CN105719225B - A kind of key recovery method of the LSB Matching steganography based on small echo absolute moment - Google Patents
A kind of key recovery method of the LSB Matching steganography based on small echo absolute moment Download PDFInfo
- Publication number
- CN105719225B CN105719225B CN201610037595.7A CN201610037595A CN105719225B CN 105719225 B CN105719225 B CN 105719225B CN 201610037595 A CN201610037595 A CN 201610037595A CN 105719225 B CN105719225 B CN 105719225B
- Authority
- CN
- China
- Prior art keywords
- small echo
- key
- subband
- absolute moment
- test path
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 22
- 238000011084 recovery Methods 0.000 title claims abstract description 12
- 238000012360 testing method Methods 0.000 claims abstract description 65
- 238000004364 calculation method Methods 0.000 claims description 16
- 238000001914 filtration Methods 0.000 claims description 12
- 238000000354 decomposition reaction Methods 0.000 claims description 8
- 230000009466 transformation Effects 0.000 claims description 6
- 238000001514 detection method Methods 0.000 description 8
- 238000003780 insertion Methods 0.000 description 7
- 230000037431 insertion Effects 0.000 description 7
- 238000005516 engineering process Methods 0.000 description 4
- 241001269238 Data Species 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000007476 Maximum Likelihood Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T1/00—General purpose image data processing
- G06T1/0021—Image watermarking
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2201/00—General purpose image data processing
- G06T2201/005—Image watermarking
- G06T2201/0065—Extraction of an embedded watermark; Reliable detection
Abstract
The invention discloses a kind of key recovery methods of LSB Matching steganography based on small echo absolute moment to perform the following steps in sequence each possible stego-key in key space or key dictionary respectively: A: setting steganography message length estimated valueL, according to each possible stego-keykGenerating length isLTest pathB: the remaining absolute average of small echo absolute moment of all positions in test path in image to be detected is calculated separatelyAnd it is located at the remaining absolute average of small echo absolute moment of all positions outside test path in image to be detected;C: it calculatesWithDifference;D: one or more maximum differences are chosenDoubtful stego-key of the corresponding stego-key as reduction.The present invention can determine doubtful stego-key under the premise of embedded location is only determined by stego-key.
Description
Technical field
The present invention relates to field of information security technology more particularly to a kind of LSB Matching based on small echo absolute moment are hidden
The key recovery method write.
Background technique
Steganography is to be embedded in secret information in the redundancy of the multi-medium datas such as image, video or audio, hidden to realize
Cover a kind of technology of communication.With the development of computer network and multimedia technology, steganography has become information security neck
One of the important technology in domain can be used for the secret communication between personal and enterprise.Concurrently, enterprise staff or government's duty
Internal information may also be hidden in using steganography and seem the multi-medium datas such as normal digital picture, audio or video by member
In leak out, with escape supervision.In numerous steganography methods, LSB matches steganography since it realizes that simple and concealment is strong
The characteristics of, expanded to a variety of different Digital Media, and develop many modified versions to differ from one another.Therefore, right
The evidence obtaining of LSB matching steganography has become one of the research emphasis in evidence obtaining steganalysis direction.
Currently, LSB matches the research of steganography evidence obtaining aspect, not only includes the hidden close image detection of LSB matching steganography, also wraps
Include the estimation of insertion rate and hidden close location estimation of LSB matching steganography.Wherein, the hidden close image detection that LSB matches steganography is current
LSB matches the emphasis that steganography evidence obtaining aspect is studied, and detection method can be divided mainly into two classes: detection method based on generic features and
Detection method based on special characteristic.The former mainly utilizes some general blind Detecting feature (such as co-occurrence matrix) training classifiers real
The hidden close image detection of existing LSB matching steganography, the latter mainly utilize some specific Stego-detection features for LSB matching steganography
(such as histogram feature function and small echo absolute moment) realizes detection.The main method of the insertion rate estimation of LSB matching steganography has most
The maximum-likelihood estimation technique, the method based on adjacent pixel to transfer and method based on machine learning etc..LSB matches the hidden close of steganography
Location estimation method mainly has the hidden close location estimation method based on Bayes and the hidden close location estimation based on small echo absolute moment
Method.
The above method enriches the means of LSB matching steganography evidence obtaining, is mentioned to prevent LSB from matching steganography by criminal's abuse
Technical support is supplied.However, the final purpose of steganography evidence obtaining is to restore the secret information of insertion.And many steganography softwares are embedding
A certain number of pixels will be chosen when entering information according to stego-key with embedding information.Therefore, how correctly to restore steganography close
Key has become a vital link in LSB matching steganography evidence obtaining.
Summary of the invention
The object of the present invention is to provide a kind of key recovery method of LSB Matching steganography based on small echo absolute moment,
Can be in known embedded location selection mechanism, and stego-key is differentiated under the premise of embedded location is only determined by stego-key
The true and false, and determine doubtful stego-key.
The present invention adopts the following technical solutions:
A kind of key recovery method of the LSB Matching steganography based on small echo absolute moment, for key space or close
Each possible stego-key in key dictionary, perform the following steps in sequence respectively:
A: setting steganography message length estimated value L, using known embedded location selection mechanism, according to each possible hidden
It writes key k and generates the test path Path that length is Lk,Pathk=i1i2…iL, wherein i1, i2 ..., iL are respectively according to can
Can stego-key k obtain the 1,2nd ..., L possible embedded locations;
B: the small echo absolute moment for calculating each position in image to be detected using small echo absolute moment residue calculation method is remaining,
Then the remaining absolute average of small echo absolute moment of all positions in test path in image to be detected is calculated separatelyAnd it is located at the remaining absolute average of small echo absolute moment of all positions outside test path in image to be detected
C: the remaining absolute average R of small echo absolute moment for all positions being located in test path is calculatedkIt is surveyed with being located at
Try the remaining absolute average R of small echo absolute moment of all positions outside pathk' difference dk,
D: the difference d corresponding to the multiple possible stego-keys being calculatedkIn, it is maximum to choose one or more
Difference dkDoubtful stego-key of the corresponding stego-key as reduction.
In the step B, the remaining absolute average of small echo absolute moment of all positions in test path
Calculation method it is as follows:
B11: carrying out level-one wavelet decomposition to image to be detected using filter, obtains low frequency sub-band, horizontal subband H, hangs down
Straight subband V and diagonal subband D;
B12: it is calculated separately based on N number of various sizes of window each in horizontal subband H, vertical subband V and diagonal subband D
The maximum a-posteriori estimation v of the local variance of position ii, calculation formula is as follows:
WhereinIndicate all coefficients in N × N neighborhood in wavelet sub-band W centered on the coefficient of i-th of position
Mean value of square, W ∈ { H, V, D },
B13: setting 0 for low frequency sub-band, carries out to each coefficient of horizontal subband H, vertical subband V and diagonal subband D
Quasi-Wiener filtering:
Wherein, RW,iIndicate the coefficient of i-th of position by filtered wavelet sub-band W, WiSmall echo before indicating filtering
The coefficient of i-th of position of subband W;
B14: by coefficients R obtained in step B13W,iInverse wavelet transform to airspace, using the value of position each after transformation as
The small echo absolute moment residue r of corresponding positioni;
B15: the remaining absolute average of small echo absolute moment of all positions in test path is acquired
In the step B, the remaining absolute average of small echo absolute moment of all positions outside test path
Calculation method it is as follows:
B21: carrying out level-one wavelet decomposition to image to be detected using filter, obtains low frequency sub-band, horizontal subband H, hangs down
Straight subband V and diagonal subband D;
B22: it is calculated separately based on N number of various sizes of window each in horizontal subband H, vertical subband V and diagonal subband D
The maximum a-posteriori estimation v of the local variance of position ii, calculation formula is as follows:
WhereinIndicate all coefficients in N × N neighborhood in wavelet sub-band W centered on the coefficient of i-th of position
Mean value of square, W ∈ { H, V, D },
B23: setting 0 for low frequency sub-band, carries out to each coefficient of horizontal subband H, vertical subband V and diagonal subband D
Quasi-Wiener filtering:
Wherein, wherein RW,iIndicate the coefficient of i-th of position by filtered wavelet sub-band W, WiBefore indicating filtering
Wavelet sub-band W i-th of position coefficient;
B24: by coefficients R obtained in step B23W,iInverse wavelet transform to airspace, using the value of position each after transformation as
The small echo absolute moment residue r of corresponding positioni;
B25: the remaining absolute average of small echo absolute moment of all positions outside test path is acquired
Level-one wavelet decomposition is carried out to image to be detected using 8-tap Daubechies filter.
Each position in horizontal subband H, vertical subband V and diagonal subband D is calculated separately using 4 various sizes of windows
The maximum a-posteriori estimation v of the local variance of ii。
The present invention firstly generates the test road that length is L to each possible key in key space or key dictionary
Then diameter calculates separately all positions in test path in image to be detected using small echo absolute moment residue calculation method
The remaining absolute average of small echo absolute momentAnd it is located at the small echo of all positions outside test path in image to be detected
The remaining absolute average of absolute momentFinally according to their differenceIt determines and differentiates the true of stego-key
Puppet greatly improves the efficiency and accuracy of the judgement of the stego-key true and false.
Detailed description of the invention
Fig. 1 is flow diagram of the invention.
Specific embodiment
The present invention is made with detailed description below in conjunction with drawings and examples:
As shown in Figure 1, the key recovery method of the LSB Matching steganography of the present invention based on small echo absolute moment,
For each possible stego-key in key space or key dictionary, it perform the following steps in sequence respectively:
A: setting steganography message length estimated value L, using known embedded location selection mechanism, according to each possible hidden
It writes key k and generates the test path Path that length is Lk,Pathk=i1i2…iL, wherein i1,i2,…,iLRespectively according to possible
Stego-key k obtain the 1,2nd ..., L possible embedded locations.
B: the small echo absolute moment for calculating each position in image to be detected using small echo absolute moment residue calculation method is remaining,
Then the remaining absolute average of small echo absolute moment of all positions in test path in image to be detected is calculated separatelyAnd it is located at the remaining absolute average of small echo absolute moment of all positions outside test path in image to be detected
In the step B, the remaining absolute average of small echo absolute moment of all positions in test path
Calculation method it is as follows:
B11: carrying out level-one wavelet decomposition to image to be detected using filter, obtains low frequency sub-band, horizontal subband H, hangs down
Straight subband V and diagonal subband D;8-tap Daubechies filter can be used in filter.
B12: it is calculated separately based on N number of various sizes of window each in horizontal subband H, vertical subband V and diagonal subband D
The maximum a-posteriori estimation v of the local variance of position ii, calculation formula is as follows:
WhereinIndicate all coefficients in N × N neighborhood in wavelet sub-band W centered on the coefficient of i-th of position
Mean value of square, W ∈ { H, V, D },In the present embodiment, 4 various sizes of windows can be used.
B13: setting 0 for low frequency sub-band, carries out to each coefficient of horizontal subband H, vertical subband V and diagonal subband D
Quasi-Wiener filtering:
Wherein, RW,iIndicate the coefficient of i-th of position by filtered wavelet sub-band W, WiSmall echo before indicating filtering
The coefficient of i-th of position of subband W;
B14: by coefficients R obtained in step B13W,iInverse wavelet transform to airspace, using the value of position each after transformation as
The small echo absolute moment residue r of corresponding positioni;
B15: the remaining absolute average of small echo absolute moment of all positions in test path is acquired
In the step B, the remaining absolute average of small echo absolute moment of all positions outside test path
Calculation method it is as follows:
B21: carrying out level-one wavelet decomposition to image to be detected using filter, obtains low frequency sub-band, horizontal subband H, hangs down
8-tap Daubechies filter can be used in straight subband V and diagonal subband D, filter;
B22: it is calculated separately based on N number of various sizes of window each in horizontal subband H, vertical subband V and diagonal subband D
The maximum a-posteriori estimation v of the local variance of position ii, calculation formula is as follows:
WhereinIndicate all coefficients in N × N neighborhood in wavelet sub-band W centered on the coefficient of i-th of position
Mean value of square, W ∈ { H, V, D },In the present embodiment, 4 various sizes of windows can be used.
B23: setting 0 for low frequency sub-band, carries out to each coefficient of horizontal subband H, vertical subband V and diagonal subband D
Quasi-Wiener filtering:
Wherein, RW,iIndicate the coefficient of i-th of position by filtered wavelet sub-band W, WiSmall echo before indicating filtering
The coefficient of i-th of position of subband W;
B24: by coefficients R obtained in step B23W,iInverse wavelet transform to airspace, using the value of position each after transformation as
The small echo absolute moment residue r of corresponding positioni;
B25: the remaining absolute average of small echo absolute moment of all positions outside test path is acquired
C: the remaining absolute average of small echo absolute moment for all positions being located in test path is calculatedIt is surveyed with being located at
Try the remaining absolute average of small echo absolute moment of all positions outside pathDifference dk,
D: the difference d corresponding to the multiple possible stego-keys being calculatedkIn, it is maximum to choose one or more
Difference dkDoubtful stego-key of the corresponding stego-key as reduction.
It is absolute due to being not embedded into the remaining absolute mean of small echo absolute moment of information position and the small echo of embedding information position
There are significant differences for the remaining absolute mean of square, therefore the present invention is located at the small echo of all positions in test path by calculating
The remaining absolute average of absolute momentIt is remaining absolutely average with the small echo absolute moment for all positions being located at outside test path
ValueDifference dk, carry out the judgement of the stego-key true and false.
For pseudo- stego-key, randomly selected in hidden close image since the test path generated in step A is equivalent to, institute
Ratio shared by the position of information and the outer embedding information of test path are embedded in the test path of generation with pseudo- stego-key
Position shared by ratio it is approximately equal with the insertion rate in whole picture image to be detected.Therefore, for pseudo- stego-key, road is tested
In diameter and the difference of ratio shared by the position of the outer embedding information of test path is approximately 0.
And for true stego-key, then there is following three situation:
1) when the test path length of generation is less than embedding information length, all positions on test path are respectively positioned on embedding
Entering in the position of information, i.e. the ratio that the position of embedding information accounts for test path is 1, and then comprising being partially submerged into outside test path
The position of information, the ratio that the position for being partially submerged into information outside test path accounts for test path external position quantity are less than whole
Insertion rate p (0 < p < 1) in width image.Therefore, ratio shared by the position of embedding information in test path and outside test path
Difference is greater than 1-p.
2) when the test path length of generation is equal to embedding information length, all positions on test path are respectively positioned on embedding
Enter in the position of information, i.e. the ratio that the position of embedding information accounts for test path is 1, and outside test path does not include any insertion
It is 0 that insertion information bit, which sets and accounts for the ratio of test path external position quantity, outside the position of information, i.e. test path.Therefore, road is tested
In diameter and the difference of ratio shared by the position of the outer embedding information of test path is equal to 1.
3) when the test path length of generation is greater than embedding information length, comprising all embedding informations on test path
Position, therefore the position of embedding information accounts for the p that is greater than of the ratio of test path, and outside test path do not include any embedding information
Position, i.e., be embedded in information bit outside test path and set that account for the ratio of test path external position quantity be 0.Therefore, in test path
The difference of ratio shared by position with embedding information outside test path should be greater than p.
In conclusion for true stego-key, ratio shared by the interior position with the outer embedding information of test path of test path
Difference be greater than 0 certainly, this makes the remaining absolute average of small echo absolute moment in test pathIt is likely to be greater than test road
The remaining absolute average of the outer small echo absolute moment of diameterTherefore, the present invention can be remaining according to small echo absolute moment in test path
Absolute averageWith the remaining absolute average of small echo absolute moment outside test pathDifference dkIt is true to carry out stego-key
Pseudo- judgement, is inferred to doubtful stego-key.
Claims (5)
1. a kind of key recovery method of the LSB Matching steganography based on small echo absolute moment, which is characterized in that for key sky
Between or key dictionary in each possible stego-key, perform the following steps in sequence respectively:
A: being arranged steganography message length estimated value L, close according to each possible steganography using known embedded location selection mechanism
Key k generates the test path Path that length is Lk,Pathk=i1i2…iL, wherein i1,i2,…,iLRespectively according to possible hidden
The key k is obtained the 1,2nd is write ..., L possible embedded locations;
B: the small echo absolute moment for calculating each position in image to be detected using small echo absolute moment residue calculation method is remaining, then
Calculate separately the remaining absolute average of small echo absolute moment for being located at all positions in image to be detected in test pathWith
And it is located at the remaining absolute average of small echo absolute moment of all positions outside test path in image to be detected
C: the remaining absolute average of small echo absolute moment for all positions being located in test path is calculatedRoad is tested with being located at
The remaining absolute average of small echo absolute moment of all positions outside diameterDifference dk,
D: the difference d corresponding to the multiple possible stego-keys being calculatedkIn, choose one or more maximum difference dk
Doubtful stego-key of the corresponding stego-key as reduction.
2. the key recovery method of the LSB Matching steganography according to claim 1 based on small echo absolute moment, feature
It is, in the step B, the remaining absolute average of small echo absolute moment of all positions in test path's
Calculation method is as follows:
B11: carrying out level-one wavelet decomposition to image to be detected using filter, obtains low frequency sub-band, horizontal subband H, vertical son
Band V and diagonal subband D;
B12: each position in horizontal subband H, vertical subband V and diagonal subband D is calculated separately based on N number of various sizes of window
The maximum a-posteriori estimation v of the local variance of ii, calculation formula is as follows:
WhereinIndicate square of all coefficients in N × N neighborhood in wavelet sub-band W centered on the coefficient of i-th of position
Mean value, W ∈ { H, V, D },
B13: setting 0 for low frequency sub-band, carries out quasi- to each coefficient of horizontal subband H, vertical subband V and diagonal subband D
Wiener filtering:
Wherein, RW,iIndicate the coefficient of i-th of position by filtered wavelet sub-band W, WiWavelet sub-band W before indicating filtering
I-th of position coefficient;
B14: by coefficients R obtained in step B13W,iInverse wavelet transform is to airspace, using the value of position each after transformation as correspondence
The small echo absolute moment residue r of positioni;
B15: the remaining absolute average of small echo absolute moment of all positions in test path is acquired
3. the key recovery method of the LSB Matching steganography according to claim 1 based on small echo absolute moment, feature
It is, in the step B, the remaining absolute average of small echo absolute moment of all positions outside test pathMeter
Calculation method is as follows:
B21: carrying out level-one wavelet decomposition to image to be detected using filter, obtains low frequency sub-band, horizontal subband H, vertical son
Band V and diagonal subband D;
B22: each position in horizontal subband H, vertical subband V and diagonal subband D is calculated separately based on N number of various sizes of window
The maximum a-posteriori estimation v of the local variance of ii, calculation formula is as follows:
WhereinIndicate square of all coefficients in N × N neighborhood in wavelet sub-band W centered on the coefficient of i-th of position
Mean value, W ∈ { H, V, D },
B23: setting 0 for low frequency sub-band, carries out quasi- to each coefficient of horizontal subband H, vertical subband V and diagonal subband D
Wiener filtering:
Wherein, wherein RW,iIndicate the coefficient of i-th of position by filtered wavelet sub-band W, WiSmall echo before indicating filtering
The coefficient of i-th of position of subband W;
B24: by coefficients R obtained in step B23W,iInverse wavelet transform is to airspace, using the value of position each after transformation as correspondence
The small echo absolute moment residue r of positioni;
B25: the remaining absolute average of small echo absolute moment of all positions outside test path is acquired
4. the key recovery method of the LSB Matching steganography according to claim 2 or 3 based on small echo absolute moment,
It is characterized in that: level-one wavelet decomposition is carried out to image to be detected using 8-tap Daubechies filter.
5. the key recovery method of the LSB Matching steganography according to claim 2 or 3 based on small echo absolute moment,
It is characterized in that: calculating separately each position in horizontal subband H, vertical subband V and diagonal subband D using 4 various sizes of windows
Set the maximum a-posteriori estimation v of the local variance of ii。
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN2015110216914 | 2015-12-31 | ||
CN201511021691 | 2015-12-31 |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105719225A CN105719225A (en) | 2016-06-29 |
CN105719225B true CN105719225B (en) | 2018-12-11 |
Family
ID=56148005
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610037595.7A Active CN105719225B (en) | 2015-12-31 | 2016-01-20 | A kind of key recovery method of the LSB Matching steganography based on small echo absolute moment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105719225B (en) |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109995503B (en) * | 2019-03-11 | 2022-03-18 | 青岛大学 | Electronic grid construction method based on secret key |
CN111047497B (en) * | 2019-12-24 | 2022-10-04 | 中国人民解放军战略支援部队信息工程大学 | JPEG image steganography information positioning method based on same-frequency sub-image filtering |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1737819A (en) * | 2005-08-29 | 2006-02-22 | 上海师范大学 | Universal digital image invisible information detecting method |
USRE40477E1 (en) * | 2001-06-22 | 2008-09-02 | The Research Foundation Of Suny | Reliable detection of LSB steganography in color and grayscale images |
CN102724041A (en) * | 2012-06-07 | 2012-10-10 | 北京航空航天大学 | Steganography-based key transmission and key updating method |
CN104008521A (en) * | 2014-05-29 | 2014-08-27 | 西安理工大学 | LSB replacement steganalysis method based on grey co-occurrence matrix statistic features |
-
2016
- 2016-01-20 CN CN201610037595.7A patent/CN105719225B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
USRE40477E1 (en) * | 2001-06-22 | 2008-09-02 | The Research Foundation Of Suny | Reliable detection of LSB steganography in color and grayscale images |
CN1737819A (en) * | 2005-08-29 | 2006-02-22 | 上海师范大学 | Universal digital image invisible information detecting method |
CN102724041A (en) * | 2012-06-07 | 2012-10-10 | 北京航空航天大学 | Steganography-based key transmission and key updating method |
CN104008521A (en) * | 2014-05-29 | 2014-08-27 | 西安理工大学 | LSB replacement steganalysis method based on grey co-occurrence matrix statistic features |
Non-Patent Citations (3)
Title |
---|
"Feature Reduction and Payload Location with WAM Steganalysis";Andrew D.Ker.et al;《Proceedings of SPIE-The International Society for Optical Engineering》;20090228;第1-8页 * |
"图像隐写定量与定位分析关键问题研究";杨春芳;《中国博士学位论文全文数据库 信息科技辑》;20130615(第06期);I138-60 * |
"针对图像空域随机LSB信息隐藏的提取攻击策略";王丽娜;《武汉大学学报(理学版)》;20150228;第61卷(第01期);第42-48页 * |
Also Published As
Publication number | Publication date |
---|---|
CN105719225A (en) | 2016-06-29 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105095856B (en) | Face identification method is blocked based on mask | |
CN106530200B (en) | Steganographic image detection method and system based on deep learning model | |
Zimba et al. | DWT-PCA (EVD) based copy-move image forgery detection | |
CN110675328B (en) | Low-illumination image enhancement method and device based on condition generation countermeasure network | |
CN113807276B (en) | Smoking behavior identification method based on optimized YOLOv4 model | |
Verma et al. | An overview of robust digital image watermarking | |
CN110033040B (en) | Flame identification method, system, medium and equipment | |
CN109584162B (en) | Image super-resolution reconstruction method based on generation network | |
CN109977832B (en) | Image processing method, device and storage medium | |
AU2011265429A1 (en) | Method and system for robust scene modelling in an image sequence | |
CN109960975B (en) | Human face generation and human face recognition method based on human eyes | |
CN109711309B (en) | Method for automatically identifying whether portrait picture is eye-closed | |
CN112052830B (en) | Method, device and computer storage medium for face detection | |
CN108650491A (en) | A kind of video watermark detection method towards monitoring system | |
CN105719225B (en) | A kind of key recovery method of the LSB Matching steganography based on small echo absolute moment | |
CN109460705A (en) | Oil pipeline monitoring method based on machine vision | |
CN110349134A (en) | A kind of piping disease image classification method based on multi-tag convolutional neural networks | |
WO2019228450A1 (en) | Image processing method, device, and equipment, and readable medium | |
CN110930384A (en) | Crowd counting method, device, equipment and medium based on density information | |
CN111815529B (en) | Low-quality image classification enhancement method based on model fusion and data enhancement | |
CN110674689B (en) | Vehicle re-identification method and system based on feature embedding space geometric constraint | |
CN108122209A (en) | A kind of car plate deblurring method based on confrontation generation network | |
CN115358952B (en) | Image enhancement method, system, equipment and storage medium based on meta-learning | |
CN108305207B (en) | Airspace image steganalysis credibility evaluation method | |
CN116645258A (en) | Watermark embedding, watermark extraction, watermark identification model construction method and electronic equipment |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
TR01 | Transfer of patent right |
Effective date of registration: 20201023 Address after: 450001 No. 62 science Avenue, hi tech Zone, Henan, Zhengzhou Patentee after: Information Engineering University of the Chinese People's Liberation Army Strategic Support Force Address before: 405, 450001, 62, science Avenue, Zhengzhou hi tech Development Zone, Henan, China Patentee before: Yang Chunfang |
|
TR01 | Transfer of patent right |