CN109819260A - Video steganography method and device based on the fusion of multi-embedding domain - Google Patents
Video steganography method and device based on the fusion of multi-embedding domain Download PDFInfo
- Publication number
- CN109819260A CN109819260A CN201910102368.1A CN201910102368A CN109819260A CN 109819260 A CN109819260 A CN 109819260A CN 201910102368 A CN201910102368 A CN 201910102368A CN 109819260 A CN109819260 A CN 109819260A
- Authority
- CN
- China
- Prior art keywords
- video
- frame
- steganography
- motion vector
- video frame
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Abstract
The present invention relates to a kind of video steganography methods and device based on the fusion of multi-embedding domain.This method is not limited to single video embedded domain and carries out the insertion of close breath, but merged motion vector field, three embedded domain in quantization parameter domain and intra-prediction carry out the insertion of close breath.For motion vector field part, the motion vector generated by motion estimation module in modification video compression coding frame is to implement steganography insertion；For quantization parameter domain, by modifying the quantization parameter of video encoding unit (macro block) to implement steganography insertion；For intra-prediction, the Intra4x4 intra prediction mode generated by intra-framed prediction module in modification video compression coding frame is to implement steganography insertion.The present invention is especially suitable for embedding capacity and the higher digital multimedia security scenario of steganography security requirement.
Description
Technical field
The present invention relates to a kind of video steganography (Video based on multi-embedding domain under H.264/AVC video encoding standard
Steganography) method, and in particular to merged one kind based on motion vector field, one kind based on quantization parameter domain and one
Video steganography method of the kind based on intra-prediction, can be used in digital medium safety field, belongs to field of information security technology
In Information hiding field.
Background technique
Steganography is one of the important means of current covert communication, and confidential information is hidden in ostensible content and passes by it
Defeated or save, the content of secrecy cannot be browsed by making opponent not only, it is also difficult to know that secret communication or secret storage are true
In the presence of.
Generally, it when designing steganographic algorithm, needs to comprehensively consider the following aspects factor: 1) not sentience:
Seeking the carrier before and after steganography is undistinguishable for human body, i.e., can not only sentence by the sense of hearing of the mankind and vision
Whether disconnected carrier passes through steganography；2) robustness: even if it is required that steganography file is by certain disturbance, such as lossy compression or transmission
Noise jamming etc. in the process, should be able to also restore secret information therein；3) embedding capacity: it is required that guaranteeing not sentience
To a certain degree under the premise of robustness, it should be embedded in secret information into carrier as much as possible；4) Embedding efficiency: it is required that
Under the premise of embedding information amount is certain, it should reduce the modification degree to carrier as far as possible；5) safety: it is required that even if possessing pair
The priori knowledge of the steganographic algorithm, existing steganalysis method also can not distinguish general carrier and warp with sufficiently high accuracy
Cross the file of the algorithm steganography.
In recent years, with the height interaction multimedias such as video on demand and live streaming media application rise and be widely current, number
Word video just gradually replaces image with its visual performance abundant and powerful information carrying capacity, becomes current entertainment industry
In communication media with strongest influence power.In addition, fast with technologies such as video compression coding, computer network and high-performance calculations
Speed development, digital video can be prepared very quickly and under conditions of keeping compared with high coding efficiency and eye fidelity in internet
Upper real time communication.Based on the above-mentioned fact, digital video is generally considered as a kind of ideal covert communications carrier, video steganography at present
The extensive concern for also therefore having attracted Information hiding area research person with video Steganalysis becomes the research heat in the field
One of point.
Generally, according to the difference of embedded domain, video steganography can be divided into spatial domain video steganography and compressed domain video steganography two
Major class.Most spatial domain video steganographic algorithms have used for reference the design philosophy of image latent writing field classic algorithm, by directly repairing
Change the original pixel value of video frame to be embedded in close breath, and using error-correcting code technique and by the way of repeating to be embedded in, enhances the close anti-view of breath
The robustness of frequency compressed encoding is to reduce the bit error rate.
The advantages of spatial domain video steganographic algorithm is that robustness is stronger, can restore it by a degree of interference
In secret information；The disadvantage is that algorithm type is fewer, and since secret information generally requires repetition insertion, insertion is held
It measures limited.
Compressed domain video steganography is by steganography embedding operation and video compression coding close-coupled, by video compression coding
Key modules in frame are disturbed, and modulation modification its generate bit stream syntax element to be embedded in close breath.It is embedding according to steganography
Enter domain, compressed domain video steganography can be further divided into: the steganography based on motion vector, is based on frame at the steganography based on transformation coefficient
The steganography of inner estimation mode, the steganography based on inter-frame forecast mode, the steganography based on quantization parameter, the steganography based on entropy coding.
The motion vector that video steganography based on motion vector generates motion estimation module in video compression coding frame
As close breath carrier, the research achievement in relation to such steganography is more.Conventional motion vector field video steganographic algorithm generallys use letter
Single screening rule, such as motion vector amplitude or phase angle, selected part motion vector carries out insertion modification.With conventional motion to
Domain video steganographic algorithm difference is measured, second generation motion vector field video steganographic algorithm reduces motion vector by using steganography code
Quantity is modified, to effectively improve Embedding efficiency and steganography safety.Third generation motion vector field video steganographic algorithm is using
On the basis of steganography code, also maintains as far as possible by the local optimum of modification motion vector, it is hidden that steganography is further enhanced with this
Property.
Video steganography based on transformation coefficient passes through modification brightness DCT (Discrete Cosine Transform) coefficient
To realize that steganography is embedded in, such algorithm usually has biggish embedding capacity.There is scholar's proposition, for using in H.264 video
Intra4x4 prediction mode carries out the 4x4 luminance block (abbreviation Intra4x4 intraframe predictive coding block) of predictive coding, can basis
The intra prediction mode of its adjacent block, selects the quantization DCT coefficient of some of them specific position to modify to implement steganography.Together
When, the distortion drift phenomenon generated there are also scholar to the modification of H.264 quantization DCT coefficient is analyzed, on this basis,
Using the quantization DCT coefficient of Intra4x4 intraframe predictive coding block in I frame as carrier, in conjunction with STC (Syndrome Trellis
Codes), content-adaptive DCT coefficient steganography insertion scheme is proposed, distortion drift phenomenon has been effectively relieved, and reached good
Good steganography safety.
Video steganography based on intra prediction mode will compress video in using intraframe predictive coding coding unit as
Close breath carrier, by modifying its intra prediction mode to implement steganography.For example, the steganographic algorithm for having scholar to propose is in telescopiny
In the quantity of intra prediction mode for needing to modify reduced by matrix coder, improve Embedding efficiency and steganography safety.
Video steganography based on inter-frame forecast mode will compress video in using inter prediction encoding coding unit as
Close breath carrier, by modifying its inter-frame forecast mode to implement steganography.Existing algorithm is generally basede on preset inter-frame forecast mode
Mapping ruler between binary bits (sequence) directly carries out the inter-frame forecast mode of macro block according to embedding close breath bit
Modulation modification.
As can be seen that the advantages of compressed domain video steganographic algorithm is that insertion selectivity is more, algorithm type is abundant.But there is also
Part relatively limited the disadvantages of causing embedding capacity to be often not so good as spatial domain video steganographic algorithm in embedded domain for modification.It is based on
The limitation of above-mentioned existing video steganography, the present invention can fill up the field of multi-embedding domain fusion to a certain extent
Blank, to enhance the technology completeness in the field.
Summary of the invention
It is an object of the present invention to be independent of each other according to the close breath of embedded domain insertion each in H.264/AVC compression video
Characteristic, design it is a kind of based on multi-embedding domain fusion video steganography method.
The present invention is compared to other video steganography methods, it is not limited to which single video embedded domain carries out the embedding of close breath
Enter, but merged motion vector field, three embedded domain in quantization parameter domain and intra-prediction carry out the insertion of close breath.For
Motion vector field part, the motion vector generated by motion estimation module in modification video compression coding frame is to implement steganography
Insertion；For quantization parameter domain, by modifying the quantization parameter of video encoding unit (macro block) to implement steganography insertion；For frame
Interior prediction domain, the Intra4x4 intra prediction mode generated by intra-framed prediction module in modification video compression coding frame is with reality
Apply steganography insertion.As it can be seen that method proposed by the present invention is different from previous any video steganography method, especially suitable for holding to insertion
Amount and the higher digital multimedia security scenario of steganography security requirement.
According to investigation, current overwhelming majority video steganography method fails active balance embedding capacity and steganography safety, table
It is present: steganography safety is sacrificed to increase embedding capacity, to compressed video bit stream specific syntax elements (such as move to
Amount, inter-frame forecast mode etc.) carry out steganography modification when, steganography safety will be adversely affected when load factor is excessive；Or
Embedding capacity is reduced in order to guarantee steganography safety.In addition, most video steganography methods are difficult to effectively keep original load
The attribute and coding parameter of volumetric video.For above-mentioned limitation, load factor, steganography safety and carrier video category are being comprehensively considered
Property the factors such as effectively keep in the case where, how to design preferably efficient video steganographic algorithm, be current digital multimedia safety
One of field urgent problem.
Specifically, The technical solution adopted by the invention is as follows:
A kind of video steganography method based on the fusion of multi-embedding domain, as shown in Figure 1, comprising the following steps:
1) video frame for needing to carry out steganography insertion is obtained；
2) according to selected steganography strategy, specified compression code element to be embedded in the frame is selected, and is advised according to mapping
Then obtain binary system carrier vector；
3) the specified compression code element to be embedded for each of the video frame, according to preset cost computation model
Obtain its steganography insertion cost；
4) it establishes private communication channel using steganography code and is embedded in secret information thereto, and according to the result modulation modification video
Correspondence compressed encoding element in frame；
5) step 2) is repeated to 4), which is encoded according to result, that is, is completed the frame and corresponded to secret information insertion；
6) remaining every frame is restarted to execute, until secret from step 1) according to the frame coded sequence of video file
Information, which has fully embedded into, to be finished, and final steganography video file is obtained.
Further, above-mentioned method further includes the extraction process of secret information, as shown in Figure 2, comprising the following steps:
1) it obtains and needs to carry out the video frame that close breath extracts；
2) according to selected steganography strategy, specified compression code element to be embedded in the frame is determined.It is double according to communication
Secret information bit in each specified embedded coding element of the Rule Extraction that side appoints in advance；
3) according to the frame decoding sequence of video file, step 1) is repeated to 2) until secret information is complete to remaining every frame
It extracts and finishes in portion.
On the basis of above scheme, the present invention is further improved, i.e., in the steganography for carrying out close breath using the present invention
When insertion, for single video frame, modify as follows to each single embedded domain.
[1] the video steganographic algorithm based on motion vector field
Video steganographic algorithm based on motion vector field is produced by motion estimation module in modification video compression coding frame
Raw motion vector is to implement steganography insertion.Specifically, when the algorithm is to motion vector MV0,0=(mvx,mvy) (wherein 0,0 table
Show current macro, mvx,mvyIndicate motion vector both horizontally and vertically) when modifying, successively execute following step
Select most suitable local optimum motion vector as final modification result from all candidate motion vectors suddenly:
A) it pre-processes.Respectively to MV0,0Four motion vector MV in contiguous range-1,0=(mvx-1,mvy)、MV1,0=
(mvx+1,mvy)、MV0,-1=(mvx,mvy- 1) and MV0,1=(mvx,mvy+ 1) local optimum sex determination is carried out；
B) best candidate set is constructed.By MV-1,0, MV1,0, MV0,-1And MV0,1In local optimum motion vector form wait
Motion vector set Ψ is selected, if Ψ non-empty, gos to step e), otherwise continues to execute step c)；
C) region of search delimited.The region of search for delimiting a motion vector, for any Candidate Motion in the region
Vector M V '=(mv 'x,mvy'), it is necessary to meet LSB (mvx+mvy)≠LSB(mvx′+mvy′)；Wherein, LSB indicates lowest bit
Position embedded mobile GIS；
D) best candidate set is constructed.For each candidate motion vector in the region of search, part is carried out most respectively
Dominance determines, determines the motion vector of wherein all local optimums, and their groups are combined into set Ψ；
E) local optimum motion vector is chosen.For each local optimum motion vector in Ψ, its rate distortion generation is calculated
Valence selects the local optimum motion vector for possessing minimum rate distortion costs as final modification result.
[2] the video steganographic algorithm based on quantization parameter
Video steganographic algorithm based on quantization parameter is by the quantization parameter of modification video encoding unit (macro block) to implement
Steganography insertion.Specifically, it when the quantization parameter of algorithm macro block in certain video frame carries out steganography insertion, successively executes following
Step:
A) carrier vector determines.Present frame is encoded, the quantization parameter of current macro is obtained, judges whether it is 0, if it is 0,
Next macro block is skipped to, otherwise, continues to execute step b)；
B) it pre-processes.Judge current macro quantization parameter last bit whether with the current bit position to embedding secret information
Match.Matching then keeps quantization parameter constant, otherwise, execute step c)；
C) it is embedded in.The operation of plus-minus one is carried out to the quantization parameter of current macro, while guaranteeing the number of quantization parameter
Value is consistently less than 51.
D) video frame next code.The subsequent coding steps for continuing to execute the video frame export gained steganography video frame
To code stream.
[3] the video steganographic algorithm based on intra prediction mode
Video steganographic algorithm based on intra prediction mode passes through intra-framed prediction module in modification video compression coding frame
The Intra4x4 intra prediction mode of generation is to implement steganography insertion.The algorithm is in telescopiny, by selecting suboptimum
Intra4x4 intra prediction mode, reduce as far as possible intra prediction mode disturbance video compression coding efficiency is generated it is negative
It influences.
The algorithm by H.264 video encoding standard 9 kinds of Intra4x4 intra prediction mode Mode_i (i=0,1 ...,
8) two groups of A and B are divided into, wherein the intra prediction mode in A group represents bit 0, and the intra prediction mode in B group represents ratio
Spy 1.
A={ Mode_0, Mode_2, Mode_4, Mode_6, Mode_8 }
B={ Mode_1, Mode_3, Mode_5, Mode_7 }
The insertion process of the algorithm is as shown.Specifically, the algorithm to Intra4x4 intra prediction mode Mode_i into
When row steganography is modified, it perform the following steps in sequence and selects most suitable suboptimum from all candidate's Intra4x4 intra prediction modes
Intra4x4 intra prediction mode is as final modification result.
A) candidate intra prediction mode grouping is determined.According to H.264/AVC video encoding standard, the index of Mode_i is determined
Number, and candidate's Intra4x4 intra prediction mode set X is determined according to its lowest bit position: if Mode_i index number is most
Low bit position is 0, then X=B；If the lowest bit position of Mode_i index number is 1, X=A；
B) optimal candidate intra prediction mode determines.For any candidate's Intra4x4 intra prediction mode in X
IMODE calculates its corresponding Lagrangian cost
J (IMODE)=D+ λ R
Wherein, D indicates to carry out encoding introduced vision distortion to corresponding 4 × 4 pieces using IMODE；R presentation code
Bit number needed for bit number needed for IMODE and corresponding 4 × 4 Block Brightness transformation coefficient；λ is LaGrange parameter, for control D and
Balance between R, is determined by encoder.The Intra4x4 intra prediction mould with minimum Lagrangian cost is determined in X
Formula, as optimal modification result；
C) next code.According to step b's) as a result, using optimal candidate Intra4x4 intra prediction mode to corresponding 4 ×
4 pieces of progress intraframe predictive codings, and next code result is exported to code stream.
It is proposed by the present invention based on the video steganography method merged based on multi-embedding domain based on above description content, wherein
The telescopiny of secret information includes the following steps (unless otherwise specified, following steps are executed by computer):
1) video frame for needing to carry out steganography insertion is obtained；
2) it is originated from first macro block of current video frame, from quantization parameter domain, motion vector field, intra-prediction three
Aspect be embedded in by bit to secret information；
3) if current video frame type is P frame and macro block (mb) type is 8x8 piecemeal, embedded mobile GIS, which is chosen, is based on quantization parameter
Video steganographic algorithm be embedded in.Determine the quantization parameter of current macro, if the quantization parameter of current macro is not 0,
Close breath insertion is carried out according to preset steganographic algorithm；
If 4) current video frame type is P frame and macro block (mb) type is 16x16 piecemeal, 16x8 piecemeal or 8x16 piecemeal,
Embedded mobile GIS is chosen the video steganographic algorithm based on motion vector and is embedded in.Determine the motion vector MV of current macro0,0=
(mvx,mvy), close breath, which is carried out, according to preset steganographic algorithm is embedded in；
If 5) current video frame type is I frame and when macro block (mb) type is Intra4x4, embedded mobile GIS is chosen based in frame
The video steganographic algorithm of prediction is embedded in, and is generated by intra-framed prediction module in modification video compression coding frame
Intra4x4 intra prediction mode is to implement steganography insertion；
6) step 3) is repeated to 5) to the remaining macro block of the frame, until traversing all macro blocks；
7) according to the frame coded sequence of video file, step 2) is repeated to 6), until secret information to remaining every frame
It has fully embedded into and finishes, obtain final steganography video file.
Further, the extraction process of the secret information based on multi-embedding domain include the following steps (unless otherwise specified, with
Lower step is executed by computer):
1) it obtains and needs to carry out the video frame that close breath extracts；
2) it is originated from first macro block of current video frame, according to quantization parameter domain, motion vector field, intra-prediction
Sequence carries out the extraction of close breath by bit；
If 3) current video frame type is P frame and macro block (mb) type is 8x8 piecemeal, there is secret information embedding in quantization parameter domain
Enter, the extraction of close breath is carried out according to preset steganographic algorithm；
If 4) current video frame type is P frame and macro block (mb) type is 16x16 piecemeal, 16x8 piecemeal or 8x16 piecemeal,
There is secret information insertion in motion vector field, the extraction of close breath is carried out according to preset steganographic algorithm；
5) if current video frame type is I frame and when macro block (mb) type is Intra4x4, there is secret letter in intra-prediction
Breath insertion, the extraction of close breath is carried out according to preset steganographic algorithm；
6) step 3) is repeated to 5) to the remaining macro block of the frame, until traversing all macro blocks；
7) according to the frame coded sequence of video file, step 2) is repeated to 6), until secret information to remaining every frame
It all extracts and finishes.
Accordingly with above method, the present invention also provides a kind of video steganography device based on the fusion of multi-embedding domain, packets
It includes:
Video frame obtains module, responsible to obtain the video frame for needing to carry out steganography insertion；
Multi-embedding domain fusion insertion module, is responsible for since first macro block of current video frame, according to quantization parameter domain,
Motion vector field and intra-prediction be embedded in by bit to secret information, until traversing all macro blocks of the video frame；So
Afterwards according to the frame coded sequence of video file, the insertion of secret information is carried out to remaining each video frame, until secret information
It has fully embedded into and finishes, obtain final steganography video file.
Further, above-mentioned apparatus further includes secret information extraction module, is responsible for extracting secret information from steganography video；
The secret information extraction module, which obtains, needs to carry out the video frame that close breath extracts；It is opened from first macro block of current video frame
Begin, carry out the extraction of close breath by bit according to quantization parameter domain, motion vector field and intra-prediction, until traversing the video frame
All macro blocks；Then according to the frame coded sequence of video file, same processing is done to remaining each video frame, until secret
Information is all extracted and is finished.
The present invention also provides a kind of computer equipments comprising memory and processor, the memory store computer
Program, the computer program are configured as being executed by the processor, and the computer program includes for executing institute above
State the instruction of each step in method.
Video steganography method of the invention has the beneficial effect that correlative technology field:
1) embedding capacity is big.Existing overwhelming majority video steganographic algorithm carries out close breath steganography only for single compressed domain,
Because relatively limited for the part of modification in compression embedded domain, it will lead to embedding capacity and calculated not as good as spatial domain video steganography
Method.Another feature of compression domain is that insertion field selectivity is more, and algorithm type is abundant, still have in terms of improving steganography load factor compared with
Large space.For above-mentioned limitation, the present invention increases embeddable appearance by carrying out close breath insertion in multiple embedded domain simultaneously
Amount, further increases algorithmic load performance.
2) highly-safe.The existing nearly all targeted steganalysis method of video steganographic algorithm, in practical application
Middle safety is lower.It is carried out simultaneously to compressed video bit stream specific syntax elements (such as motion vector, inter-frame forecast mode)
When steganography is modified, steganography safety will be adversely affected when load factor is excessive.For this limitation, the present invention will be secret
Information dispersion is embedded in different compression embedded domain, guarantees that the load factor of each embedded domain guarantees in certain Low threshold with this
Under, while meeting higher total load rate, guarantee certain steganography safety.
3) attribute and coding parameter of steganography front and back initial carrier video are kept.Existing video steganographic algorithm is not examined usually
The essential attribute parameter (such as frame per second, code rate, resolution ratio) for considering initial carrier video, so that original load can be destroyed to a certain extent
The audio visual quality of volumetric video causes distortion or sound to draw phenomena such as asynchronous.For above-mentioned limitation, the present invention is in steganography
Before insertion, first the essential attribute parameter of initial carrier video is quickly identified or estimated, steganography is regarded according to testing result
The compressed encoding parameter of frequency carries out reasonable set, and steganography video is enabled effectively to keep the audio visual matter of initial carrier video
Amount, audio frequency parameter simultaneously maintain sound to draw synchronization, improve steganography safety to a certain extent.
4) intermediate file is not generated.Existing video steganographic algorithm can generally generate intermediate text when carrying out steganography insertion
Part (such as YUV file) is added significantly to the space expense of software operation, reduces operational efficiency, greatly constrains being applicable in for software
Range.For above-mentioned limitation, the present invention improves the insertion process and frame of existing video steganographic algorithm, is carrying out
When steganography is embedded in, by memory to the generation of intermediate file, the operations such as use and discharge control, effectively prevent
Between file generation, alleviate I/O load, significantly reduce software operation space expense, improve software operation efficiency and use
Range.
5) it is adapted to high-resolution video.The applicable carrier video resolution of existing video steganographic algorithm is often CIF, point
Resolution is lower.Now with capture apparatus keep updating and the continuous promotion of processor performance, major video differentiate
Code rate is also higher and higher, and existing video steganographic algorithm often will appear place when the video to high-resolution carries out steganography
Phenomena such as managing longer time or even collapse.For above-mentioned limitation, the present invention carries out secret letter while to coding and decoding video
The insertion of breath is capable of handling the steganography insertion of the high-resolutions video such as 1080P.
Detailed description of the invention
Fig. 1 is the overview flow chart of secret information insertion of the invention.
Fig. 2 is the overview flow chart that secret information of the invention extracts.
Fig. 3 is the steganography process schematic in the present invention based on motion vector.
Fig. 4 is the steganography process schematic in the present invention based on quantization parameter.
Fig. 5 is the steganography process schematic in the present invention based on intra prediction.
Fig. 6 is that the secret information of the invention based on the fusion of multi-embedding domain is embedded in flow chart.
Fig. 7 is that the secret information of the invention based on the fusion of multi-embedding domain extracts flow chart.
Specific embodiment
Below by specific implementation and the invention will be further described in conjunction with attached drawing 6-7.
Before carrying out the insertion of secret information steganography using the present invention, the data that can be first embedded in needs are encrypted, and are obtained
To random binary data stream.The secret information insertion of video steganographic algorithm proposed by the present invention based on the fusion of multi-embedding domain
Process is as shown in fig. 6, concrete operations details is as follows:
1) video frame for needing to carry out steganography insertion is obtained；
2) it is originated from first macro block of current video frame, according to quantization parameter domain, motion vector field, intra-prediction
Sequence be embedded in by bit to secret information；
3) if current video frame type is P frame and macro block (mb) type is 8x8 piecemeal, embedded mobile GIS, which is chosen, is based on quantization parameter
Video steganographic algorithm be embedded in.It determines the quantization parameter of current macro, if the quantization parameter of current macro is not 0, sentences
Whether the quantization parameter last bit of disconnected current macro matches with the current bit position to embedding secret information.Matching then keeps quantization parameter
It is constant, the operation of plus-minus one is otherwise carried out to the quantization parameter of current macro, while guaranteeing that the numerical value of quantization parameter is consistently less than
51.Steganography process based on quantization parameter is as shown in Figure 4.
If 4) current video frame type is P frame and macro block (mb) type is 16x16 piecemeal, 16x8 piecemeal or 8x16 piecemeal,
Embedded mobile GIS is chosen the video steganographic algorithm based on motion vector and is embedded in.Determine the motion vector MV of current macro0,0=
(mvx,mvy), while region of search delimited, for each candidate motion vector in the region of search, local optimum is carried out respectively
Sex determination, determines the motion vector of wherein all local optimums, and by their composite set Ψ.Most for each part in Ψ
Excellent motion vector and to calculate its rate distortion costs, select the local optimum motion vector for possessing minimum rate distortion costs as most
Modification result eventually.Steganography process based on motion vector is as shown in Figure 3.
If 5) current video frame type is I frame and when macro block (mb) type is Intra4x4, embedded mobile GIS is chosen based in frame
The video steganographic algorithm of prediction is embedded in.It is determined in Intra4x4 frame according to the current bit position of embedding secret information first
The Candidate Set of prediction mode, and the Lagrangian cost of all modes in the Candidate Set is calculated, wherein having minimum Lagrange
The Intra4x4 intra prediction mode of cost is as optimal modification result.Steganography process based on intra prediction is as shown in Figure 5.
6) step 3) is repeated to 5) to the remaining macro block of the frame, until traversing all macro blocks；
7) according to the frame coded sequence of video file, step 2) is repeated to 6) to remaining every frame, it is known that secret information
It has fully embedded into and finishes, obtain final steganography video file.
The secret information of video steganographic algorithm proposed by the present invention based on the fusion of multi-embedding domain extracts process such as Fig. 7 institute
Show, concrete operations details is as follows:
1) video frame for needing to carry out secret information extraction is obtained；
2) it is originated from first macro block of current video frame, according to quantization parameter domain, motion vector field, intra-prediction
Sequence carries out by bit extraction secret information；
If 3) current video frame type is P frame and macro block (mb) type is 8x8 piecemeal, it is hidden to choose the video based on quantization parameter
Algorithm is write to extract.It determines the quantization parameter of current macro, if the quantization parameter of current macro is not 0, extracts current macro
Block quantization parameter the last as extract secret information bit；
If 4) current video frame type is P frame and macro block (mb) type is 16x16 piecemeal, 16x8 piecemeal or 8x16 piecemeal,
The video steganographic algorithm based on motion vector is chosen to extract.Determine the motion vector MV of current macro0,0=(mvx,
mvy), obtain the lowest bit position i.e. LSB (mv of the sum of its horizontal and vertical componentx+mvy), and believe as the secret of extraction
Cease bit；
If 5) current video frame type is I frame and when macro block (mb) type is Intra4x4 intra prediction mode, selection is based on
The video steganographic algorithm of intra prediction extracts.According to Intra4x4 intra prediction mode as the secret information ratio extracted
It is special；
6) step 3) is repeated to 5) to the remaining macro block of the frame, until traversing all macro blocks；
7) according to the frame coded sequence of video file, step 2) is repeated to 6) to remaining every frame, it is known that secret information
It all extracts and finishes.
Illustrate that proposed by the present invention is a kind of video steganography method of highly concealed type to protrude, to MP4 format and resolution
Rate is that the H.264/AVC video of 1080p carries out the insertion of close breath, then carries out steganalysis detection by ADB_S method.Specifically
Video sample collection is prepared using following configuration and carries out steganalysis experiment:
The MP4 format video that 40 resolution ratio are 1080p is chosen, after carrying out close breath insertion, is carried out by ADB_S method
Steganalysis detection, wherein 29 pairs of training, 11 pairs of detections, and statistical correction rate, the results are shown in Table 1.
Table 1 tests and analyzes the video steganography method merged based on multi-embedding domain by ADB_S method
Pass through table 1, it can be seen that be directed to ADB_S method, when Rate of average load is more than 5%, detection accuracy does not surpass still
Cross 60%.
Another embodiment of the present invention provides a kind of video steganography devices based on the fusion of multi-embedding domain comprising: video frame
Module is obtained, it is responsible to obtain the video frame for needing to carry out steganography insertion；Multi-embedding domain fusion insertion module, is responsible for from current video
First macro block of frame starts, and carries out secret information by bit according to quantization parameter domain, motion vector field and intra-prediction
Insertion, until traversing all macro blocks of the video frame；Then according to the frame coded sequence of video file, to remaining each video
Frame carries out the insertion of secret information, finishes until secret information has fully embedded into, obtains final steganography video file.
Above-mentioned apparatus further includes secret information extraction module, is responsible for extracting secret information from steganography video；The secret
Information extraction modules, which obtain, needs to carry out the video frame that close breath extracts；Since first macro block of current video frame, according to amount
Change the extraction that parameter field, motion vector field and intra-prediction carry out close breath by bit, until traversing all macro of the video frame
Block；Then according to the frame coded sequence of video file, same processing is done to remaining each video frame, until secret information whole
Extraction finishes.
Another embodiment of the present invention provides a kind of computer equipments comprising memory and processor, the memory are deposited
Computer program is stored up, the computer program is configured as being executed by the processor, and the computer program includes for holding
The instruction of each step in row method described above.
The above embodiments are merely illustrative of the technical solutions of the present invention rather than is limited, the ordinary skill of this field
Personnel can be with modification or equivalent replacement of the technical solution of the present invention are made, without departing from the spirit and scope of the present invention, this
The protection scope of invention should subject to the claims.
Claims (10)
1. a kind of video steganography method based on the fusion of multi-embedding domain, which comprises the following steps:
1) video frame for needing to carry out steganography insertion is obtained；
2) since first macro block of current video frame, according to quantization parameter domain, motion vector field and intra-prediction to secret
Confidential information be embedded in by bit, until traversing all macro blocks of the video frame；
3) according to the frame coded sequence of video file, step 2) is repeated to remaining each video frame, until secret information
It has fully embedded into and finishes, obtain final steganography video file.
2. the method according to claim 1, wherein step 2) includes:
2.1) if current video frame type is P frame and macro block (mb) type is 8x8 piecemeal, embedded mobile GIS is chosen based on quantization parameter
Video steganographic algorithm, the quantization parameter by modifying video encoding unit are embedded in；
2.2) if current video frame type is P frame and macro block (mb) type is 16x16 piecemeal, 16x8 piecemeal or 8x16 piecemeal, embedding
Enter video steganographic algorithm of the algorithm picks based on motion vector, is produced by motion estimation module in modification video compression coding frame
Raw motion vector is embedded in；
2.3) if current video frame type is I frame and macro block (mb) type is Intra4x4, embedded mobile GIS, which is chosen, is based on intra prediction
Video steganographic algorithm, pass through modification video compression coding frame in intra-framed prediction module generate Intra4x4 intra prediction mould
Formula is embedded in.
3. according to the method described in claim 2, it is characterized in that, the video steganographic algorithm based on quantization parameter include with
Lower step:
A) present frame is encoded, the quantization parameter of current macro is obtained, judges whether it is 0, if it is 0, skip to next macro block, it is no
Then, step b) is continued to execute；
B) judge whether the quantization parameter last bit of current macro matches with the current bit position to embedding secret information, matching is then kept
Quantization parameter is constant, otherwise, executes step c)；
C) it is embedded in, i.e., carries out the operation of plus-minus one to the quantization parameter of current macro, while guaranteeing the numerical value of quantization parameter
It is consistently less than 51；
D) subsequent coding steps for continuing to execute the video frame export gained steganography video frame to code stream.
4. according to the method described in claim 2, it is characterized in that, the video steganographic algorithm based on motion vector field when pair
Motion vector MV0,0=(mvx, mvy) when modifying, perform the following steps in sequence and select most to close from all candidate motion vectors
Suitable local optimum motion vector is as final modification result:
A) respectively to MV0,0Four motion vector MV in contiguous range- 1,0=(mvx- 1, mvy)、MV1,0=(mvx+ 1, mvy)、
MV0, -1=(mvx, mvy- 1) and MV0,1=(mvx, mvy+ 1) local optimum sex determination is carried out；
B) by MV- 1,0, MV1,0, MV0, -1And MV0,1In local optimum motion vector composition candidate motion vector set close Ψ, if Ψ
Non-empty then gos to step e), otherwise continues to execute step c)；
C) region of search for delimiting a motion vector, for any candidate motion vector MV '=(mv ' in the regionx,
mvy'), it is necessary to meet LSB (mvx+mvy)≠LSB(mvx′+mvy′)；
D) for each candidate motion vector in the region of search, local optimum sex determination is carried out respectively, and determination is wherein owned
The motion vector of local optimum, and their groups are combined into set Ψ；
E) for each local optimum motion vector in Ψ, its rate distortion costs is calculated, selection possesses minimum rate distortion costs
Local optimum motion vector as final modification result.
5. according to the method described in claim 2, it is characterized in that, the video steganographic algorithm based on intra prediction mode exists
When carrying out steganography modification to Intra4x4 intra prediction mode Mode_i, it perform the following steps in sequence from all candidate Intra4x4
Select most suitable suboptimum Intra4x4 intra prediction mode as final modification result in intra prediction mode:
A) it according to H.264/AVC video encoding standard, determines the index number of Mode_i, and is determined and waited according to its lowest bit position
Select Intra4x4 intra prediction mode set X: if the lowest bit position of Mode_i index number is 0, X=B；If ModeOri
The lowest bit position of index number is 1, then X=A；
B) for any candidate Intra4x4 intra prediction mode IMODE in X, its corresponding Lagrangian cost is calculated:
J (IMODE)=D+ λ R
Wherein, D indicates to carry out encoding introduced vision distortion to corresponding 4 × 4 pieces using IMODE；R presentation code IMODE institute
Bit number needed for needing bit number and corresponding 4 × 4 Block Brightness transformation coefficient；λ is LaGrange parameter, for controlling between D and R
Balance, is determined by encoder；The Intra4x4 intra prediction mode with minimum Lagrangian cost is determined in X, by it
As optimal candidate intra prediction mode；
C) according to step b's) as a result, using optimal candidate Intra4x4 intra prediction mode to pre- in corresponding 4 × 4 pieces of progress frame
Coding is surveyed, and next code result is exported to code stream.
6. the method according to claim 1, wherein carrying out the extraction of secret information using following steps:
[1] it obtains and needs to carry out the video frame that close breath extracts；
[2] since first macro block of current video frame, according to quantization parameter domain, motion vector field and intra-prediction by than
Spy carries out the extraction of close breath, until traversing all macro blocks of the video frame；
[3] according to the frame coded sequence of video file, step [2] are repeated to remaining each video frame, until secret is believed
Breath is all extracted and is finished.
7. according to the method described in claim 6, it is characterized in that, step [2] includes:
[2.1] if current video frame type is P frame and macro block (mb) type is 8x8 piecemeal, there is secret information embedding in quantization parameter domain
Enter, the extraction of close breath is carried out according to preset steganographic algorithm；
[2.2] if current video frame type is P frame and macro block (mb) type is 16x16 piecemeal, 16x8 piecemeal or 8x16 piecemeal,
Motion vector field has secret information insertion, and the extraction of close breath is carried out according to preset steganographic algorithm；
[2.3] if current video frame type is I frame and when macro block (mb) type is Intra4x4, there is secret information in intra-prediction
Insertion, the extraction of close breath is carried out according to preset steganographic algorithm.
8. a kind of video steganography device based on the fusion of multi-embedding domain characterized by comprising
Video frame obtains module, responsible to obtain the video frame for needing to carry out steganography insertion；
Multi-embedding domain fusion insertion module, is responsible for since first macro block of current video frame, according to quantization parameter domain, movement
Vector field and intra-prediction be embedded in by bit to secret information, until traversing all macro blocks of the video frame；Then it presses
According to the frame coded sequence of video file, the insertion of secret information is carried out to remaining each video frame, until secret information whole
Insertion finishes, and obtains final steganography video file.
9. device according to claim 8, which is characterized in that further include secret information extraction module, be responsible for regarding from steganography
Secret information is extracted in frequency；The secret information extraction module, which obtains, needs to carry out the video frame that close breath extracts；From current video
First macro block of frame starts, and carries out the extraction of close breath by bit according to quantization parameter domain, motion vector field and intra-prediction,
Until traversing all macro blocks of the video frame；Then according to the frame coded sequence of video file, remaining each video frame is done
Same processing is finished until secret information all extracts.
10. a kind of computer equipment, which is characterized in that including memory and processor, the memory stores computer program,
The computer program is configured as being executed by the processor, and the computer program includes requiring 1-7 for perform claim
The instruction of each step in middle any claim the method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910102368.1A CN109819260B (en) | 2019-02-01 | 2019-02-01 | Video steganography method and device based on multi-embedded domain fusion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910102368.1A CN109819260B (en) | 2019-02-01 | 2019-02-01 | Video steganography method and device based on multi-embedded domain fusion |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109819260A true CN109819260A (en) | 2019-05-28 |
| CN109819260B CN109819260B (en) | 2020-11-06 |
Family
ID=66606303
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910102368.1A Expired - Fee Related CN109819260B (en) | 2019-02-01 | 2019-02-01 | Video steganography method and device based on multi-embedded domain fusion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109819260B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111263157A (en) * | 2020-02-27 | 2020-06-09 | 武汉大学 | Video multi-domain steganalysis method based on motion vector consistency |
| CN112738355A (en) * | 2020-12-24 | 2021-04-30 | 华南理工大学 | Non-degraded HEVC video steganography method capable of resisting deep learning network detection |
| CN113347435A (en) * | 2019-06-21 | 2021-09-03 | 杭州海康威视数字技术股份有限公司 | Method and device for decoding and encoding prediction mode |
| CN113840146A (en) * | 2021-09-30 | 2021-12-24 | 中南大学 | Video information hiding method and system |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080180571A1 (en) * | 2007-01-31 | 2008-07-31 | Liang-Shing Chuang | Method and Related Apparatus for Hiding Data Inside Video Signals and Transmitting the Video Signals to a Display Device |
| US7965838B2 (en) * | 2007-11-30 | 2011-06-21 | Faraday Technology Corp. | Watermark generating circuit and related method thereof |
| CN102223540A (en) * | 2011-07-01 | 2011-10-19 | 宁波大学 | Information hiding method facing to H.264/AVC (automatic volume control) video |
| CN102547297A (en) * | 2012-02-28 | 2012-07-04 | 中国传媒大学 | MPEG2 (Moving Picture Experts Group 2) video watermarking realization method based on DC (Discrete Cosine) coefficient |
| CN102905134A (en) * | 2012-10-22 | 2013-01-30 | 山东省计算中心 | Adaptive video digital steganography method |
| CN103338376A (en) * | 2013-07-02 | 2013-10-02 | 中国科学技术大学 | Video steganography method based on motion vector |
| CN103826128A (en) * | 2014-02-25 | 2014-05-28 | 宁波工程学院 | Encryption domain H.264/AVC video reversible data hiding method |
| CN104581176A (en) * | 2014-12-26 | 2015-04-29 | 宁波大学 | H.264/AVC (advanced video coding) compressed domain robust video watermark embedding and extracting methods free from intra-frame error drift |
-
2019
- 2019-02-01 CN CN201910102368.1A patent/CN109819260B/en not_active Expired - Fee Related
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080180571A1 (en) * | 2007-01-31 | 2008-07-31 | Liang-Shing Chuang | Method and Related Apparatus for Hiding Data Inside Video Signals and Transmitting the Video Signals to a Display Device |
| US7965838B2 (en) * | 2007-11-30 | 2011-06-21 | Faraday Technology Corp. | Watermark generating circuit and related method thereof |
| CN102223540A (en) * | 2011-07-01 | 2011-10-19 | 宁波大学 | Information hiding method facing to H.264/AVC (automatic volume control) video |
| CN102547297A (en) * | 2012-02-28 | 2012-07-04 | 中国传媒大学 | MPEG2 (Moving Picture Experts Group 2) video watermarking realization method based on DC (Discrete Cosine) coefficient |
| CN102905134A (en) * | 2012-10-22 | 2013-01-30 | 山东省计算中心 | Adaptive video digital steganography method |
| CN103338376A (en) * | 2013-07-02 | 2013-10-02 | 中国科学技术大学 | Video steganography method based on motion vector |
| CN103826128A (en) * | 2014-02-25 | 2014-05-28 | 宁波工程学院 | Encryption domain H.264/AVC video reversible data hiding method |
| CN104581176A (en) * | 2014-12-26 | 2015-04-29 | 宁波大学 | H.264/AVC (advanced video coding) compressed domain robust video watermark embedding and extracting methods free from intra-frame error drift |
Non-Patent Citations (4)
| Title |
|---|
| DAWEN XU; RANGDING WANG; JICHENG WANG: "Prediction mode modulated data-hiding algorithm for H.264/AVC", 《JOURNAL OF REAL-TIME IMAGE PROCESSING》 * |
| GAOBO YANG, JUNJIE LI, YINGLIANG HE, ZHIWEI KANG: "《International Journal of Electronics and Communications(AEU)》", 30 April 2011 * |
| TAMER SHANABLEH: "《IEEE Transactions on Information Forensics and Security》", 30 April 2012 * |
| 张弘，尤玮珂，赵险峰: "视频隐写分析技术研究综述", 《信息安全学报》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113347435A (en) * | 2019-06-21 | 2021-09-03 | 杭州海康威视数字技术股份有限公司 | Method and device for decoding and encoding prediction mode |
| CN113347435B (en) * | 2019-06-21 | 2022-03-29 | 杭州海康威视数字技术股份有限公司 | Method and device for decoding and encoding prediction mode |
| CN111263157A (en) * | 2020-02-27 | 2020-06-09 | 武汉大学 | Video multi-domain steganalysis method based on motion vector consistency |
| CN112738355A (en) * | 2020-12-24 | 2021-04-30 | 华南理工大学 | Non-degraded HEVC video steganography method capable of resisting deep learning network detection |
| CN113840146A (en) * | 2021-09-30 | 2021-12-24 | 中南大学 | Video information hiding method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109819260B (en) | 2020-11-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109819260A (en) | Video steganography method and device based on the fusion of multi-embedding domain | |
| CN105981389B (en) | Picture coding device, picture decoding apparatus, encoding stream converting means, image encoding method and picture decoding method | |
| CN105325000B (en) | Picture coding device, image encoding method, picture decoding apparatus and picture decoding method | |
| CN102223540B (en) | Information hiding method facing to H.264/AVC (automatic volume control) video | |
| CN103338376B (en) | A kind of video steganography method based on motion vector | |
| CN107925763A (en) | The transcoding, coding transform method and apparatus of the selection of block level transforming and implicit signaling in Multi-level segmentation | |
| CN106559669A (en) | The method and device of image prediction | |
| CN101841700B (en) | H.264 compression bit stream oriented video watermark method | |
| CN104159117B (en) | Method for hiding HEVC video information | |
| CN109952577A (en) | Digital frame is encoded/decoded by down-sampling/up-sampling and enhancement information | |
| CN105933711B (en) | Neighborhood optimum probability video steganalysis method and system based on segmentation | |
| CN106101714B (en) | A kind of and tightly coupled H.264 video information hiding method of compression encoding process | |
| CN104010193B (en) | Video steganographic method based on macro block partition mode disturbance | |
| CN108965887B (en) | Video information hiding method and device based on inter-block decoupling | |
| CN102685500B (en) | H.264 large-capacity information hiding method based on information variable-length packet structure | |
| CN108028945A (en) | The apparatus and method of conversion are performed by using singleton coefficient update | |
| CN104602016A (en) | HEVC video information hiding method based on intra-frame prediction mode difference | |
| CN105681803B (en) | A kind of HEVC video information hiding methods of large capacity | |
| CN106791828B (en) | Video transcoding method and its transcoder based on machine learning | |
| CN105979269B (en) | Motion vector field video steganography method based on novel insertion cost | |
| CN104853215B (en) | The video steganography method kept based on motion vector local optimality | |
| Fu et al. | Efficient depth intra frame coding in 3D-HEVC by corner points | |
| Dong et al. | Adaptive HEVC steganography based on steganographic compression efficiency degradation model | |
| CN105915916A (en) | Video steganalysis method based on motion vector rate-distortion performance estimation | |
| CN107135391B (en) | A kind of video-aware hash method for H.264 compression domain |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20201106 Termination date: 20210201 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |