CN1141835C - Method for protecting multimedia video data - Google Patents
Method for protecting multimedia video data Download PDFInfo
- Publication number
- CN1141835C CN1141835C CNB021151067A CN02115106A CN1141835C CN 1141835 C CN1141835 C CN 1141835C CN B021151067 A CNB021151067 A CN B021151067A CN 02115106 A CN02115106 A CN 02115106A CN 1141835 C CN1141835 C CN 1141835C
- Authority
- CN
- China
- Prior art keywords
- frame
- watermark
- video
- dimensional
- perform
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Images
Landscapes
- Editing Of Facsimile Originals (AREA)
- Image Processing (AREA)
Abstract
The present invention relates to a method for protecting multimedia video data, which is a kind of digital watermark technology based on wavelet transformation and digital communication technology and belongs to the field of multimedia signal processing. The present invention comprises the following steps that the video data is decomposed by means of the wavelet transformation; a watermark is preprocessed by the digital communication technology; a wavelet transformation domain is embedded with the watermark; video data embedded with the watermark is obtained through inverse wavelet transformation. Compared with the original video data, human eyes can not distinguish both differences and a digital watermark provides a distinguishable means for the problem, wherein the watermark can be any meaningful digital file. The present invention can protect the video data or the files propagated through Internet.
Description
The present invention relates to a kind of multimedia video data guard method, is a kind of digital watermark technology based on wavelet transformation, digital communication technology, belongs to field of multimedia signal processing.
In recent years, digital watermark technology had embodied its importance in many applications, and had obtained paying attention to widely.Although the research of present most of data hiddens and document concentrate on image watermark, digital video watermarking and Information Hiding Techniques also are being subjected to increasing attention.
The type that hides Info in the digital watermarking algorithm can be a random sequence, also can be significant content (as text, image, video, audio frequency etc.).The watermark that is made of the former can only provide the answer of " watermark is arranged " or " no watermark ", the secret information of lbit also promptly only is provided, thereby its application is restricted.In most of application scenarios, the information that requires to embed is readable or visual, as meaningful character string (possessory name, address, sign etc.), image (trade mark, seal etc.), video (scene that has the TV user of certain access rights just can see) etc., therefore meaningful data hiding has application scenario widely.
Digital watermarking algorithm can be divided into two classes: need the private system of original media and do not need the publicly-owned system of original media.Usually, because the auxiliary of original media arranged, the robustness of private system is better than blind-detection system, yet because in most of application scenarios, difficulty is all compared in obtaining of original media, makes blind-detection system have more applications and be worth.Because the huge data volume of video, Blind Detect Algorithm is even more important for video watermark.
The video watermark algorithm is divided into two kinds according to watermarked data field: compression domain algorithm and uncompressed domain algorithm.Wherein most of algorithm belongs to the latter.It is main that present most video watermark algorithms adopt random sequence, less even significant watermarking algorithm also exists the hiding data amount, and sane inadequately deficiency.One of them famous method is a kind of direct method watermarked in non-compression video that Hartungand Girod proposes, and for ntsc video, the data transfer rate of watermark is 528byte/s, and the data that every frame can embed are 17.6byte.Do not having under the situation under attack, the bit error rate that detects data is 5 * 10
-7Watermarking algorithm on another famous compressed video territory is that Hartung andGirod proposes, and for ntsc video, watermarked data transfer rate is several bytes each second, and when the MPEG-2 compressed bit stream was 7Mbps, bit error rate (BER) was 5 * 10
-3In addition, existing most of video watermark algorithm directly expands to video with the algorithm on the image and obtains.Therefore lack effective frame synchronizing, the performance that the antagonism time shaft is attacked a little less than.
The objective of the invention is to propose a kind of multimedia signal dispose method, make the detection that hides Info not need original media; Reduced the error rate of the information that detects; Solve the distinctive robustness problem of video watermark simultaneously.
To achieve these goals, the inventive method adopts following watermark to embed and extract two processes, the watermark embed process step is as follows: 1) luminance component with every frame in the video sequence carries out 3 grades of 2-D DWT (Discrete WaveletTransform), extracts DWT low frequency sub-band coefficient; 2) error correction coding is carried out in watermark; 3) the DWT low frequency sub-band coefficient of each frame of video carries out 3-D (Dimension) and interweaves; 4) in the DWT low frequency sub-band coefficient after pretreated digital watermarking is embedded into 3-D interweaves; 5) the DWT low frequency sub-band coefficient after the embed digital watermark being carried out the 3-D reciprocal cross knits; 6) four jiaos at low frequency sub-band coefficient figure embed the synchronizing informations of passing through error correction codings; 7) wavelet inverse transformation obtains watermarked frame of video;
The watermark extraction process step is as follows: 1) luminance component with every frame in the video sequence carries out 3 grades of 2-D DWT (Discrete Wavelet Transform), extracts DWT low frequency sub-band coefficient; 2) extract synchronizing information and decoding, recover synchronously according to synchronizing information; 3) the DWT low frequency sub-band coefficient of each frame of video carries out 3-D (Dimension) and interweaves; 4) extract the information that embeds; 5) carry out error-correcting decoding, obtain watermark.
Watermark can be the digital document of text, numeral, figure or signature, by certain scanning sequency it is formed the data of one dimension when embedding.
Below the inventive method is further described.
Watermark embed process of the present invention is:
1) luminance component with every frame in the video sequence carries out 3 grades of 2-D DWT conversion, obtains 10 subbands
{ LH
i, HL
i, HH
i, LL
3, i=1-3}, wherein LL
3Subband is that the low pass of original video frame is approximate, and the present invention selects LL
3
Sub-band coefficients embeds information;
2) watermark W is carried out the BCH error correction coding, generate X.The present invention adopts the BCH code of (r, 8) form,
Wherein r represents that BCH code is long.The present invention is with a code book with 256 BCH code words, S
m, 0≤m<256,
Come 256 probable values of a corresponding byte.Treat each byte W of hiding data
i, corresponding in the set of signals
A prototype signal S
m, be mapped as a BCH code word:
Compromise capacity and the error correcting capability of considering frame of video of definite needs of r.R is big more, the Hamming distance between two code words
From big more, error correcting capability is strong more, but suppressible data volume can be more little simultaneously.It is 61 that the present invention selects r.With
Hexadecimal representation, the generator polynomial of BCH is:
g(z)={2759262D5D?506D}
According to the error correction coding principle, the Hamming distance between above-mentioned any two BCH code words is 27, can correct 13
Individual random error.X{X
iCan regard a bit stream as.
3) at every frame LL
3Sub-band coefficients figure extracts at the middle part coefficient block of a N * N, and wherein N is 2 power power,
Again the coefficient block of N frame is put together, form the three-dimensional coefficient block B of a N * N * N, such one
Three-dimensional coefficient block adopts three-dimensional sliding window interweaving method to interweave, the N * N * N coefficient that has obtained interweaving
Piece B '.If what embed is significant information, but not random sequence, then frame losing or large-scale video cut
Disconnected, the continuous mistake (information bit of many embeddings has all been lost) of the position that will cause hiding Info is corresponding to communication
In burst error.The interweaving encoding technology is the effective ways of opposing burst error in the communication theory.So this
Bright introducing 3-D interleaving technology is to strengthen the robustness of watermark signal to antiburst error.
4) X is embedded among the B ' successively by bit, generates B "; The present invention proposes with representative information bit " 0 " to be embedded
Minimum 4 significance bits of the value displacement wavelet conversion coefficient integer part of " 1 ".Wherein representative information position " 1 "
The value of " 0 " is respectively " 1100b " and " 0100b ".If watermark image meets with to attack distortion appears, only
The variation of wavelet coefficient of wanting hiding data is in-4~3 scope, and then embedding data can correctly be examined
Measure.
5) " carry out the 3-D reciprocal cross and knit, the reciprocal cross organization method is identical with deinterleaving method with B;
6) will carry out with the frame number fn of a byte representation as BCH error correction coding identical in the step 2, generation length is
The fn ' of 61 bits, and in four jiaos of embeddings of the low frequency sub-band coefficient figure of each frame.The present invention uses respectively
Minimum 5 significance bits of " 11000b " and " 01000b " displacement wavelet conversion coefficient embed the synchronizing information position
" 0 " and " 1 ".
7) wavelet inverse transformation obtains watermarked frame of video.
The leaching process of watermark does not need original video among the present invention, and step is as follows:
1)DWT。The luminance component of each frame of video to be detected is carried out 3 grades of wavelet decomposition obtain 10 subbands.
Select the low frequency sub-band coefficient.
2) extract synchronizing information, recover frame synchronization.Extract 61 coefficients of each frame DWT low frequency sub-band four angle part
The 5th least significant bit of integer part carries out BCH decoding then and obtains frame number as synchronizing information.
According to the principle that frame number increases progressively the order of each frame is adjusted.If adjusting the frame number of adjacent two frames in back is respectively
I and J, if J-I ≠ 1, the LL of I frame
3The LL of I frame and J frame is gone in the subband interpolation
3Between the subband, repeat
J-I-1 time, with realize correct synchronously.
3) 3-D interweaves.Extract the piece of N * N in the LL3 wavelet coefficient that Synchronous Processing is crossed, wherein N is 2 power
Power, the piece of the N * N of N frame is formed the three-dimensional coefficient block B of a N * N * N
*, to B
*Carry out 3-D
Interweave, generate B
*
4) detect the actual information that embeds.Note coef
*Be B
*In wavelet coefficient to be detected, x
*For detecting the watermark letter
Breath.Coef
*The 4th significance bit of integer part of absolute value promptly be x
*, all x
*Form bit stream
x
**。
5) BCH decoding.If embedding data is character strings, then with x
*Be divided into length and be 61 subsequence, right
Pass through at S in each subfamily
mThe code word of middle search and its Hamming distance minimum can obtain the watermark word
Joint.
The present invention and existing video watermark technology relatively have following advantage:
1), rises as BCH chnnel coding, 3-D interleaving technology and simultaneous techniques combination with the technology in the digital communication theory
Introduce in the video watermark algorithm, reduced the error rate of the information that detects.
2) a kind of effective frame synchronizing has been proposed.Make the testing process can motor synchronizing, resisting video watermark effectively may
Attack on the time shaft that suffers is as frame losing, frame exchange, frame replacement etc.
3) data volume of the algorithm embedding of the present invention's proposition is bigger, and robustness is fine.For the CIF layout sequence, but every frame embedding
Go into about 16.8 bytes (redundancy section that does not comprise Bose-Chaudhuri-Hocquenghem Code).In the experiment, we are 352 * 288 of 96 frames
Sequence in, embedded 1150 characters.Under the situation of not attacked, watermarked can zero defect
Be restored.The robustness of watermarked video sequence behind the MPEG-2 coding is: smaller for motion
The Salesman sequence, when compression bit rate is 2.7Mb, CR (CompressionRatio)=11.16,
During PSNR=41.12dB, can detect the character string of embedding zero defect.For complicated movement, range of movement
Bigger Mobile sequence, when compression bit rate is 6Mb, CR=5.07, during PSNR=37.19dB, can indifference
Detect the character string of embedding wrongly.
Fig. 1-the 5th, the experimental result picture of video standard cycle tests Salesman and Mobile (every frame sign is 352 * 288, totally 96 frames).Watermark is that length is the text of 1150 bytes.
Fig. 1 is the digital watermarking figure of the video watermark algorithm that proposes of the present invention to video sequence Salesman.
Fig. 2 is the digital watermarking figure of the video watermark algorithm that proposes of the present invention to video sequence Mobile.
Fig. 3 is the robustness resolution chart (transverse axis be frame loss rate, the longitudinal axis be character error rate) of the video watermark algorithm that proposes of the present invention to frame losing.
(a) among the Salesman watermark to the robustness of continuous frame losing; (b) among the Salesman watermark to the robustness of interval frame losing.
Fig. 4 is that the video watermark algorithm that the present invention proposes is tested (transverse axis is a frame loss rate, and the longitudinal axis is a character error rate) to the robustness of frame losing
(a) among the Mobile watermark to the robustness of continuous frame losing; (b) among the Mobile watermark to the robustness of interval frame losing.
Fig. 5 is the robustness (transverse axis be the average peak signal to noise ratio of decoded video, the longitudinal axis be character error rate) of the video watermark algorithm that proposes of the present invention to the MPEG-2 coding
(a) robustness that watermark is encoded to MPEG-2 to watermark among robustness (b) Mobile of MPEG-2 coding among the Salesman.
Among Fig. 1, (a) be the original graph of the 20th frame of Salesman sequence, watermarked frame of video is presented among Fig. 1 (b), and its PSNR (Peak Signal Noise Ratio) is 49.23dB.The frame that has embedded watermark as seen from the figure visually with primitive frame insentience difference.Satisfy the requirement of invisibility.
Among Fig. 2, (a) be the original graph of the 17th frame of Mobile sequence, watermarked frame of video effect is presented among Fig. 2 (b), and its PSNR (Peak Signal Noise Ratio) is 49.26dB.The frame that has embedded watermark as seen from the figure visually with primitive frame insentience difference.Satisfy the requirement of invisibility.
Fig. 3 and Fig. 4 are respectively the watermark detection performance after Salesman and the frame losing of Mobile video sequence.Watermark embed process in the by specification obtains watermark video, then frame losing.Divide continuous frame losing and frame losing at interval.If one group is made up of 32 frames, then frame loss rate is:
Fig. 3 (a) and 3 (b) have shown that respectively Salesman detects the character error rate of the watermark that obtains under the various frame loss rates of frame losing and frame losing at interval continuously.Fig. 4 (a) and 4 (b) have shown that respectively Mobile detects the character error rate of the watermark that obtains under the various frame loss rates of frame losing and frame losing at interval continuously.Contrasted employing 3-D interleaving technology among each figure (among Fig. 3 (a) and 4 (a) corresponding to interleave consecutive frame loss, among Fig. 3 (b) and 4 (b) corresponding to interleavediscrete flame loss) and the robustness when not adopting interleaving technology (among Fig. 3 (a) and 4 (a) corresponding to original consecutiveflame loss, among Fig. 3 (b) and 4 (b) corresponding to original discrete frame loss).As seen from the figure
1) for continuous frame losing, in fr<0.5 o'clock, the robustness that adopts the 3-D interleaving technology when not adopting interleaving technology has bright
Show and improve.After adopting 3-D to interweave, in 32 frames continuously the number of lost frames greater than 12 o'clock, detected watermark
Just go wrong.
2), similarly, in fr<0.5 o'clock, adopt the 3-D interleaving technology all when not adopting interleaving technology for the interval frame losing
Robustness has clear improvement.
3) when the attack of antagonism on the time shaft, utilizing synchronizing information to recover on the synchronous basis, adopt the algorithm that interweaves than not
Robustness when employing interweaves algorithm has clear improvement.And experiment shows that method that the present invention proposes is for the frame losing tool
Good robustness is arranged.
Fig. 5 is the robustness test result of the method that proposes of the present invention to the MPEG-2 compression.For the Salesman sequence, when compression bit rate is 2.7Mb, CR=11.16 during PSNR=41.12dB, can detect the character string of embedding zero defect.When compression multiple was within 13.81 times, the character error rate of the character string of detected embedding was less than 1%.Fig. 5 (a) has provided sane performance curve.For the Mobile sequence, when compression bit rate is 6Mb, CR=5.07 during PSNR=37.19dB, can detect the character string of embedding zero defect.Fig. 5 (b) has provided sane performance curve.By experimental result as can be known, the method for the present invention's proposition is better than some existed algorithms for the robustness of MPEG-2 coding.
Common processing also has the spatial domain to amplify to video watermark.The algorithm that this paper proposes after watermarked video sequence is exaggerated, the information that detects embedding that can zero defect.
Claims (3)
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CNB021151067A CN1141835C (en) | 2002-04-19 | 2002-04-19 | Method for protecting multimedia video data |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CNB021151067A CN1141835C (en) | 2002-04-19 | 2002-04-19 | Method for protecting multimedia video data |
Publications (2)
Publication Number | Publication Date |
---|---|
CN1377180A CN1377180A (en) | 2002-10-30 |
CN1141835C true CN1141835C (en) | 2004-03-10 |
Family
ID=4743457
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CNB021151067A Expired - Fee Related CN1141835C (en) | 2002-04-19 | 2002-04-19 | Method for protecting multimedia video data |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN1141835C (en) |
Families Citing this family (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7929409B2 (en) | 2004-01-13 | 2011-04-19 | Interdigital Technology Corporation | Orthogonal frequency division multiplexing (OFDM) method and apparatus for protecting and authenticating wirelessly transmitted digital information |
CN1315091C (en) * | 2005-03-23 | 2007-05-09 | 中山大学 | A Feature-Based Digital Image Authentication Method |
FR2886787A1 (en) * | 2005-06-06 | 2006-12-08 | Thomson Licensing Sa | METHOD AND DEVICE FOR ENCODING AND DECODING AN IMAGE SEQUENCE |
CN103854250A (en) * | 2014-03-17 | 2014-06-11 | 广东石油化工学院 | Self-synchronizing image watermarking method applied to wireless multimedia sensor network |
CN115052078A (en) * | 2022-05-27 | 2022-09-13 | 武汉兴图新科电子股份有限公司 | High-security network camera based on digital watermarking technology |
-
2002
- 2002-04-19 CN CNB021151067A patent/CN1141835C/en not_active Expired - Fee Related
Also Published As
Publication number | Publication date |
---|---|
CN1377180A (en) | 2002-10-30 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Fan et al. | An enhanced fragile watermarking scheme to digital image protection and self-recovery | |
US7389420B2 (en) | Content authentication and recovery using digital watermarks | |
Singh et al. | Robust and secure multiple watermarking in wavelet domain | |
Kalantari et al. | A robust image watermarking in the ridgelet domain using universally optimum decoder | |
Huang et al. | Reliable information bit hiding | |
CN101004831A (en) | Methof for embedding and extracting watermark based on statistical model of coefficient in transform domain of digital images | |
JP4024153B2 (en) | Digital watermark embedding method and encoding device and decoding device capable of using the method | |
Singh et al. | Hybrid technique for robust and imperceptible dual watermarking using error correcting codes for application in telemedicine | |
CN1256705C (en) | Wavelet Domain Digital Watermarking Method Based on Image Target Area | |
CN109636707B (en) | Blind detection information hiding and extracting method based on image texture features | |
CN1141835C (en) | Method for protecting multimedia video data | |
Abdulaziz et al. | Robust data hiding for images | |
Zain | Strict authentication watermarking with JPEG compression (SAW-JPEG) for medical images | |
CN1414778A (en) | Watermark method using geometry calibrated anti geometry conversion image | |
Moloi et al. | Performance Analysis for Reed Solomon codes and Bose Chaudhuri-Hocquenghem codes in digital image watermarking | |
Akter et al. | Digital watermarking with a new algorithm | |
CN1529282A (en) | Method and device for unified digital watermarking based on singular value decomposition and spread spectrum technology | |
CN100344145C (en) | Lossless data embedding | |
Chen et al. | Digital image watermarking based on mixed error correcting code | |
Anjum et al. | Performance evaluation of DWT based image watermarking using error correcting codes | |
CN1620003A (en) | Anti-jamming Information Hiding Method Based on Turbo Code and Image Carrier | |
Hajjaji et al. | A digital watermarking algorithm based on quantization of the DCT: Application on medical imaging | |
Hazra et al. | A survey on different techniques for covert communication using steganography | |
Kurniawan et al. | Multiple watermarking on digital medical images for tamper detection and integrity control | |
Lan et al. | A digital watermarking algorithm based on dual-tree complex wavelet transform |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
C17 | Cessation of patent right | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20040310 |