganography methods that H.264 interframe drift is distorted, and solve
Video quality caused by being embedded in the prior art due to data is declined, uneasy during Data Embedding Capacity is insufficient and transmission process
The problem of full property.
The technical solution adopted in the present invention is to reduce the robustness video steganography method of H.264 interframe drift distortion, tool
Body is implemented according to the following steps：
Step 1, interframe drift distortion is analyzed, obtains drift distortion Producing reason, then drift about and lose to caused by
It is really classified according to sizes, the embedded location of DCT luminance factors is selected according to the size of embedded drift distortion；
Step 2, to information sequence M=(m to be embedded1m2...mn) carry out convolutional encoding after output sequence be M '=
(m′1m′2...m′2n), wherein each input symbols corresponds to two output symbols, then information sequence output sequence to be embedded
The expression of M ' is as follows：
C=MG∞ (3)
In formula (3), G∞For generator matrix；
Step 3, the output sequence M ' of step 2 is embedded into the DCT coefficient of step 1, after the completion of embedded, to H.264/
AVC video flowings carry out entropy coding and form the close video sequence of load；
Step 4, to through step 3 treated H.264/AVC video flowing carry out entropy decoding, extract original DCT coefficient C, obtain
The output sequence M ' after convolutional encoding is taken, to output sequence M ' into row decoding, exports information sequence M to be embedded, completion waits for embedding
The extraction of the information sequence M entered.
The features of the present invention also characterized in that
H.264/AVC video flowing includes at least a P piece by I frames, B frames and P group of picture at, P frame images, if P pieces by
Dry P macro block and several I macro blocks form.
The detailed process of step 1 is as follows：
Interframe drift distortion is analyzed by formula (1), obtains drift distortion Producing reason：
In formula (1), De(n) it is that the insertion that hiding information generates in n-th of video frame quantization DCT coefficient is distorted, p n-th
The probability of hiding information in a video frame, a α constant related with video content,To possess the residual error of embedding data
Reconstructed value,For the residual error reconstructed value not comprising embedding data；
It drifts about to be distorted to caused by by formula (2) and be classified according to sizes, form DCT coefficient matrix C, wherein C=
C1UC2UC3, the embedded location of DCT luminance factors is selected further according to the size of embedded drift distortion：
In formula (2), C1, C2, C3For the set of 4 × 4 macro 16 coefficients classification in the block, ρijFor modification P 4 × 4DCT of frame systems
Independent distortion, r " caused by numberijFor carry privacy information reconstructive residual error block error amount, r 'ijNot carry privacy letter
The error amount of the reconstructive residual error block of breath, QstepFor quantization step, α is a constant related with video content, and i, j are DCT systems
Several positions, z 'ijIt is the coefficient value at i, j for position.
The detailed process of step 3 is as follows：
Step 3.1, in built-in end, to carrier, H.264/AVC video flowing is handled as unit of piece, to H.264/AVC
Video flowing carries out entropy decoding, obtains DCT coefficient matrix C, is scanned with Z-shaped, obtains 16 binary systems being made of 0 and 1
Sequence further obtains DCT coefficient matrix；
Step 3.2, judge whether current slice is P pieces, if current slice is P pieces, luminance macroblock therein is chosen, current
The embedded location of private data to be embedded is selected in 4 × 4DCT coefficient matrixes of macro block；
Step 3.3, the information embedding operation of H.264/AVC video flowing is carried out to the embedded location selected through step 3.2；
Step 3.1 is specifically implemented according to the following steps：
Step 3.1.1, tables look-up according to Coeff_token and NC, obtains nonzero coefficient number TotalCoeffs and hangover is
Keep count of TrailingOnes；
Step 3.1.2 obtains hangover coefficient number by the hangover number of coefficients of step 3.1.1, and input hangover coefficient corresponds to
Symbolic coding code stream, the hangover coefficient value exported；
Step 3.1.3 tables look-up to obtain prefix LevelPrefix, suffix lengths SuffixLength, root according to input code flow
LevelCode is obtained according to prefix and suffix, is expressed as formula (7), level is formula (8) if LevelCode is even number, if
LevelCode is that then level is formula (9) to odd number, then the threshold value by setting determines whether to update Suffixlegth；
LevelCode=(levelprefix<<suffixlength)+levelsuffix (7)
Level=(level+2)/2 (8)
Level=(- level-1)/2 (9)；
Step 3.1.4 tables look-up to obtain TotalZeros according to TotalCoeffs and input code flow, initializes i=
TotalCoeffs-1, zeroleft=TotalZeros parse the number of nonzero coefficient leading zero；Decoding finishes, by remaining member
Plain 0 polishing, inverted sequence arrange to obtain DCT coefficient matrix C；
Step 3.2 is specifically implemented according to the following steps：
Step 3.2.1 after carrying out dct transform to the DCT coefficient matrixes of original P frame luminance macroblocks, obtains original not measuring
The DCT coefficient matrix Y of change：
In formula (10), R is the residual matrix by dct transform, CiFor dct transform matrix；
Original non-quantized DCT coefficient matrix Y is scaled again again and re-quantization, the DCT coefficient after being quantified
Matrix Z：
Z=round [Yg×PFg/Qstep] (11)
In formula (11), QstepFor quantization step, PF is quantization matrix；
Step 3.2.2 carries out entropy coding to the DCT coefficient matrix Z after the quantization that is handled through step 3.2.1；In coding
Feedback stage by can inverse quantization ICT, obtain the luminance factor matrix of not no embedding dataThe DCT for obtaining luminance block again is residual
Poor coefficient matrix
In formula (13), CiFor dct transform matrix；
Step 3.2.3, if after completing data insertion, corresponding 4 × 4 quantization DCT luminance factor squares are obtained by formula (14)
Battle array Z '：
Z '=Z+ Δs (14)
In formula (14), Δ=(δij)4×4For the DCT coefficient matrix Z ' after embedded private data and the DCT coefficient after quantization
Corresponding error matrix between matrix Z；
Step 3.2.4 obtains 4 × 4 luminance DCT coefficients matrixes with embedding data by formula (14)
4 × 4 luminance DCT coefficients matrixes with embedding data are obtained by formula (15)Corresponding error matrix
Step 3.2.5 calculates the 4 × 4DCT coefficient matrixes with embedding data obtained by step 3.2.4It is corresponding
Residual matrixThe not reconstructive residual error block of embedding dataBetween pixel luminance errors E：
Step 3.2.6 does not consider the rounding operation to DCT coefficient, using mean square error MSE formula (2) calculate due to
It is independently distorted ρ caused by modification P frame 4 × 4DCT coefficient matrixesij；
In formula (2),By formula (2) to 4 × 4DCT luminance factor squares
The distortion that is separately embedded caused by battle array is classified：
Obtain individual distortion ρijSummation is equal to whole be distorted
Step 3.3 is specifically implemented according to the following steps：
The information sequence output sequence M ' to be embedded that step 2 is handled is embedded into step by step 3.3.1 in the current frame
In 3.2 embedded locations chosen, pre-defined mapping ruler S is：
Step 3.3.2, in untreated DCT coefficient matrix, if there are nonzero coefficient, on nonzero coefficient not into
Row data embedding operation modifies to nonzero coefficient after eliminating ambiguity, completes data insertion：
In formula (21), C ' is modified DCT coefficient；
Step 3.3.3 records currently embedded information bit quantity in private data telescopiny to be embedded
sum；
If sum<2n, wherein 2n are embedded in total information bit quantity, and next in H.264/AVC video flowing is waited for
The piece of processing is as current slice, if the current slice is I pieces, without any processing, direct coding；If the current slice is P pieces,
Return to step 3.3.1 is continued to execute, and is completed until all private data bits to be embedded are embedded in.
Step 4 is specifically implemented according to the following steps：
Step 4.1, entropy decoding is carried out to H.264/AVC video flowing in decoding end, obtains the prediction of current slice P frame luminance blocks
Information quantifies the later DCT coefficient C ' of the modification obtained from receiving terminal, extracts original DCT luminance factors C：
In formula (22), C ' is the later DCT coefficient of modification obtained from receiving terminal；
Step 4.2, output sequence is extracted according to the later coefficient C ' of step 4.1 modification that treated obtains from receiving terminal
M ' is arranged, if M'<Sum then continues with recovery output sequence by next as current slice：
Step 4.3, the output sequence M ' handled to step 4.2 using Viterbi decoding is exported to be embedded into row decoding
Information sequence M.
The invention has the advantages that
(1) a kind of reduction of present invention robustness video steganography method that H.264 interframe drift is distorted, reduces due to data
Interframe drift distortion, improves video quality caused by embedded；
(2) a kind of reduction of present invention robustness video steganography method that H.264 interframe drift is distorted, by being embedded in information
Improvement, can under the premise of not influencing visual quality per bit be embedded in 2 bit datas, improve Data Embedding Capacity；
(3) a kind of reduction of present invention robustness video steganography method that H.264 interframe drift is distorted, when secret information inputs
By carrying out convolutional encoding to it, anti-attack ability of the video information in transmission process is improved, video robustness is improved.
Description of the drawings
Fig. 1 is the information embedding operation stream that the present invention reduces the H.264 robustness video steganography method of interframe drift distortion
Cheng Tu；
Fig. 2 is the DCT luminance factors selection that the present invention reduces the H.264 robustness video steganography method of interframe drift distortion
Distribution map；
Fig. 3 is the C that the present invention reduces the H.264 robustness video steganography method of interframe drift distortion1、C2、C3Set is whole
The comparison diagram of distortion；
Fig. 4 is the convolutional encoding schematic diagram that the present invention reduces the H.264 robustness video steganography method of interframe drift distortion；
Fig. 5 is that the present invention reduces the original image for the robustness video steganography method that H.264 interframe drift is distorted and carries close
The comparison diagram of image；
Fig. 6 present invention reduces the information extraction flow chart of the H.264 robustness video steganography method of interframe drift distortion.
Specific implementation mode
The following describes the present invention in detail with reference to the accompanying drawings and specific embodiments.
H.264/AVC video is mainly by I frames, B frames and P group of picture at due to considering interframe drift distortion, so selection
Insertion frame be P frames.P frame images include at least a P piece, and P pieces are made of several P macro blocks and several I macro blocks.P macro blocks
Coding mode be divided into SKIP, Inter_16 × 16, Inter_16 × 8, Inter_8 × 16 and the classes of P8 × 8 five.When P macro blocks
When being encoded using the coding modes of Inter_16 × 16, P macro blocks are divided into 1 16 × 16 moving sub-block；When P macro blocks use
When the coding modes of Inter_16 × 8 are encoded, P macro blocks are divided into 2 16 × 8 moving sub-blocks；When P macro blocks use Inter_
When 8 × 16 coding modes are encoded, P macro blocks are divided into 28 × 16 moving sub-blocks；When P macro blocks encode mould using P8 × 8
When formula is encoded, P macro blocks are divided into 48 × 8 moving sub-blocks.8 × 8 sub-block have Inter_8 × 8, Inter_4 ×
8, Inter_8 × 4, Inter_4 × 4 four kinds of coding mode.When 8 × 8 sub-blocks are encoded using the coding modes of Inter_8 × 8
When, 8 × 8 sub-blocks are divided into 18 × 8 moving sub-block；When 8 × 8 sub-blocks are encoded using the coding modes of Inter_8 × 4
When, 8 × 8 sub-blocks are divided into 28 × 4 moving sub-blocks；When 8 × 8 sub-blocks are encoded using the coding modes of Inter_4 × 8
When, 8 × 8 sub-blocks are divided into 24 × 8 moving sub-blocks；When 8 × 8 sub-blocks are encoded using the coding modes of Inter_4 × 4
When, 8 × 8 sub-blocks are divided into 44 × 4 moving sub-blocks.Each 4 × 4 luminance block change quantization obtains one 4 × 4 later
DCT (Discrete Cosine Transform) coefficient matrix, DCT [i] [j], 0≤i≤3,0≤j≤3, DCT [0] [0] are
Lowest frequency coefficient, DCT [3] [3] are most high frequency coefficient.DCT coefficient matrix is along its upper left corner to the frequency of lower right corner diagonal coefficient
Rate is from low to high.The matrix that the coding mode of final P macro blocks forms 4 × 4 indicates.
The present invention reduces the robustness video steganography method of H.264 interframe drift distortion, is specifically implemented according to the following steps：
Step 1, by the analysis to interframe drift distortion, drift distortion Producing reason, then the drift to caused by are obtained
Distortion is classified according to sizes, and the embedded location of DCT coefficient is selected according to the size of embedded drift distortion；
Interframe drift distortion is analyzed by formula (1), obtains drift distortion Producing reason：
In formula (1), De(n) it is that the insertion that hiding information generates in n-th of video frame quantization DCT coefficient is distorted, p n-th
The probability of hiding information in a video frame, a α constant related with video content.To possess the residual error of embedding data
Reconstructed value,For the residual error reconstructed value not comprising embedding data；
As formula (1) it can be seen that embedded distortion is the distortion α D as caused by being embedded in a upper video framee(n-1) and it is current
The distortion of video frame insertionWhat addition obtained, generation distortion drift can be obtained by formula (1)
The reason of it is as follows：
(1) distortion drift will increase whole embedded distortion in frame caused by being embedded in data in each frame.
(2) with the increase of number of frames, the embedded distortion that each frame generates can gradually be accumulated, and next frame data are influenced
It is embedded, generate interframe distortion drift；
Two above hypothesis shows interframe drift distortion Producing reason；
It is learnt by formula (2), the drift distortion that different embedded locations generates is of different sizes, and the drift distortion to caused by is pressed
Size is classified, and C, C=C are denoted as1UC2UC3Wherein C1, C2, C3Represent the set of 4 × 4 macro 16 coefficients classification in the block；
Then select suitable DCT coefficient position as embedded location, formula (2) according to the embedded size being independently distorted generated
It is as follows：
In formula (2), ρijTo be independently distorted caused by modification P frame 4 × 4DCT coefficients, r "ijTo carry privacy information
The error amount of reconstructive residual error block, r 'ijFor do not carry privacy information reconstructive residual error block error amount, QstepFor quantization step, α is
A constant related with video content, i, j are the position of DCT coefficient, z 'ijIt is the coefficient value at i, j for position；
Step 2, to information sequence M=(m to be embedded1m2...mn) carry out convolutional encoding after output sequence be M '=
(m′1m′2...m′2n), wherein each input symbols corresponds to two output symbols, then the expression of output sequence M ' is as follows：
M '=MG∞ (3)
In formula (3), G∞For generator matrix；
Step 3, the output sequence M ' of step 2 is embedded into the DCT coefficient of step 1, after the completion of embedded, to H.264/
AVC video flowings carry out entropy coding and form the close video sequence of load；
Step 3.1, in built-in end, to carrier, H.264/AVC video flowing is handled as unit of piece, to H.264/AVC
Video flowing carries out entropy decoding, obtains DCT coefficient matrix C, is scanned with zigzag, obtains 16 binary systems being made of 0 and 1
Sequence further obtains DCT coefficient matrix；
Step 3.1.1, tables look-up according to Coeff_token and NC, obtains nonzero coefficient number TotalCoeffs and hangover is
Keep count of TrailingOnes；
Step 3.1.2 obtains hangover coefficient number by the hangover number of coefficients of step 3.1.1, and input hangover coefficient corresponds to
Symbolic coding code stream, the hangover coefficient value exported；
Step 3.1.3, suffix lengths SuffixLength table look-up to obtain prefix LevelPrefix, root according to input code flow
LevelCode is obtained according to prefix and suffix, is expressed as formula (7), level is formula (8) if LevelCode is even number, if
LevelCode is that then level is formula (9) to odd number, is determined whether to update Suffixlegth by the threshold value set；
LevelCode=(levelprefix<<suffixlength)+levelsuffix (7)
Level=(level+2)/2 (8)
Level=(- level-1)/2 (9)；
Step 3.1.4 tables look-up to obtain TotalZeros according to TotalCoeffs and input code flow, initializes i=
TotalCoeffs-1, zeroleft=TotalZeros parse the number of nonzero coefficient leading zero；Decoding finishes, by remaining member
Plain 0 polishing, inverted sequence arrangement can be obtained by DCT coefficient matrix；
Step 3.2, judge whether current slice is P pieces, if current slice is P pieces, choose luminance macroblock therein, current macro
The embedded location of private data to be embedded is selected in 4 × 4DCT coefficient matrixes of block；
Step 3.2.1 after carrying out dct transform to the DCT coefficient matrixes of original P frame luminance macroblocks, obtains original not measuring
The DCT coefficient matrix Y of change：
In formula (10), Y is original non-quantized DCT coefficient matrix, and R is the residual matrix by dct transform, CiFor DCT
Transformation matrix；
Original non-quantized DCT coefficient matrix Y is scaled again again and re-quantization, the DCT coefficient after being quantified
Matrix Z, expression formula are：
Z=round [Y·×PF·/Qstep] (11)
In formula (11), Z is the DCT coefficient matrix after quantization, QstepFor quantization step, PF is quantization matrix, × ,/be
The XOR operation of matrix；
Step 3.2.2 carries out entropy coding, in the feedback rank of coding to the DCT coefficient matrix Z after quantifying in step 3.2.1
Section by can inverse quantization ICT, obtain the DCT coefficient matrix of not no embedding dataThe reconstructive residual error of not embedding data is obtained again
Block
If step 3.2.3 obtains corresponding 4 × 4DCT coefficient matrixes Z ' after completing data insertion by formula (14)：
Z '=Z+ Δs (14)
In formula (14), Δ=(δij)4×4For the DCT coefficient matrix Z ' after embedded private data and the DCT coefficient after quantization
Corresponding error matrix between matrix Z；
4 × 4DCT coefficient matrixes with embedding data are calculated by formula (14) by step 3.2.4
4 × 4DCT coefficient matrixes with embedding data are calculated by formula (15)Corresponding residual matrix
Step 3.2.5 calculates the 4 × 4DCT coefficient matrixes with embedding data obtained by step 3.2.4It is corresponding residual
Poor matrixThe not reconstructive residual error block of embedding dataBetween pixel luminance errors E：
Step 3.2.6 does not consider the rounding operation to DCT coefficient, is calculated using mean square error MSE due to changing P frames 4
It is independently distorted ρ caused by × 4DCT coefficient matrixesij, i.e. formula (2)；
In formula (2), due toTherefore, it is possible to 4 × 4DCT coefficient matrixes
The caused distortion that is separately embedded is classified：
In formula (18), C1、C2、C3Represent the set of 16 DCT coefficients classification of 4 × 4 macro blocks；
Since the transformation of 4 × 4 macro blocks is linear, independent distortion ρ can be derived thatijSummation is equal to whole be distorted
From the foregoing, it will be observed that when the drift distortion that data insertion generates can classify, descending DCT brightness system
Manifold is combined into C1, C2, C3, so in each embedding data, the C of all P frames luminance blocks is selected first1Set is embedded in, when
C1When DCT coefficient off-capacity in set, then C is considered successively2, C3DCT coefficient in set is embedded in, using such
Embedding strategy can reduce the generation of drift distortion, improve video quality；
To verify above-mentioned conclusion, surveyed according to H.264/AVC standard encoding and decoding software JM8.5 in metlab2014a
Examination.Test video Foreman is 100 frames, and coding frame per second is 30fps, and quantization parameter QP is 28, the gop structure of document No.
For IPPP；The rearwards continuous P frames insertion successively since first p frame of each GOP, in the DCT luminance factors of each P frames
Embedded message digit.It is embedded in area in data, belongs to C1, C2And C3DCT luminance factors be respectively 2810,12230 and 14646, respectively
Belonging to C1, C2And C3DCT luminance factors in embedded 2000 message digits, generate three carrier videos, and generate such as Fig. 3 institutes
The whole distortion shown, it can be seen that embed data into C1In coefficient can introduce minimum distortion；
Step 3.3, the H.264/AVC information embedding operation of video flowing；
Output sequence M ' in step 3.3 is embedded into the insertion chosen in step 3.2 by step 3.3.1 in the current frame
In position, pre-defined insertion mapping ruler S is：
In formula (20), m '1, m '2For the output sequence after convolutional encoding；
Step 3.3.2, in untreated DCT coefficient matrix, if there are nonzero coefficient, on nonzero coefficient not into
Row data embedding operation modifies to nonzero coefficient after eliminating ambiguity, completes data insertion：
In formula (21), C ' is modified DCT coefficient；
Step 3.3.3 records currently embedded information bit quantity in private data telescopiny to be embedded
sum；
If sum<2n, wherein 2n are embedded in total information bit quantity, and next in H.264/AVC video flowing is waited for
The piece of processing is as current slice, if the current slice is I pieces, without any processing, direct coding；If the current slice is P pieces,
Return to step 3.3 continues to execute, and is completed until all secret information bits are embedded in；
The present invention is tested according to H.264/AVC standard encoding and decoding software JM18.5, and each test video is encoded to 300
A frame, coding frame per second are 30fps, and quantization parameter QP is 28, and gop structure when document No. is IPPPPPPPPP；Each video
Sequence only has first frame to be encoded to I frames, remaining is P frames coding, in testing, using 3 segment standard video test sequences, respectively
For：Foreman, mobile, akiyo.
Table 1 is the comparison of PSNR under different Q P, reflects the objective standard of measurement video quality after embedding data；It is different
QP values corresponding to embedding capacity and PSNR.As can be drawn from Table 1：QP is smaller, and the visual quality of the higher videos of PSNR is just
Better, drift distortion is just smaller；The more complicated video image of the more violent image of movement, PSNR is relatively low, and video quality is
Decline；
Table 1
Table 2 is the comparison of embedding capacity under different Q P, it can be deduced that：Embedding capacity depends on the value of video content and QP,
In the case of QP is identical, the object of which movement more intense environment in video image is more complicated, and non-zero quantised DCT coefficient is more, embedded
Capacity is smaller；With the increase of QP, zero DCT coefficient more quantified will produce at this time, the embedded location of this paper is exactly in zero amount
Change in DCT coefficient, so embedding capacity is consequently increased, the proportional relationship of the two；
Table 2
As shown in figure 5, listing the image of three the 20th frames of test video, reflect that the present invention is carrying in terms of subjective quality
Under the premise of high embedding capacity, the visual quality of video is not interfered with, can be seen that inventive algorithm from the comparison of hypograph
Maintain the clarity before and after visual quality, the variation for the video quality that human eye can not be discovered.
Step 4, entropy decoding is carried out to H.264/AVC video flowing, extracts original DCT coefficient C, obtained defeated after convolutional encoding
Go out sequence M ', to output sequence M ' into row decoding, exports information sequence M to be embedded, complete carrying for information sequence M to be embedded
It takes；
Step 4.1, entropy decoding is carried out to H.264/AVC video flowing in decoding end, obtains the prediction of current slice P frame luminance blocks
Information quantifies the modification obtained from receiving terminal later DCT coefficient C ', extraction original DCT coefficient C：
In formula (22), C ' is the later DCT coefficient of modification obtained from receiving terminal；
Step 4.2, the output after obtaining convolutional encoding is extracted according to the later coefficient C ' of the modification obtained from receiving terminal
Sequence M ', if sum>2n then continues with recovery output sequence by next as current slice：
Step 4.3, using Viterbi decoding to output sequence M '=(m ' after obtaining convolutional encoding1m′2...m′2n) into
Row decoding exports information sequence M=(m to be embedded1m2...mn)。
Table 3 be different packet loss rate under robustness comparison, proposition under conditions of different packet loss rate, do not make any place
Reason, is separately added into the video sequence survival rate shown under conditions of Bose-Chaudhuri-Hocquenghem Code and convolutional code used herein, it can be seen that with
The increase survival rate for packet loss is reducing, compared to Bose-Chaudhuri-Hocquenghem Code present invention introduces convolutional code almost can completely restore
Original video information, although can have fraction error, the video restored is not interfered with and is regarded in human eye tolerance interval
Frequency visual quality.In conjunction with the embedded mobile GIS in the present invention, it can realize that a drift distortion is small, embedding capacity is big, robustness
Strong video steganographic algorithm；
Table 3
The present invention reduces the robustness video steganographic algorithm of interframe drift distortion, passes through the optimization choosing to embedded location coefficient
It selects, reduces interframe drift distortion, improve video quality；It is being passed by carrying out convolutional encoding to secret information to improve video data
The ability of resistance passive attack during defeated；By the improvement to embedded mobile GIS, under the premise of not influencing video quality, carry
High embedding capacity.
Claims (8)
1. reducing the robustness video steganography method of H.264 interframe drift distortion, which is characterized in that specifically real according to the following steps
It applies：
Step 1, interframe drift distortion is analyzed, obtains drift distortion Producing reason, then drift distortion is pressed to caused by
Size is classified, and the embedded location of DCT luminance factors is selected according to the size of embedded drift distortion；
Step 2, to information sequence M=(m to be embedded1m2...mn) carry out convolutional encoding after output sequence be M '=(m '1m
′2...m′2n), wherein each input symbols corresponds to two output symbols, then the table of information sequence output sequence M ' to be embedded
Show as follows：
C=MG∞ (3)
In formula (3), G∞For generator matrix；
Step 3, the output sequence M ' of step 2 is embedded into the DCT coefficient of step 1, after the completion of embedded, to H.264/AVC regarding
Frequency stream carries out entropy coding and forms the close video sequence of load；
Step 4, to through step 3 treated H.264/AVC video flowing carry out entropy decoding, extract original DCT coefficient C, obtain volume
Output sequence M ' after product coding exports information sequence M to be embedded, completes to be embedded to output sequence M ' into row decoding
The extraction of information sequence M.
2. the reduction according to claim 1 robustness video steganography method that H.264 interframe drift is distorted, feature exist
In the H.264/AVC video flowing is by I frames, B frames and P group of picture at the P frames image includes at least a P piece, the P
Piece is made of several P macro blocks and several I macro blocks.
3. the reduction according to claim 2 robustness video steganography method that H.264 interframe drift is distorted, feature exist
In the detailed process of step 1 is as follows：
Interframe drift distortion is analyzed by formula (1), obtains drift distortion Producing reason：
In formula (1), De(n) it is that the insertion that hiding information generates in n-th of video frame quantization DCT coefficient is distorted, p is n-th of video
The probability of hiding information in frame, a α constant related with video content,Residual error to possess embedding data is rebuild
Value,For the residual error reconstructed value not comprising embedding data；
It drifts about to be distorted to caused by by formula (2) and be classified according to sizes, form DCT coefficient matrix C, wherein C=C1∪C2∪
C3, the embedded location of DCT luminance factors is selected further according to the size of embedded drift distortion：
In formula (2), C1, C2, C3For the set of 4 × 4 macro 16 coefficients classification in the block, ρijFor modification P 4 × 4DCT of frame coefficients institute
The independent distortion of generation, r "ijFor carry privacy information reconstructive residual error block error amount, r 'ijNot carry privacy information
The error amount of reconstructive residual error block, QstepFor quantization step, α is a constant related with video content, and i, j are DCT coefficient
Position, z 'ijIt is the coefficient value at i, j for position.
4. the reduction according to claim 3 robustness video steganography method that H.264 interframe drift is distorted, feature exist
In the detailed process of step 3 is as follows：
Step 3.1, in built-in end, to carrier, H.264/AVC video flowing is handled as unit of piece, to H.264/AVC video
Stream carries out entropy decoding, obtains DCT coefficient matrix C, is scanned with Z-shaped, obtains 16 binary sequences being made of 0 and 1,
Further obtain DCT coefficient matrix；
Step 3.2, judge whether current slice is P pieces, if current slice is P pieces, luminance macroblock therein is chosen, in current macro
4 × 4DCT coefficient matrixes in select the embedded location of private data to be embedded；
Step 3.3, the information embedding operation of H.264/AVC video flowing is carried out to the embedded location selected through step 3.2.
5. the reduction according to claim 4 robustness video steganography method that H.264 interframe drift is distorted, feature exist
In step 3.1 is specifically implemented according to the following steps：
Step 3.1.1, tables look-up according to Coeff_token and NC, obtains nonzero coefficient number TotalCoeffs and hangover coefficient number
Mesh TrailingOnes；
Step 3.1.2 obtains hangover coefficient number, the corresponding symbol of input hangover coefficient by the hangover number of coefficients of step 3.1.1
Number encoding code stream, the hangover coefficient value exported；
Step 3.1.3 tables look-up to obtain prefix LevelPrefix, suffix lengths SuffixLength, according to preceding according to input code flow
Sewing and obtain LevelCode with suffix, is expressed as formula (7), level is formula (8) if LevelCode is even number, if
LevelCode is that then level is formula (9) to odd number, then the threshold value by setting determines whether to update Suffixlegth；
LevelCode=(levelprefix<<suffixlength)+levelsuffix (7)
Level=(level+2)/2 (8)
Level=(- level-1)/2 (9)；
Step 3.1.4 tables look-up to obtain TotalZeros according to TotalCoeffs and input code flow, initializes i=
TotalCoeffs-1, zeroleft=TotalZeros parse the number of nonzero coefficient leading zero；Decoding finishes, by remaining member
Plain 0 polishing, inverted sequence arrange to obtain DCT coefficient matrix C.
6. the reduction according to claim 4 robustness video steganography method that H.264 interframe drift is distorted, feature exist
In step 3.2 is specifically implemented according to the following steps：
Step 3.2.1 is obtained original non-quantized after carrying out dct transform to the DCT coefficient matrix of original P frame luminance macroblocks
DCT coefficient matrix Y：
In formula (10), R is the residual matrix by dct transform, CiFor dct transform matrix；
Original non-quantized DCT coefficient matrix Y is scaled again again and re-quantization, the DCT coefficient matrix after being quantified
Z：
Z=round [Yg×PFg/Qstep] (11)
In formula (11), QstepFor quantization step, PF is quantization matrix；
Step 3.2.2 carries out entropy coding to the DCT coefficient matrix Z after the quantization that is handled through step 3.2.1；In the feedback of coding
Stage by can inverse quantization ICT, obtain the luminance factor matrix of not no embedding dataThe DCT residual error coefficients of luminance block are obtained again
Matrix
In formula (13), CiFor dct transform matrix；
Step 3.2.3, if after completing data insertion, corresponding 4 × 4 quantization DCT luminance factor matrixes Z ' is obtained by formula (14)：
Z '=Z+ Δs (14)
In formula (14), Δ=(δij)4×4For the DCT coefficient matrix Z ' and DCT coefficient matrix Z after quantization after embedded private data
Between corresponding error matrix；
Step 3.2.4 obtains 4 × 4 luminance DCT coefficients matrixes with embedding data by formula (14)
4 × 4 luminance DCT coefficients matrixes with embedding data are obtained by formula (15)Corresponding error matrix
Step 3.2.5 calculates the 4 × 4DCT coefficient matrixes with embedding data obtained by step 3.2.4Corresponding residual error square
Battle arrayThe not reconstructive residual error block of embedding dataBetween pixel luminance errors E：
Step 3.2.6 does not consider the rounding operation to DCT coefficient, is calculated using mean square error MSE formula (2) due to changing P
It is independently distorted ρ caused by frame 4 × 4DCT coefficient matrixesij；
In formula (2),By formula (2) to 4 × 4DCT luminance factors matrix institute
Caused by be separately embedded distortion and classify：
Obtain individual distortion ρijSummation is equal to whole be distorted
7. the reduction according to claim 6 robustness video steganography method that H.264 interframe drift is distorted, feature exist
In step 3.3 is specifically implemented according to the following steps：
The information sequence output sequence M ' to be embedded that step 2 is handled is embedded into step 3.2 by step 3.3.1 in the current frame
In the embedded location of selection, pre-defined mapping ruler S is：
Step 3.3.2, in untreated DCT coefficient matrix, if there are nonzero coefficient, without number on nonzero coefficient
It according to embedding operation, modifies to nonzero coefficient after eliminating ambiguity, completes data insertion：
In formula (21), C ' is modified DCT coefficient；
Step 3.3.3 records currently embedded information bit quantity sum in private data telescopiny to be embedded；
If sum<2n, wherein 2n are the information bit quantity being embedded in total, will be next pending in H.264/AVC video flowing
Piece as current slice, if the current slice be I pieces, without any processing, direct coding；If the current slice is P pieces, return
Step 3.3.1 is continued to execute, and is completed until all private data bits to be embedded are embedded in.
8. the reduction according to claim 4 robustness video steganography method that H.264 interframe drift is distorted, feature exist
In step 4 is specifically implemented according to the following steps：
Step 4.1, entropy decoding is carried out to H.264/AVC video flowing in decoding end, obtains the prediction letter of current slice P frame luminance blocks
Breath quantifies the later DCT coefficient C ' of the modification obtained from receiving terminal, extracts original DCT luminance factors C：
In formula (22), C ' is the later DCT coefficient of modification obtained from receiving terminal；
Step 4.2, output sequence is extracted according to the later coefficient C ' of step 4.1 modification that treated obtains from receiving terminal
M ', if M'<Sum then continues with recovery output sequence by next as current slice：
Step 4.3, the output sequence M ' handled to step 4.2 using Viterbi decoding exports information to be embedded into row decoding
Sequence M.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810549353.5A CN108769700A (en) | 2018-05-31 | 2018-05-31 | Reduce the robustness video steganography method of H.264 interframe drift distortion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810549353.5A CN108769700A (en) | 2018-05-31 | 2018-05-31 | Reduce the robustness video steganography method of H.264 interframe drift distortion |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108769700A true CN108769700A (en) | 2018-11-06 |
Family
ID=64001243
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810549353.5A Pending CN108769700A (en) | 2018-05-31 | 2018-05-31 | Reduce the robustness video steganography method of H.264 interframe drift distortion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108769700A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109862395A (en) * | 2019-03-29 | 2019-06-07 | 中国人民解放军战略支援部队信息工程大学 | A kind of video flowing steganographic detection method and apparatus |
| CN110062242A (en) * | 2019-03-04 | 2019-07-26 | 中山大学 | A kind of H.264 video steganographic algorithm based on UED |
| CN111263165A (en) * | 2019-12-27 | 2020-06-09 | 郑州师范学院 | Embedding method and extracting method of HEVC video steganography information |
| CN112565779A (en) * | 2020-12-12 | 2021-03-26 | 四川大学 | Video steganography method based on distortion drift |
| CN115565114A (en) * | 2022-11-17 | 2023-01-03 | 中央广播电视总台 | Video picture drift identification method and device, computer equipment and storage medium |
| CN116320471A (en) * | 2023-05-18 | 2023-06-23 | 中南大学 | Video information hiding method, system, equipment and video information extracting method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103152578A (en) * | 2013-03-25 | 2013-06-12 | 中国传媒大学 | H.264 video watermark embedding and extraction method based on mixed coding/decoding |
| CN106101714A (en) * | 2016-07-14 | 2016-11-09 | 中国科学院信息工程研究所 | One and the tightly coupled H.264 video information hiding method of compression encoding process |
| CN107172434A (en) * | 2017-06-27 | 2017-09-15 | 郑州师范学院 | A kind of method that built-in steganography is carried out to H.265 video |
| CN107948655A (en) * | 2017-11-22 | 2018-04-20 | 中山大学 | A kind of H.264/AVC encrypted video steganography method based on CABAC |
-
2018
- 2018-05-31 CN CN201810549353.5A patent/CN108769700A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103152578A (en) * | 2013-03-25 | 2013-06-12 | 中国传媒大学 | H.264 video watermark embedding and extraction method based on mixed coding/decoding |
| CN106101714A (en) * | 2016-07-14 | 2016-11-09 | 中国科学院信息工程研究所 | One and the tightly coupled H.264 video information hiding method of compression encoding process |
| CN107172434A (en) * | 2017-06-27 | 2017-09-15 | 郑州师范学院 | A kind of method that built-in steganography is carried out to H.265 video |
| CN107948655A (en) * | 2017-11-22 | 2018-04-20 | 中山大学 | A kind of H.264/AVC encrypted video steganography method based on CABAC |
Non-Patent Citations (2)
| Title |
|---|
| YUANZHIYAO 等: "Inter-frame distortion drift analysis for reversible data hiding in encrypted H.264/AVC video bitstreams", 《SIGNAL PROCESSING 》 * |
| 王林等: "减小H.264帧间漂移失真的鲁棒性视频隐写算法", 《计算机系统应用》 * |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110062242A (en) * | 2019-03-04 | 2019-07-26 | 中山大学 | A kind of H.264 video steganographic algorithm based on UED |
| CN110062242B (en) * | 2019-03-04 | 2021-05-04 | 中山大学 | H.264 video steganography algorithm based on UED |
| CN109862395A (en) * | 2019-03-29 | 2019-06-07 | 中国人民解放军战略支援部队信息工程大学 | A kind of video flowing steganographic detection method and apparatus |
| CN109862395B (en) * | 2019-03-29 | 2021-05-04 | 中国人民解放军战略支援部队信息工程大学 | Video stream hidden information detection method and device |
| CN111263165A (en) * | 2019-12-27 | 2020-06-09 | 郑州师范学院 | Embedding method and extracting method of HEVC video steganography information |
| CN111263165B (en) * | 2019-12-27 | 2022-07-01 | 郑州师范学院 | Embedding method and extracting method of HEVC video steganography information |
| CN112565779A (en) * | 2020-12-12 | 2021-03-26 | 四川大学 | Video steganography method based on distortion drift |
| CN112565779B (en) * | 2020-12-12 | 2021-10-29 | 四川大学 | Video steganography method based on distortion drift |
| CN115565114A (en) * | 2022-11-17 | 2023-01-03 | 中央广播电视总台 | Video picture drift identification method and device, computer equipment and storage medium |
| CN115565114B (en) * | 2022-11-17 | 2023-03-21 | 中央广播电视总台 | Video picture drift identification method and device, computer equipment and storage medium |
| CN116320471A (en) * | 2023-05-18 | 2023-06-23 | 中南大学 | Video information hiding method, system, equipment and video information extracting method |
| CN116320471B (en) * | 2023-05-18 | 2023-08-22 | 中南大学 | Video information hiding method, system, equipment and video information extracting method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108769700A (en) | Reduce the robustness video steganography method of H.264 interframe drift distortion | |
| CN106101729B (en) | Logic frame mode name in HEVC Video coding | |
| CN103414892B (en) | The Image Hiding that a kind of Large Copacity is incompressible | |
| CN107040787B (en) | A kind of 3D-HEVC inter-frame information hidden method of view-based access control model perception | |
| CN102223540B (en) | Information hiding method facing to H.264/AVC (automatic volume control) video | |
| Liu et al. | A robust and improved visual quality data hiding method for HEVC | |
| CN101841700B (en) | H.264 compression bit stream oriented video watermark method | |
| CN104581176B (en) | The insertion of H.264/AVC compression domain robust video watermark and extracting method without frame in error drift | |
| CN107087183A (en) | The method decoded to intra prediction mode | |
| CN104994382B (en) | A kind of optimization method of perception rate distortion | |
| CN108366259A (en) | A kind of reversible steganography method of three coefficient of H.264/AVC video of high-fidelity | |
| CN106101714B (en) | A kind of and tightly coupled H.264 video information hiding method of compression encoding process | |
| CN105898300B (en) | A kind of improvement transformation coefficient sign bit hidden method based on recovery transformation coefficient | |
| CN108965887B (en) | Video information hiding method and device based on inter-block decoupling | |
| CN107710763A (en) | For method, coding/decoding method, device and the related computer program encoded to digital picture | |
| CN109819260A (en) | Video steganography method and device based on the fusion of multi-embedding domain | |
| Tian et al. | A semi-fragile video watermarking algorithm based on chromatic residual DCT | |
| Dong et al. | Multi-channel HEVC steganography by minimizing IPM steganographic distortions | |
| CN105681803B (en) | A kind of HEVC video information hiding methods of large capacity | |
| Ma et al. | A CNN-based image compression scheme compatible with JPEG-2000 | |
| CN109427032B (en) | Scalable video watermark embedding and extracting method based on compressed domain | |
| Ma et al. | A cross channel context model for latents in deep image compression | |
| Dong et al. | Adaptive HEVC steganography based on steganographic compression efficiency degradation model | |
| Cappellari et al. | Resolution scalable image coding with reversible cellular automata | |
| Tang et al. | Reversible data hiding based on a modified difference expansion for H. 264/AVC video streams |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181106 |