CN100416595C - Blind detecting system and method for digital watermarking flooding - Google Patents
Blind detecting system and method for digital watermarking flooding Download PDFInfo
- Publication number
- CN100416595C CN100416595C CNB2006101142984A CN200610114298A CN100416595C CN 100416595 C CN100416595 C CN 100416595C CN B2006101142984 A CNB2006101142984 A CN B2006101142984A CN 200610114298 A CN200610114298 A CN 200610114298A CN 100416595 C CN100416595 C CN 100416595C
- Authority
- CN
- China
- Prior art keywords
- detection
- algorithm
- linear dependence
- watermark
- detects
- 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
- 238000000034 method Methods 0.000 title claims abstract description 32
- 238000001514 detection method Methods 0.000 claims abstract description 190
- 238000004458 analytical method Methods 0.000 claims abstract description 30
- 230000004927 fusion Effects 0.000 claims abstract description 11
- 238000012360 testing method Methods 0.000 claims description 41
- 230000008569 process Effects 0.000 claims description 23
- 239000013598 vector Substances 0.000 claims description 13
- 230000000295 complement effect Effects 0.000 claims description 8
- 238000001914 filtration Methods 0.000 claims description 7
- 239000000284 extract Substances 0.000 claims description 4
- 239000011159 matrix material Substances 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 4
- 239000012141 concentrate Substances 0.000 claims description 3
- 238000002386 leaching Methods 0.000 claims description 3
- 230000009977 dual effect Effects 0.000 claims 1
- 238000013507 mapping Methods 0.000 claims 1
- 238000007500 overflow downdraw method Methods 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 5
- 230000006870 function Effects 0.000 description 5
- 238000011156 evaluation Methods 0.000 description 4
- 230000001427 coherent effect Effects 0.000 description 3
- 238000007689 inspection Methods 0.000 description 3
- 238000005259 measurement Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 238000003786 synthesis reaction Methods 0.000 description 3
- 238000012795 verification Methods 0.000 description 3
- 230000002087 whitening effect Effects 0.000 description 3
- 230000003321 amplification Effects 0.000 description 2
- 238000010835 comparative analysis Methods 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 238000003199 nucleic acid amplification method Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000010998 test method Methods 0.000 description 2
- 238000012384 transportation and delivery Methods 0.000 description 2
- 208000035126 Facies Diseases 0.000 description 1
- 238000007630 basic procedure Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000012467 final product Substances 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000013439 planning Methods 0.000 description 1
- 239000000047 product Substances 0.000 description 1
Images
Landscapes
- Editing Of Facsimile Originals (AREA)
- Image Processing (AREA)
Abstract
A general blind detection system for digital watermark and method, which contains a number of detecting algorithm software, such as the order detection software on the digital images, the fusion analysis software according to the result of these detection algorithm, and the detection rule database software which is compared with these algorithm and fusion analysis. In the condition of on original watermark, it can use the detection algorithm according to the embedding algorithm detects several different embedding ones, and make fusion analysis on the result by the rule, to detect the watermark of digital image.
Description
Technical field
The present invention relates to a kind of digital watermarking flooding blind-detection system and detection method, belong to information security field,
Background technology
Digital watermark technology adds the data of needs in the carrier data to, can detect wherein information at any time by certain means.Therefore, this technical requirement information in the carrier of adding to do not destroy carrier data normal use this be the hidden property requirement of digital watermark.Equally, carrier data may be through overcompression so that transmission, various signal Processing are to adapt to various demands, therefore, the watermark information of this moment energy equally is detected, promptly digital watermark also should have robustness.Through many researchers' hard effort in the past few years, find that digital watermark technology must have the one-way requirement, otherwise this digital watermark can't prove that to correctness, the uniqueness that is hidden in the information in the carrier data promptly the authenticity of information is suspicious at this moment.Subsequently, a kind of watermark detection technique based on the statistical dependence analysis has changed this situation.Detected carrier data is only used in the detection of this technology, if contain appointed information in the data, and the correctness that can hide Info really then.This technique guarantee the one-way of testing process.Current main detection technique is based on the statistical dependence analysis mostly and comes, and puts forth effort on hidden property, robustness and the one-way that guarantees digital watermarking.The digital watermarking flooding blind-detection system is under the jurisdiction of a kind of like this detection technique field.
The detection technique of digital watermarking exists many algorithms and scheme at present, relatively more classical have coherent detection and related coefficient detection algorithm and a statistical study detection algorithm, be directed to detection algorithm its to embed algorithm also inequality, and be directed to the used detection algorithm of various algorithms self and all deepened continuously at the one's best quality that shows aspect robustness and the invisibility.
Extensive blind Detecting technology has following several form: sequence detection method, characteristic analysis method, data fusion method, dedicated test method etc.The sequence detection method is in the long as far as possible time, adopts multiple detection algorithm to detect successively; The characteristic analysis method is exactly directly to use some detection algorithms, fast detecting according to the characteristics of image; The data fusion method is on the detection basis of multiple detection algorithm, carries out convergence analysis for the result, thereby obtains testing result accurately; The dedicated test method is set up special-purpose fast detection algorithm at the image watermark of certain kind.
The watermark detection analysis is carried out according to image intrinsic feature and embedding algorithm usually, and data are carried out statistical study.Existing digital watermarking detection technique is pointed for different embedded objects and embedding algorithm, and there is not a unified standard criterion algorithm, multimedia data information in the face of magnanimity when being used for the detection of internet does not often possess extensive characteristic, and the detection that has now based on original image has a lot of restrictions for detecting high efficiency and availability.
Summary of the invention
The purpose of this invention is to provide a kind of digital watermarking flooding blind-detection system and detection method, set up a complete extendible digital watermarking flooding blind-detection system, thereby can carry out extensive blind Detecting for the watermark carrier that different known embedding algorithm process is crossed, and the robustness of check algorithm, reliability is carried out safety evaluation to existing algorithm.
The technical solution adopted for the present invention to solve the technical problems is: a kind of digital watermarking flooding blind-detection system includes linear dependence detection module, albefaction linear dependence detection module, block-based linear dependence detection module, block-based albefaction linear dependence detection module, Viterbi detection module, detects rule database and data fusion analysis module.
After the reading images with each pixel and template acquiring correlative value, the linear dependence detection module that compares with threshold value then;
What adopt is that 11 * 11 matrix carries out the filtering convolution algorithm to original image, mainly concentrates in the image boundary, is limited to the albefaction linear dependence detection module of the computing within image and the template pixel coverage;
Carry out watermark extracting, earlier content is carried out some pre-service, produce a vector in the sign space, dimension is lower than original vector, determine again whether the sign that has extracted comprises watermark information, carry out the inverse process of leaching process, new vector is mapped to Channel Space again, thereby obtained adding the works of watermark; Or carry out treatment of picture, and the piece that extracts 8*8 adds up then and averages, and the piece that is directed to 8*8 then carries out the block-based linear dependence detection module that watermark embeds;
Testing result is at the block-based blind embedding algorithm of algorithm, utilize the block-based watermarking algorithm of fixedly normalization linear dependence and have block-based watermarking algorithm that fixing robustness embeds to embed and errorlessly to detect in 1 o'clock, the block-based albefaction linear dependence detection module of false dismissal when detecting embedding 0 value, occurs;
Adopt the mode of delivery complementation to travel through 8 states, in the template that generates, reduce to the related coefficient of different templates minimum simultaneously, thereby can under the restriction of detection threshold, accurately judge watermark information, when related coefficient absolute value during greater than the threshold value thresholding, judge that then wherein corresponding Template Information is 1, when related coefficient absolute value during less than the threshold value thresholding, judge that then wherein corresponding Template Information is 0, be weighted the Viterbi detection module of the specifying information of expression embedding by corresponding binary digit;
The detection rule database that detection threshold, relevant parameters and the testing result of single detection algorithm in this in store system are arranged;
There is at system employs single detection algorithm wherein image is traveled through when detecting the data fusion analysis module that the testing result and the rule that detects in the rule database of each detection algorithm compared.If be complementary, show to detect digital watermarking with certain rule; If each detection algorithm and strictly all rules all do not match, show not detect digital watermarking.In order to reduce the alert rate of mistake, may need many rules to come in the process decision chart picture whether under cover digital watermarking sometimes.
The known testing process that the present invention is based on the coherent detection algorithm combines, and at first by utilizing several known algorithms to set up unified detection model, then its amplification is expanded as a feasible detection system.Following five kinds of detection algorithms are mainly adopted in the foundation of extensive detection model, and linear dependence detects, and block-based linear dependence detects, and the albefaction linear dependence detects, and block-based albefaction linear dependence detects and Viterbi detection; These five kinds of algorithms are detection algorithms of the function admirable of selecting out from 13 kinds of detection algorithms, can reach the purpose that low false dismissal detects for detecting the mistake alarm probability.
When whether having watermark information in adjudicating detected object, adopt the database matching comparative analysis, the array information in the database is to verify by experiment and the obtaining of posterior probability analysis.When the testing result of a certain row embedding algorithm in testing result and the database is complementary, can judge that then this detected object is the watermark of algorithm embedding thus; If the database association rule that embeds with no watermark is complementary then judges no watermark existence in the detected object.The accuracy that detects depends on the mistake alarm probability of the extensive blind Detecting of image digital watermark, and the capacity of database changes according to the expansion of detection algorithm, as a kind of system that can expand, current basic model framework is discussed.
A kind of digital watermarking flooding blind checking method has the convergence analysis method of the testing result of a plurality of detection algorithms, and the detection rule database is arranged, and for different embedding algorithms, selects the relevant detection algorithm to extract watermark information under the situation that does not need original image;
May further comprise the steps:
Step 1, there not being under the original work watermark situation to use the detection algorithm detection corresponding with embedding algorithm, a plurality of different embedding algorithms are detected;
Step 2, the testing result of algorithms of different is carried out convergence analysis according to detecting rule;
Described step 1 comprises: linear dependence detects, the albefaction linear dependence detects, block-based linear dependence detects, block-based albefaction linear dependence detects and Viterbi detection;
Described step 2 comprises: the combination and the flow process that detect rule database and convergence analysis method.
Beneficial effect of the present invention:
Utilize extensive blind-detection system can be directed to picture in the network, even expansion can effectively be monitored it to Digital Medias such as audio frequency and videos, utilize known detection algorithm to carry out the modeling optimization system, and it is carried out safety evaluation, the purpose of safety evaluation is to test the robustness of detection algorithm, makes it be suitable for the basic demand of extensive blind Detecting.
The present invention adopts sequence detection and data fusion method, utilizing multiple detection algorithm to treat detected object detects, remedied the unicity of existing different detection algorithms, characteristics with the alert rate of low mistake, and has an extensibility, can add new detection algorithm, be adapted to the continuous variation of current algorithm and application, make this model have higher verification and measurement ratio and range of application widely.
The present invention is based on existing detection algorithm and screens, the detection model that combination and optimization obtain, and its advantage is can be with different embedding algorithm synthesis relatively, and the detection algorithm of selecting to be fit to finally extracts watermark information.
Principal feature of the present invention is to merge multiple detection technique, forms the testing mechanism of unified standard; And the possible outcome that will detect information is included the database schema management in, the extensive blind-detection system of digital watermarking is constantly perfect as an extendible model, and effective canned data capacity of database does not need very big, is very suitable for current requirement from the conserve space angle.
Description of drawings
Fig. 1 digital watermarking flooding blind-detection system;
Fig. 2 linear dependence testing process figure;
Fig. 3 albefaction linear dependence testing process figure;
The block-based linear dependence testing process of Fig. 4 figure;
The block-based albefaction linear dependence of Fig. 5 testing process figure;
Fig. 6 Viterbi detection process flow diagram.
Embodiment
A kind of digital watermarking flooding blind checking method, at first, treating detected object uses linear dependence detection algorithm, albefaction linear dependence detection algorithm, block-based linear dependence detection algorithm, block-based albefaction linear dependence detection algorithm, viterbi-detection-algorithm to detect under the prerequisite of embedment strength and detection threshold assurance verification and measurement ratio optimum, the testing result of each detection algorithm and the rule in the database are compared, if be complementary with certain rule, show to detect digital watermarking, withdraw from; If each detection algorithm and strictly all rules all do not match, show not detect digital watermarking.
Adopt the sequence detection method to detect for image to be detected, utilize data fusion method analysis-by-synthesis testing result then, judge watermark information wherein at last, and planning becomes corresponding array database according to the testing result analysis-by-synthesis, and and then testing result compared judgement, thereby seek out corresponding detection algorithm.The core missions of extensive blind-detection system are the rule of testing result and database is compared judgement.
System of the present invention constitutes as Fig. 1: include a plurality of detection modules, these detection modules are detected digital picture successively, the convergence analysis module that adapts with the testing result of these detection modules is with the detection regular data library module of these detection modules and convergence analysis module facies analysis contrast.This system can pass through to increase new detection algorithm or reduce detection algorithm, and sequence detection module and convergence analysis module are made amendment with the expansion or the modification of realization system with detection regular data library module.
A kind of digital watermarking flooding blind-detection system includes linear dependence detection module, albefaction linear dependence detection module, linear dependence detection module, block-based albefaction linear dependence detection module, Viterbi detection module, detects rule database and data fusion analysis module.Can also increase other or new module according to the development of the needs of detected image or detection algorithm software.
Detection method of the present invention selects the relevant detection algorithm to extract watermark information under the situation that does not need original image at different embedding algorithms.
Be divided into the following step of detection method of the present invention:
Steps A 1, there not being under the original work watermark situation to use the detection algorithm detection corresponding with embedding algorithm, a plurality of different embedding algorithms are detected;
Steps A 2, the testing result of algorithms of different is carried out convergence analysis according to detecting rule;
Described steps A 1 comprises: linear dependence detects, the albefaction linear dependence detects, block-based linear dependence detects, block-based albefaction linear dependence detects and Viterbi detection;
Described steps A 2 comprises: the combination and the flow process that detect rule database and convergence analysis method.
Detection method of the present invention specifically realizes by following process: treat detected object and use linear dependence detection algorithm, albefaction linear dependence detection algorithm, block-based linear dependence detection algorithm, block-based albefaction linear dependence detection algorithm, viterbi-detection-algorithm to detect under the prerequisite of embedment strength and detection threshold assurance verification and measurement ratio optimum; The testing result of each detection algorithm and the rule in the database are compared,, show to detect digital watermarking, withdraw from if be complementary with certain rule; If each detection algorithm and strictly all rules all do not match, show not detect digital watermarking.
In extensive blind-detection system of the present invention, comprise following detection algorithm:
Linear dependence detects, as shown in Figure 2, the basic step that linear dependence detects be after the reading images with each pixel and template acquiring correlative value, compare with threshold value then.Its improvement mainly is embodied in the selection of threshold value, can detect undistorted image and whether be embedded into watermark.When detecting that fixed linear is relevant to embed the embedded object of device, detection threshold is 0.0299, and and if only if, and wr*5 embeds, and detects (wr is the watermark template) with wr/10.
The albefaction linear dependence detects, as shown in Figure 3, particularly albefaction linear dependence detection algorithm adopts is that 11 * 11 matrix carries out the filtering convolution algorithm to original image, mainly concentrates in the image boundary, therefore is only limited to the computing within image and the template pixel coverage.
Block-based linear dependence detects, as shown in Figure 4, detecting device was divided into for three steps to carry out, and the first step is a watermark extracting, earlier content is carried out some pre-service, produce a vector in the sign space, dimension is lower than original vector, and second step determined whether the sign that has extracted comprises watermark information again, and the 3rd step was the inverse process of leaching process, new vector is mapped to Channel Space again, thereby has obtained adding the works of watermark.In the treatment of picture process, the piece that extracts 8*8 adds up then and averages, and the piece that is directed to 8*8 then carries out watermark and embeds.Extracting 64 dimensional vectors is key issues, and can tie up by extensive one-tenth 16, and forms such as 128 dimensions are mainly decided according to the size of image.
Block-based albefaction linear dependence detects, and as shown in Figure 5, the result that computing obtains makes wave filter be unlikely to surpass predetermined size.But testing result is at the block-based blind embedding algorithm of algorithm, utilize the block-based watermarking algorithm of fixedly normalization linear dependence and have block-based watermarking algorithm that fixing robustness embeds to embed and errorlessly to detect in 1 o'clock, false dismissal when detecting embedding 0 value, occurs.
Viterbi detection, as shown in Figure 6, Viterbi detection is that the mode of delivery complementation travels through 8 states this employing, rather than adopts the mode of grid coding, but its ultimate principle is consistent.The just easier realization of herein adopting of mode, and be convenient to understand.In the template that generates, also reduce to the related coefficient of different templates minimum simultaneously.Thereby can under the restriction of detection threshold, accurately judge watermark information.When related coefficient absolute value during greater than the threshold value thresholding, judge that then wherein corresponding Template Information is 1, when related coefficient absolute value during, judge that then wherein corresponding template information is 0, is weighted the specifying information that expression embeds by corresponding binary digit less than the threshold value thresholding.
Extensive blind-detection system and method among the present invention, according to the respective detection threshold value after optimizing, detected object is detected, utilize the data fusion method to carry out analyzing and processing then for testing result, classify and sums up according to its testing result that whether meets in the database rule, and adjudicate to draw whether there is digital watermarking.
Decision mechanism:
The known testing process that the present invention is based on the coherent detection algorithm combines, and at first by utilizing several known algorithms to set up unified detection model, then its amplification is expanded as a feasible detection system.Following five kinds of detection algorithms are mainly adopted in the foundation of extensive detection model, and linear dependence detects, and block-based linear dependence detects, and the albefaction linear dependence detects, and block-based albefaction linear dependence detects and Viterbi detection; These five kinds of algorithms are detection algorithms of the function admirable of selecting out from 13 kinds of detection algorithms, can reach the purpose that low false dismissal detects for detecting the mistake alarm probability.
When whether having watermark information in adjudicating detected object, adopt the database matching comparative analysis, the array information in the database is to verify by experiment and the obtaining of posterior probability analysis.When the testing result of a certain row embedding algorithm in testing result and the database is complementary, can judge that then this detected object is the watermark of algorithm embedding thus; If the database association rule that embeds with no watermark is complementary then judges no watermark existence in the detected object.The accuracy that detects depends on the mistake alarm probability of the extensive blind Detecting of image digital watermark, and the capacity of database changes according to the expansion of detection algorithm, as a kind of system that can expand, current basic model framework is discussed.
The safety evaluation of system:
Be directed to the detection mistake alarm probability of algorithm, mainly by adjusting for algorithm parameter and being optimized realization for algorithm.The parameter of extensive blind-detection system depends on single detection algorithm and improves.
In the last table:
The 1-linear dependence detects
3-albefaction linear dependence detects
The block-based linear dependence of 5-detects
The block-based albefaction linear dependence of 11-detects
The 12-Viterbi detection
The validity of digital watermarking system is generally weighed by just inspection rate, false alarm rate and false dismissed rate.The just inspection rate of extensive blind-detection system, false alarm rate and false dismissed rate are
False alarm rate: 1.818%
False dismissed rate: 2.727%
Just inspection rate: 95.455%=100%-1.818%-2.727%
Adopt 5 detection algorithms of the extensive blind-detection system of watermark to carry out sequence detection to digital picture, then testing result and database information are mated.The selection of the threshold value of each detection algorithm is as follows:
Linear dependence is surveyed the selection of detection threshold:
Use respective function in the time of enforcement and carry out the related coefficient that correlation value calculation is promptly calculated two-dimensional function, main characteristics is that it utilizes normalized effect.Detection threshold is chosen as 0.001 when being unlikely to image fault according to embedment strength.The embedment strength of different embedding algorithms is difference to some extent, but the basic demand that embeds is not make the distortion of image too serious, cause watermark information as seen.
The selection of albefaction linear dependence detection threshold:
The main optimal way that the albefaction linear dependence detects is that threshold value is chosen as 0.005 before the stack with image and given prewhitening filter, and the accuracy rate of detection this moment is than higher.Efficient with respect to detection threshold after the whitening filtering is higher.The probability of false dismissal is also just lower, and is higher when therefore the threshold ratio that adopts does not carry out filtering.
The selection of block-based linear dependence detection threshold:
The piece that 8*8 is at first extracted image to be detected in block-based computing carries out the related coefficient computing with the template of big or small 8*8 then.Setting detection threshold is 0.09, calculates related coefficient this moment and adopts the method for getting inner product after the normalization to realize.
The selection of block-based albefaction linear dependence detection threshold:
Especially in the process of block-based piecemeal computing, when remapping back Channel Space when extracting piece 64 dimensional vectors of 8*8 and with new vector, need be when the difference vector of each piece calculates in the piecemeal rank, cycle applications is then in pixel scale, and the related coefficient of then calculating between 64 dimensional vectors and the reference template during detection gets final product.Detection threshold is chosen as 0.2.
The selection of Viterbi detection threshold value:
Detection threshold is chosen as 0.005, and can distinguish the difference of embedding 0 and embedding 1 this moment.And embedment strength is 0.00225, and be unlikely and make former figure distortion too serious this moment, can errorlessly detect during detection.
When forming the extensive blind-detection system of digital watermarking, act on the spirit of above-mentioned algorithm substantially, carry out corresponding correction according to the rule that is fit to sequence detection, strive making it to detect the mistake alarm probability and reach minimum.And carry out replenishing of a key, adopted the data fusion mode to carry out matching detection.Before setting up database, carried out DCO, and be directed to the different situations that may occur and carry out forecast analysis for the randomness of detected object and digital watermarking and mistake alarm probability.To detect false-alarm probability and false dismissal probability reduces to minimum.Utilize known close algorithm to carry out sequence detection after setting up database for image to be detected, testing result with each algorithm is stored in the buffer memory then, then data in result and the database are mated calculating, final judgement shows in the detected object whether have watermark information.Provide foundation for further analyzing watermark information.And the check and analysis process when every correlation that detects respective algorithms meets the requirements, is then listed the suspection scope in, strives making it to detect false dismissed rate and reduces to minimum.
Basic procedure shown in Figure 1, can implement the function of extensive blind-detection system:
Step 1 linear dependence detects.
Step 2 detects for the albefaction linear dependence.Contrast step 1 is that it carried out whitening filtering before computing.
Step 3 is that block-based linear dependence detects.The correlativity of piecemeal arithmograph picture and watermark template.
Step 4 is that block-based albefaction linear dependence detects.Contrast step 3 is that it carried out whitening filtering before computing.
Step 5 is a Viterbi detection.Stack embeds 8 templates on the basis of above-mentioned steps.8 information have been reached during embedding.
Step 6 is the sequence detection method, adopts known algorithm to treat detected image and carries out the feature that sequence detection does not still change image.
Step 7 is may various case when detecting and definite detection matrix for detecting rule database, building the storehouse.
Step 8 is the data fusion analysis module, and the testing result of fusion 6 is analyzed, and mates differentiation alternately with the detection rule database, for meeting rule in 7, then can judge whether there is watermark information in the detected image, it be the results are shown in the step 9, be the net result that detects.
Extensive blind-detection system and method among the present invention, to the different digital watermark detection algorithms, according to the respective detection threshold value, detected object is detected, utilize the data fusion method to carry out analyzing and processing then for testing result, classify and sums up according to its testing result that whether meets in the database rule, and adjudicate to draw whether there is digital watermarking.
Claims (2)
1. digital watermarking flooding blind checking method, it is characterized in that: the convergence analysis method that the testing result of a plurality of detection algorithms is arranged, the detection rule database is arranged,, select the relevant detection algorithm under the situation that does not need original image, to extract watermark information for different embedding algorithms;
May further comprise the steps:
Step 1, there not being under the original work watermark situation to use the detection algorithm detection corresponding with embedding algorithm, a plurality of different embedding algorithms are detected;
Step 2, the testing result of algorithms of different is carried out convergence analysis according to detecting rule;
Described step 1 comprises: linear dependence detects, the albefaction linear dependence detects, block-based linear dependence detects, block-based albefaction linear dependence detects and Viterbi detection;
Described step 2 comprises: the combination and the flow process that detect rule database and convergence analysis method.
2. digital watermarking flooding blind-detection system is characterized in that:
With each pixel and template acquiring correlative value, compare with threshold value then after the linear dependence detection module reading images;
Albefaction linear dependence detection module adopts is that 11 * 11 matrix carries out the filtering convolution algorithm to original image, mainly concentrates in the image boundary, is limited to the computing within image and the template pixel coverage;
Block-based linear dependence detection module carries out watermark extracting, earlier content is carried out some pre-service, produce a vector in the sign space, whether the sign that decision has been extracted comprises watermark information, carry out the inverse process of leaching process, new DUAL PROBLEMS OF VECTOR MAPPING is arrived Channel Space, thereby obtained adding the works of watermark; Or carry out treatment of picture, and the piece that extracts 8*8 adds up then and averages, and the piece that is directed to 8*8 then carries out watermark and embeds;
Block-based albefaction linear dependence detection module, utilize block-based blind embedding algorithm, comprise the block-based watermarking algorithm of fixedly normalization linear dependence and have when block-based watermarking algorithm that fixing robustness embeds embeds 1 value and errorlessly to detect, false dismissal when detecting embedding 0 value, occurs;
Detect detection threshold, relevant parameters and the testing result of single detection algorithm in the in store system of rule database;
When the data fusion analysis module travels through detection at system employs single detection algorithm wherein to image, the testing result of each detection algorithm and the rule that detects in the rule database are compared;
If the testing result that is complementary with the rule that detects in the rule database is arranged in the data fusion analysis module, shows to detect digital watermarking; If the testing result of each detection algorithm and strictly all rules all do not match, show not detect digital watermarking.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2006101142984A CN100416595C (en) | 2006-11-03 | 2006-11-03 | Blind detecting system and method for digital watermarking flooding |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2006101142984A CN100416595C (en) | 2006-11-03 | 2006-11-03 | Blind detecting system and method for digital watermarking flooding |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1945597A CN1945597A (en) | 2007-04-11 |
| CN100416595C true CN100416595C (en) | 2008-09-03 |
Family
ID=38044996
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2006101142984A Expired - Fee Related CN100416595C (en) | 2006-11-03 | 2006-11-03 | Blind detecting system and method for digital watermarking flooding |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100416595C (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106096668B (en) * | 2016-08-18 | 2019-06-18 | 携程计算机技术（上海）有限公司 | The recognition methods and identifying system of watermarked image |
| CN118282779B (en) * | 2024-05-31 | 2024-07-26 | 杭州海康威视数字技术股份有限公司 | Neural network-based security defense method and device for dense multimedia data |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6415040B1 (en) * | 1997-05-29 | 2002-07-02 | Koninklijke Philips Electronics N.V. | Device for optically scanning a record carrier |
| CN1581221A (en) * | 2003-08-07 | 2005-02-16 | 国际商业机器公司 | Method and apparatus for inserting and detecting watermark in picture derived from source picture |
| CN1790420A (en) * | 2005-12-01 | 2006-06-21 | 北京北大方正电子有限公司 | Method and apparatus for embedding and detecting digital watermark in text file |
-
2006
- 2006-11-03 CN CNB2006101142984A patent/CN100416595C/en not_active Expired - Fee Related
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6415040B1 (en) * | 1997-05-29 | 2002-07-02 | Koninklijke Philips Electronics N.V. | Device for optically scanning a record carrier |
| US20020168088A1 (en) * | 1997-05-29 | 2002-11-14 | Koninklijke Philips Electronics N.V. | Method and arrangement for detecting a watermark |
| CN1581221A (en) * | 2003-08-07 | 2005-02-16 | 国际商业机器公司 | Method and apparatus for inserting and detecting watermark in picture derived from source picture |
| CN1790420A (en) * | 2005-12-01 | 2006-06-21 | 北京北大方正电子有限公司 | Method and apparatus for embedding and detecting digital watermark in text file |
Non-Patent Citations (4)
| Title |
|---|
| DCT变换域盲图像水印的自适应检测. 伯晓晨，沈林成，常文森，牛轶峰.计算机研究与发展，第39卷第04期. 2002 |
| DCT变换域盲图像水印的自适应检测. 伯晓晨，沈林成，常文森，牛轶峰.计算机研究与发展，第39卷第04期. 2002 * |
| 一种新的盲图像水印检测算法. 伯晓晨，沈林成，常文森.计算机学报，第24卷第12期. 2001 |
| 一种新的盲图像水印检测算法. 伯晓晨，沈林成，常文森.计算机学报，第24卷第12期. 2001 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN1945597A (en) | 2007-04-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Bianchi et al. | Improved DCT coefficient analysis for forgery localization in JPEG images | |
| Kadam et al. | [Retracted] Efficient Approach towards Detection and Identification of Copy Move and Image Splicing Forgeries Using Mask R‐CNN with MobileNet V1 | |
| US20120076422A1 (en) | System and Method for Image Authentication | |
| CN109302410A (en) | A method, system and computer storage medium for detecting abnormal behavior of internal users | |
| EP3523776A1 (en) | Systems and methods for detection and localization of image and document forgery | |
| CN109543674B (en) | Image copy detection method based on generation countermeasure network | |
| Wei et al. | Controlling neural learning network with multiple scales for image splicing forgery detection | |
| Lou et al. | Construction information authentication and integrity using blockchain-oriented watermarking techniques | |
| Deng et al. | Deep multi-scale discriminative networks for double JPEG compression forensics | |
| Chen et al. | Identifying tampering operations in image operator chains based on decision fusion | |
| Xie et al. | Adaptive image steganography using fuzzy enhancement and grey wolf optimizer | |
| CN115983874A (en) | Wine anti-counterfeiting tracing method and system | |
| CN116911883A (en) | Agricultural product anti-counterfeiting tracing method and cloud platform based on AI (advanced technology) authentication technology and tracing quantification | |
| CN100416595C (en) | Blind detecting system and method for digital watermarking flooding | |
| Pei et al. | Vision Transformer‐Based Video Hashing Retrieval for Tracing the Source of Fake Videos | |
| Abed et al. | Efficient cover image selection based on spatial block analysis and DCT embedding | |
| Dong et al. | Double compression detection based on Markov model of the first digits of DCT coefficients | |
| Frick et al. | Detecting double compression and splicing using benfords first digit law | |
| CN117314713B (en) | A digital watermark-based data processing method for certificate anti-counterfeiting | |
| Patil et al. | Securing visual integrity: machine learning approaches for forged image detection | |
| CN117527389A (en) | Industrial intrusion detection method of deformable visual transducer | |
| Cristin et al. | Image tampering detection in image forensics using earthworm‐rider optimization | |
| Lu et al. | A study on JPEG steganalytic features: Co-occurrence matrix vs. Markov transition probability matrix | |
| Lin et al. | Passive forgery detection for JPEG compressed image based on block size estimation and consistency analysis | |
| Wang et al. | Detection of Image Tampering Using Multiscale Fusion and Anomalousness Assessment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| 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 | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Assignee: Beijing Qizhi Haotian Technology Co. Ltd. Assignor: Beijing Venus Information Technology Co., Ltd. Contract fulfillment period: 2008.9.20 to 2013.12.31 contract change Contract record no.: 2008110000099 Denomination of invention: Blind detecting system and method for digital watermarking flooding Granted publication date: 20080903 License type: Exclusive license Record date: 2008.11.7 |
|
| LIC | Patent licence contract for exploitation submitted for record |
Free format text: EXCLUSIVE LICENSE; TIME LIMIT OF IMPLEMENTING CONTACT: 2008.9.20 TO 2013.12.31; CHANGE OF CONTRACT Name of requester: BEIJING QIZHI HAOTIAN SCIENCE CO., LTD. Effective date: 20081107 |
|
| C56 | Change in the name or address of the patentee | ||
| CP03 | Change of name, title or address |
Address after: 188#, 12 South Avenue, Beijing, Haidian District, Zhongguancun Patentee after: Beijing Venus Information Technology Co., Ltd. Address before: 188#, 12 South Avenue, Beijing, Haidian District, Zhongguancun Patentee before: Beijing Qiming Xingchen Information Technology Co., Ltd. |
|
| C56 | Change in the name or address of the patentee |
Owner name: BEIJING QIMINGXINGCHEN INFORMATION TECHNOLOGY CO., Free format text: FORMER NAME: BEIJING QIMING XINGCHEN INFORMATION TECHNOLOGY CO. LTD. |
|
| C17 | Cessation of patent right | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20080903 Termination date: 20131103 |