CN100377175C - Self-adaptive watermark embedding method based on partial quality estimation - Google Patents
Self-adaptive watermark embedding method based on partial quality estimation Download PDFInfo
- Publication number
- CN100377175C CN100377175C CNB2006100652231A CN200610065223A CN100377175C CN 100377175 C CN100377175 C CN 100377175C CN B2006100652231 A CNB2006100652231 A CN B2006100652231A CN 200610065223 A CN200610065223 A CN 200610065223A CN 100377175 C CN100377175 C CN 100377175C
- Authority
- CN
- China
- Prior art keywords
- watermark
- signal
- local
- block
- quality evaluation
- 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 46
- 238000013441 quality evaluation Methods 0.000 claims abstract description 33
- 230000003044 adaptive effect Effects 0.000 claims description 11
- 230000000903 blocking effect Effects 0.000 claims description 11
- 230000014509 gene expression Effects 0.000 claims description 4
- 230000007274 generation of a signal involved in cell-cell signaling Effects 0.000 claims description 2
- 230000002194 synthesizing effect Effects 0.000 claims description 2
- 238000005516 engineering process Methods 0.000 abstract description 3
- 230000035945 sensitivity Effects 0.000 abstract 1
- 230000006870 function Effects 0.000 description 12
- 235000002566 Capsicum Nutrition 0.000 description 5
- 239000006002 Pepper Substances 0.000 description 5
- 241000722363 Piper Species 0.000 description 5
- 235000016761 Piper aduncum Nutrition 0.000 description 5
- 235000017804 Piper guineense Nutrition 0.000 description 5
- 235000008184 Piper nigrum Nutrition 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 5
- 230000008447 perception Effects 0.000 description 5
- 230000000007 visual effect Effects 0.000 description 5
- 230000006835 compression Effects 0.000 description 3
- 238000007906 compression Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 238000009795 derivation Methods 0.000 description 2
- 239000003607 modifier Substances 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 238000007493 shaping process Methods 0.000 description 2
- 238000007619 statistical method Methods 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- NAWXUBYGYWOOIX-SFHVURJKSA-N (2s)-2-[[4-[2-(2,4-diaminoquinazolin-6-yl)ethyl]benzoyl]amino]-4-methylidenepentanedioic acid Chemical group C1=CC2=NC(N)=NC(N)=C2C=C1CCC1=CC=C(C(=O)N[C@@H](CC(=C)C(O)=O)C(O)=O)C=C1 NAWXUBYGYWOOIX-SFHVURJKSA-N 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
- 230000007547 defect Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000000691 measurement method Methods 0.000 description 1
- 238000001303 quality assessment method Methods 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
Images
Abstract
The present invention relates to a self-adaptive watermark embedding method based on local quality evaluation, which belongs to the technical field of a digital watermark. The existing watermark technology uses various quality evaluation methods to measure the leading distortion of a watermark, namely that the strength of an embedded watermark is always adjusted according to an overall quality index, so that the local characteristic of a signal is not thoroughly considered; in addition, in order to guarantee the local signal quality, the overall watermark strength is frequently reduced, so that the robustness of the watermark is lowered. The present invention evaluates the quality of the signal in a block dividing mode, wherein the embedded strength of the watermark in each block is adjusted in a self-adaptive mode according to the local distortion condition. With the method of the present invention, the embedded watermark can fully utilize the local characteristic of the signal, which improves the robustness of the watermark and obtains the good unknown local sensitivity simultaneously.
Description
Technical Field
The invention belongs to the technical field of digital watermarks, and particularly relates to a self-adaptive watermark embedding method based on local quality evaluation.
Background
The use and distribution of digital media information has increased explosively over the past decade. With the internet, people can conveniently publish and acquire various digital information including images, audio, video, text and the like, and many online services. At the same time, however, piracy becomes much easier, and unlimited copying and uncontrolled transmission makes digital content non-copyrighted. The management and protection of digital content is a problem that the industry is eagerly required to solve.
Digital watermarking is an emerging copyright protection technology. It studies how to hide a certain amount of additional information, such as ownership, right of use, or company identification of a work, in the original data, and achieves the purpose of verifying copyright by extracting and recognizing the hidden information. A good digital watermarking system should meet two important requirements: imperceptibility and robustness. The imperceptibility includes two aspects, one is visual imperceptibility (the same requirement for hearing), that is, the change of the image caused by embedding the watermark should be imperceptible to the visual system of the observer, and most ideally, the watermark image is identical to the original image in vision, which is the requirement that most watermarking algorithms should meet; on the other hand, the watermark is not recoverable by statistical methods, for example, even for a large number of information products which have been processed by the same method and watermark, the watermark cannot be extracted or the presence of the watermark cannot be determined by statistical methods. By robustness is meant that a digital watermark should be able to withstand a large number of different physical and geometric distortions, including intentional (e.g., malicious attacks) or unintentional (e.g., image compression, filtering, scanning and copying, noise pollution, dimensional changes, etc.). However, the two contradict each other. How to improve robustness and ensure that no significant distortion of the signal is caused has not been well solved.
In the prior art, various signal quality evaluation methods are used in the digital watermark embedding method to quantify the distortion amount between the watermark-embedded signal and the original signal, such as the mean square error, the signal-to-noise ratio, the peak signal-to-noise ratio, various perception models and the like are commonly used, and the watermark strength is adjusted according to the distortion amount to obtain the imperceptibility of the watermark. Typical perceptual shaping watermarks (c.i. podilchuk et al. Image-adaptive watermark using visual modules, ieee j.selected Areas Communications, may 1998, 16 (4): 525-539) weight the watermark signals with maximum insignificant modifiers (JND's) of the individual components of the original signal, such that the shaped watermark signals conform to human perceptual system characteristics. It uses a global quality index to measure the distortion caused by watermark, thus obtaining a global watermark strength adjusting factor.
Liu Jiufen et al propose a simple wavelet domain watermarking method (refer to patent CN 02115174.1). The method uses simple linear modulation to embed the watermark into the low frequency region of the signal, and a global stretching factor is used for controlling the embedding strength of the watermark.
Tian Li et al devised an image watermarking method (see patent CN 01114581.1). The method comprises the steps of layering and blocking an image, then randomly selecting an image block and embedding a watermark in an integer DCT (discrete cosine transform) domain. The method can avoid the error caused by signal transformation and improve the security of the watermark.
G.f.g. delbovia et al invented a method and apparatus for embedding a watermark in an information signal (see patent CN 01806071.4). It uses data of the original information signal to determine local weighting factors for the watermark signal such that the shaped watermark signal is substantially imperceptible after superposition with the original signal. This idea is similar to perceptual shaping watermarks, where local weighting factors are associated with each signal component.
Guo Baolong and so on propose a wavelet domain digital watermarking method based on image target area (see patent CN 03134437.2). Firstly, performing wavelet transformation on an original image, and determining a visual target area and a background by using the variance of a wavelet coefficient; then, the watermark signal is embedded into the high-frequency wavelet coefficient block corresponding to the visual target area by using quantization modulation.
Us Gu Taizhi et al devised an apparatus and method for embedding and detecting digital watermarks in images (see patent CN 200510009072.3). It divides an image into a plurality of partial images and embeds a digital watermark in each partial image. To ensure that the watermark is invisible, they modify only the least significant bit or the two least significant bits.
The known watermark technology measures distortion caused by the watermark globally when the watermark is embedded, and a global adjustment factor is often used for integrally adjusting the embedding strength of the watermark to be imperceptible. In many cases, a global quality indicator does not accurately reflect the local quality of the signal, e.g. a large picture or a long piece of music, and it is not flexible to adjust the local signal quality, and in order to obtain all local watermark imperceptibility, a sufficiently small global adjustment factor must be used, thus reducing the watermark robustness. Although some digital watermarking methods are based on block division, the evaluation of the signal quality still adopts a global index, and the above problem still exists.
Disclosure of Invention
Aiming at the defects in the prior art, the invention aims to provide a local quality evaluation-based adaptive watermark embedding method, which can adaptively adjust the embedding strength of a watermark according to the characteristics of an original signal, reduce the watermark strength of a noise sensitive area in the signal, increase the watermark embedding strength of a noise insensitive area and finally obtain a good compromise between the watermark robustness and the imperceptibility.
In order to achieve the purpose, the invention adopts the technical scheme that: a self-adaptive watermark embedding method based on local quality evaluation comprises the following steps:
(1) Watermark signal generation: generating a corresponding watermark signal W according to the information M to be embedded;
(2) Signal blocking: the original signal X o Divided into p non-overlapping regions or blocks, called signal blocks, denoted X, in chronological or spatial arrangement o ＝X 1 ‖X 2 ‖…‖X p (ii) a At the same time, the watermark signal W is also divided into p watermark blocks, denoted W = W 1 ‖W 2 ‖…‖W p ；
(3) Selecting a proper quality evaluation method, and setting a quality index of each signal block, which is recorded as t 1 ，t 2 ，…，t p For expressing the amount of distortion locally allowable for the signal;
(4) Embedding a corresponding watermark block in each signal block according to a set quality index;
(5) Synthesizing a watermarked signal: each signal block X to which a watermark is to be added k w (k =1,2, …, p) are combined in the original order to obtain a carrier signal containing a watermark, which is recorded as
Further, in step (1), in order to ensure the security of the watermark, the generation of W depends on the key K.
Further, in step (2), the size of the signal block is adapted to the selected quality evaluation method, so that the local quality index can faithfully reflect the subjective quality of the local signal.
Furthermore, when the original signal is an image and the mean square error, the signal-to-noise ratio or the peak signal-to-noise ratio is adopted, the signal block is not larger than 8 × 8 pixels, and preferably 5 × 5 pixels; when the Watson model based on an 8 × 8 block DCT transform is used, the signal block size is chosen to be 24 × 24 pixels.
Further, in the step (3), the quality evaluation method includes: mean square error, signal-to-noise ratio, peak signal-to-noise ratio, and perceptual model.
Further, the perceptual model includes: a Watson model based on a block discrete cosine transform and a wavelet domain human perception model.
Further, in step (3), in order to avoid blocking effect possibly caused by blocking some signals, the quality index t of two adjacent signal blocks is selected k And t k+1 The following equation is established
Wherein σ k 2 And σ k+1 2 Respectively embedded with signal blocks X before watermarking k And X k+1 The variance of (a) is calculated,and
to embed the variance of the corresponding signal block after watermarking, δ is a constant between 0 and 1, and is generally taken as δ ∈ [0,0.25 ]. Obtaining a displayed selection t according to the conditional expression and the quality evaluation function k And t k+1 The conditional inequality that must be satisfied.
Further, the quality evaluation function is a signal-to-noise ratio, and the conditional inequality is
Here:
where L is the size of the signal block,
and
respectively, the mean of the corresponding signal blocks.
Further, in step (4), the watermark signal is embedded in the original carrier signal block in blocks, or is repeatedly embedded in each carrier signal block as a whole block.
Further, in step (4), when the watermark is embedded, each signal block defines an independent quality index, and the indexes are the same or different, so that the indexes of the noise sensitive area are increased, and the indexes of the insensitive area are reduced.
Further, in step (4), each signal block X k The watermark embedding method is based on a mixing function f (-) of the watermark signal and the original information to make the watermark embedded signal block X k w The quality of (2) reaches the index in the step (3).
Further, for each signal block X k The optimum mixing function f (-) is designed so that the amount of distortion per signal block is measuredAnd (4) when the index in the step (3) is reached, the performance of the watermark detector is optimal.
Still further, each signal block has an independent watermark strength adjustment factor alpha k K =1,2, …, p, α when the optimal mixing function f (·) is determined k For flexible adjustment of X k The embedding strength of the watermark, i.e. the local imperceptibility of the signal.
Still further, in step (4), p local adjustment factors { α ] are used k K =1,2, …, p) independently adjust the respective local qualities of the signals without affecting each other; when the optimal mixing function f (-) is determined, α k Entirely by local quality index t k And determining or adjusting the embedding strength of the watermark in the noise-insensitive area according to the local characteristics of the actual signal, and reducing the embedding strength of the watermark in the noise-sensitive area, so that the local imperceptibility of the watermark is improved while the same robustness is obtained.
The invention has the following effects: using the method of the invention, the local quality of the signal is measured by using blocks, and each block obtains an adjustment factor alpha 1 ，α 2 ，…，α 3 (ii) a The embedding strength of the watermark in one block is adjusted without causing the change of other blocks, and the adjustment of the local quality of the signal is very flexible; can increase the original signal X o The watermark embedding strength of a medium noise insensitive area (such as a texture area of an image signal) is reduced, and the watermark embedding strength of a noise sensitive area (such as a flat area of the image signal) is reduced, so that the local characteristics of the signal are fully utilized by the watermark embedding, and the performance of the detector is improved; as long as the perceptual quality measurement method is properly selected, the quality index of each block is reasonably set, and alpha is adjusted 1 ，α 2 ，…，α p All areas of the signal can be madeX 1 w ，S 2 w ，…，X p w Can all reach the quality index t 1 ，t 2 ，…，t p So that all parts of the watermarked carrier signal are well suited for watermark invisibilityTherefore, the problem that the quality of some local signals is low due to the adoption of global perception quality evaluation is avoided; in conclusion, α is adjusted by combination 1 ，α 2 ，…，α p A good compromise between watermark robustness and imperceptibility can be obtained.
Drawings
FIG. 1 is a basic flow diagram of the present invention;
FIG. 2 is a diagram of the original "pepper";
FIG. 3 is a diagram of a "pepper" after embedding a watermark using the present method;
FIG. 4 is an error map between FIG. 3 and FIG. 2;
fig. 5 is a diagram of a "pepper" with a global quality index t =30dB and a watermark embedded using a linear modulation strategy;
FIG. 6 is a graph of the error between FIG. 5 and FIG. 2;
fig. 7 is a graph of the output of the watermark detector under a JPEG attack.
Detailed Description
An embodiment of the present invention is given below by describing a simple spatial domain image watermark with reference to the accompanying drawings, and further illustrating the effects of the present invention.
Example 1:
a self-adaptive watermark embedding method based on local quality evaluation comprises the following steps:
(1) Assuming the original signal X o ＝{x ij } M×N Is an image, such as a typical "pepper" image in this embodiment. The information M to be embedded is 0 or 1. In this embodiment, a random number generator is used and a large integer K is selected as a seed to generate an independent and identically distributed Gaussian sequence { w } ij ) M×N If M =1, take W = { W ij } M×N As watermark signal, otherwiseW＝{-w ij } M×N As watermark signal it is assumed here, without loss of generality, that the dimensions of the watermark signal correspond to those of the carrier signal. In order to secure the watermark signal, the key K of the watermark should be kept secret absolutely.
(2) Next, image X o Is divided into p non-overlapping blocks, denoted X, by spatial position o ＝X 1 ‖X 2 ‖…‖X p For simplicity, the block size is L × L in this embodiment, so. The excessive pixels can be discarded and can also be subdivided into a plurality of small blocks for use. At the same time, the same blocking operation is performed on the watermark signal W to obtain W = W 1 ‖W 2 ‖…‖W p 。
(3) In this embodiment, the simplest signal-to-noise ratio (SNR) is adopted to evaluate the local quality of the signal, and the SNR, which is an acceptable perceptual quality index on each block, is set to be t 1 ，t 2 ，…，t p 。
(4) For simplicity, the watermark is embedded by directly modifying the pixel values within each block, and using a linear modulation strategy as follows
Here, the watermark strength adjustment factor can be obtained by setting the signal-to-noise ratio at the kth block as follows
Here, | - | denotes the distance calculation, i.e.
Combine the above two types to be pushed out
(5) Finally, the blocks with embedded watermark are combined in the original order to obtain the carrier signal containing watermark, i.e.
Specifically, in step (3), to avoid blocking artifacts possibly caused by block embedding, the quality index t of two adjacent signal blocks is selected k And t k+1 The following equation is established
Wherein σ k 2 And σ k+1 2 Respectively embedded with signal blocks X before watermarking k And X k+1 The variance of (a) is determined,
and
to embed the variance of the corresponding signal block after watermarking, δ is a constant between 0 and 1, and is generally taken as δ ∈ [0,0.25 ]. In addition, the signal block X k w And X k+1 w Can be expressed as
And
where QA (-) denotes a quality evaluation function. If the variance in equation (1) is replaced with a statistical measure of standard deviation, and (1), (2), and (3) are combined, a display of t can be obtained k And t k+1 In between. In this embodiment, t is given when QA (-) is the signal-to-noise ratio k And t k+1 A specific derivation of the constraint between QA (-) and QA (-) can be performed by following the process when QA (-) is other quality evaluation functions. Obtaining a displayed selection t according to the conditional expression (1) and combining the used quality evaluation function k And t k+1 The conditional inequality that must be satisfied. The quality evaluation function used in this embodiment is the signal-to-noise ratio, and this conditional inequality is
The derivation process of the constraint condition that the quality indexes of the two adjacent images need to satisfy is given below. For image block X k Let W be k And X k If it is irrelevant and the mean value is zero, then the variance is obtained from the two ends of the formula (4)
Wherein σ wk 2 Indicating a watermark block W k The variance of (c). If the standard deviation is used to estimate σ wk 2 Then there is
Wherein the content of the first and second substances,
representing the mean value of the watermark blocks Wk, i.e.
Further, the formula (7) can be represented by
Is obtained by the formula (5)
Using equations (8) and (9), and considering
Formula (6) can be changed into
If the standard deviation is used to estimate σ k 2 Similarly, relational expressions can also be obtained
Wherein the content of the first and second substances,
representing a block of signals X k Is measured. The above formula is substituted by the formula (10)
Similarly, for image block X k+1 Is provided with
Substituting the equations (11) and (12) into the inequality (1) and simplifying to obtain
Wherein:
inequality (13) is the constraint condition that the quality index selection of two adjacent images needs to meet. Obviously, if δ =0, t is k+1 ＝t k . Therefore, in this embodiment, the quality index on each block is set to t 1 ＝t 2 ＝…＝t p ＝30dB。
Detecting the watermark may use a conventional correlation detector. First, a Gaussian sequence w is reproduced with a watermark key K as a seed for a random number generator ij } M×N . Next, the signal { w is calculated as follows ij } M×N With the carrier signal X to be detected w A linear correlation coefficient therebetween.
Then, ρ is compared with a threshold τ (τ > 0): if p > τ, a carrier signal X is indicated w Contains information M =1; if ρ < - τ, the carrier signal X is indicated w Contains information M =0. The choice of τ is related to the false alarm rate. In this way, the information hidden in the carrier signal is extracted. Detecting the watermark may also use a more complex detector.
In order to show the significant effect of the present invention, the present example shows some experimental results obtained by using the examples. As mentioned above, the original signal uses a typical "pepper" image, which is 512 × 512 as shown in fig. 2, and the corresponding experimental conditions are: the information to be embedded is M =1, the sub-block size is L =5, and the quality index on each block is t 1 ＝t 2 ＝…＝t p =30dB, the carrier signal is divided into p =102 × 102 sub-blocks. Fig. 3 shows a diagram of a watermarked carrier obtained using the method described in the examples; fig. 4 is a graph of the error between fig. 3 and the original signal, the value of each pixel point being magnified 10 times for clarity of presentation. For comparison, the embodiment also shows that a global quality index t =30dB and linearity are usedA carrier map containing the watermark obtained by the modulation strategy is shown in fig. 5; fig. 6 is its corresponding error map, with the value of each pixel point magnified by a factor of 10. It can be seen that at the same quality level of 30dB, the invisibility of the watermark in fig. 3 is significantly better than that in fig. 5. This is because in fig. 3 the distortion on each local area is not below 30dB, whereas fig. 5 only shows a global quality of up to 30dB, so some local quality of the image is below 30dB. The superiority and inferiority of the invisibility of the watermark is more evident in comparison with fig. 4 and 6. Although watermark embedding only adopts a simple linear modulation strategy, the invention adjusts the factor alpha by local watermark strength 1 ，α 2 ，…，α p Effectively distributing the watermark energy over the texture and edge regions of the carrier signal, as in fig. 4, while only using a unique global adjustment factor, the watermark energy is randomly distributed, as in fig. 6.
Further, the present embodiment evaluates the robustness of the watermark to information loss through JPEG compression attack. In order to perform fair comparison on the premise of the same watermark invisibility, the embodiment appropriately reduces the watermark embedding strength of fig. 5; then, fig. 3 and 5 are compressed into JPEG maps of a series of quality factors, respectively, and the response of the watermark detector is observed. If the specified false alarm rate is P fp ＝10 -8 The corresponding threshold can be calculated as τ =0.83 (see x.s.zhuetal. Betterbuseof humanvisualmodel in watermarkingbasedonlinearprediction synthesis filter. Nectur enotes computer science (LNCS) 3304, 2005. Fig. 7 shows the detector output response for different quality watermarked carrier signals, where the abscissa indicates the signal-to-noise ratio between the watermarked carrier signal and the original signal, and the watermark embedding strength of fig. 5 is reduced by a factor of 1.2 (taking 1.2 is merely illustrative, since watermark embedding uses a linear modulation strategy, the difference in performance of the watermark at other factors can also be seen in fig. 7). It can be seen that the invention significantly improves the robustness of the watermark, and even in the spatial domain, the embodiment still achieves the effect of resisting the JPEG compression attack with distortion of about 14.4 dB. This is thatBecause the embodiment utilizes the local information of the image to enhance the embedding strength of the watermark when embedding the watermark,and thus is resistant to greater information loss.
The experimental result obtained in this embodiment can show that the method of the present invention not only realizes the optimal watermark local imperceptibility, but also greatly improves the robustness of the watermark.
The embodiment gives the implementation effect of image watermarking, but the invention is not limited to image watermarking, and the invention is also applicable to digital watermarking of streaming media such as digital music, video and the like.
Example 2:
the difference from the embodiment 1 is that,
original signal X of step (1) o ＝{x ij } M×N The Lena image is subjected to 8 multiplied by 8 blocking DCT to obtain inverted frequency coefficients, the size of the inverted frequency coefficients is 240 multiplied by 240, and the information M to be embedded is 0;
in step (2), the sub-blocks are 24 × 24 in size, so that both the carrier signal and the watermark signal are divided into 10 × 10 sub-blocks;
in the step (3), a Watson perception model based on 8 multiplied by 8 blocking DCT is adopted to evaluate the local quality of the signal, and an acceptable perception quality index t on each block is set 1 ＝t 2 ＝…＝t p ＝0.3。
In the step (4), firstly, the maximum invisible modifier of each frequency point is obtained by using a Watson model, notation D = { D ij } M×N Dividing D into 10 × 10 sub-blocks according to the method of step (2), and recording the kth block as D k Then in each sub-block X k Embedding corresponding watermark block W k Using the following embedding function
Where "o" represents the multiplication of the corresponding components of the two matrices, the watermark strength adjustment factor can be obtained by using the quality index of the k-th block, as shown in the following formula
Combine the above two types to be pushed out
1≤k≤p。
(5) Finally, the blocks with embedded watermark are combined in the original order to obtain the carrier signal containing watermark, i.e.
Then, inverse 8 × 8 block DCT transform is carried out to obtain a Lena image containing the watermark.
The present invention may be embodied in other specific forms without departing from its spirit or essential characteristics. The described embodiments are to be considered in all respects (e.g. carrier signal type, quality assessment method, watermark embedding function and various parameters etc.) merely as illustrative and not restrictive. The scope of the invention is, therefore, indicated by the appended claims rather than by the foregoing description. All changes which come within the meaning and range of equivalency of the claims are to be embraced within their scope.
Claims (11)
1. A self-adaptive watermark embedding method based on local quality evaluation comprises the following steps:
(1) Watermark signal generation: generating a corresponding watermark signal W according to the information M to be embedded;
(2) Signal blocking: the original signal X 0 Divided into p non-overlapping regions or blocks, called signal blocks, denoted X, in chronological or spatial arrangement 0 ＝X 1 ‖X 2 ‖…‖X p (ii) a At the same time, the watermark signal W is also divided into p watermark blocks, denoted W = W 1 ‖W 2 ‖…‖W p ；
(3) Selecting a proper quality evaluation method, and setting a quality index of each signal block, which is recorded as t 1 ，t 2 ，…，t p For expressing the amount of distortion locally allowable for the signal;
(4) Embedding a corresponding watermark block in each signal block according to a set quality index;
(5) Synthesizing a watermarked signal: each signal block X to which a watermark is to be added k w K =1,2, …, p, which are combined together in the original order to obtain a carrier signal containing a watermark, which is recorded as
In step (2), the size of the signal block is adapted to the selected quality evaluation method, so that the local quality indicator can faithfully reflect the subjective quality of the local signal.
2. The adaptive watermark embedding method based on local quality evaluation as claimed in claim 1, characterized in that: in step (1), W is generated in dependence on the secret key K.
3. An adaptive watermark embedding method based on local quality evaluation as claimed in claim 2, characterized in that: when the original signal is an image and the mean square error, the signal-to-noise ratio or the peak signal-to-noise ratio is adopted, the signal block is not more than 8 multiplied by 8 pixels, and 5 multiplied by 5 pixels are preferably selected; when using the Watson model based on an 8 × 8 block DCT transform, the signal block size is chosen to be 24 × 24 pixels.
4. The adaptive watermark embedding method based on local quality evaluation as claimed in claim 1, characterized in that: in step (3), in order to avoid blocking effect possibly caused by blocking some signals, selecting quality index t of two adjacent signal blocks k And t k+1 The following equation is established
Wherein σ k 2 And σ k+1 2 Respectively embedded with signal blocks X before watermarking k And X k+1 The variance of (a) is determined,
and
for the variance of the corresponding signal block after embedding watermark, δ is a constant between 0 and 1, and is generally selected from δ ∈ [0,0.25); obtaining a displayed selection t according to the conditional expression and the quality evaluation function k And t k+1 The conditional inequality that must be satisfied.
5. The adaptive watermark embedding method based on local quality evaluation as claimed in claim 4, wherein: the quality evaluation function is signal-to-noise ratio, and the conditional inequality is
Here:
where L is the size of the signal block,
and
respectively, the mean of the corresponding signal blocks.
6. The adaptive watermark embedding method based on local quality evaluation as claimed in claim 1, characterized in that: in step (4), the watermark signal is embedded into the original carrier signal block in blocks, or repeatedly embedded into each carrier signal block as a whole block.
7. An adaptive watermark embedding method based on local quality evaluation as claimed in claim 1, characterized in that: in the step (4), when the watermark is embedded, each signal block defines an independent quality index, and the indexes are the same or different, so that the indexes of the noise sensitive area are improved, and the indexes of the insensitive area are reduced.
8. The adaptive watermark embedding method based on local quality evaluation as claimed in claim 1, characterized in that: in step (4), each signal block X k The watermark embedding method is based on a mixing function f (-) of the watermark signal and the original information to make the watermark embedded signal block X k w The quality of (2) reaches the index in the step (3).
9. The adaptive watermark embedding method based on local quality evaluation according to claim 8, characterized in that: for each signal block X k Designing an optimal mixing function f (-) such that the performance of the watermark detector is optimal when the distortion amount of each signal block reaches the index described in step (3).
10. The adaptive watermark embedding method based on local quality evaluation as claimed in claim 9, wherein: each signal block has an independent watermark strength adjustment factor alpha k ，k＝1，2，…, p, α, when the optimal mixing function f (·) is determined k For flexible adjustment of X k The embedding strength of the watermark, i.e. the local imperceptibility of the signal.
11. An adaptive watermark embedding method based on local quality evaluation as claimed in claim 10, characterized in that: in step (4), p local adjustment factors { alpha ] are used k K =1,2, …, p } independently adjusts each local quality of the signal without affecting each other; when the optimal mixing function f (-) is determined, α k Totally by local quality index t k And determining or adjusting the embedding strength of the watermark in the noise-insensitive area according to the local characteristics of the actual signal, and reducing the embedding strength of the watermark in the noise-sensitive area, so that the local imperceptibility of the watermark is improved while the same robustness is obtained.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2006100652231A CN100377175C (en) | 2006-03-20 | 2006-03-20 | Self-adaptive watermark embedding method based on partial quality estimation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2006100652231A CN100377175C (en) | 2006-03-20 | 2006-03-20 | Self-adaptive watermark embedding method based on partial quality estimation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1845174A CN1845174A (en) | 2006-10-11 |
| CN100377175C true CN100377175C (en) | 2008-03-26 |
Family
ID=37064102
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2006100652231A Expired - Fee Related CN100377175C (en) | 2006-03-20 | 2006-03-20 | Self-adaptive watermark embedding method based on partial quality estimation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100377175C (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102238388B (en) * | 2011-06-29 | 2013-03-20 | 西安电子科技大学 | Self-adaptive robust video watermarking method based on AVS (Audio Video Standard) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1455578A (en) * | 2003-05-10 | 2003-11-12 | 合肥工业大学 | Image waterprint method for copyright protection |
| CN1517855A (en) * | 2003-01-16 | 2004-08-04 | 成都市宇飞信息工程有限公司 | Image digital watermark method |
| CN1738353A (en) * | 2005-08-16 | 2006-02-22 | 北京交通大学 | Digital watermark technology for resisting rotary extension and displacement attack |
-
2006
- 2006-03-20 CN CNB2006100652231A patent/CN100377175C/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1517855A (en) * | 2003-01-16 | 2004-08-04 | 成都市宇飞信息工程有限公司 | Image digital watermark method |
| CN1455578A (en) * | 2003-05-10 | 2003-11-12 | 合肥工业大学 | Image waterprint method for copyright protection |
| CN1738353A (en) * | 2005-08-16 | 2006-02-22 | 北京交通大学 | Digital watermark technology for resisting rotary extension and displacement attack |
Non-Patent Citations (1)
| Title |
|---|
| 面向版权保护的数字水印技术研究. 朱新山.中国学位论文全文数据库. 2005 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN1845174A (en) | 2006-10-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Chou et al. | A perceptually tuned watermarking scheme for color images | |
| US6707930B2 (en) | Method and arrangement for embedding a watermark in an information signal | |
| Parthasarathy et al. | An improved method of content based image watermarking | |
| Wang et al. | Blind image data hiding based on self reference | |
| Zhao et al. | Effective watermarking scheme in the encrypted domain for buyer–seller watermarking protocol | |
| Divecha et al. | Implementation and performance analysis of DCT-DWT-SVD based watermarking algorithms for color images | |
| CN110189243B (en) | Color image robust watermarking method based on tensor singular value decomposition | |
| US20020087864A1 (en) | Method and arrangement for embedding a watermark in an information signal | |
| CN111127291A (en) | Image watermark embedding and extracting method and system based on spatial frequency domain JND conversion | |
| Kim et al. | An entropy masking model for multimedia content watermarking | |
| Al-Gindy et al. | Watermarking of colour images in the DCT domain using Y channel | |
| Rawat et al. | DCT and SVD based new watermarking scheme | |
| Maity et al. | An image watermarking scheme using HVS characteristics and spread transform | |
| CN100377175C (en) | Self-adaptive watermark embedding method based on partial quality estimation | |
| Liu et al. | A hybrid approach of DWT and DCT for rational dither modulation watermarking | |
| Ustubioglu et al. | DCT based image watermarking method with dynamic gain | |
| Zheng et al. | Image quality measurement by using digital watermarking | |
| Shaikh et al. | Digital Image Watermarking In DCT Domain | |
| Chang et al. | Spatial domain image hiding scheme using pixel-values differencing | |
| Yan et al. | Block-based adaptive image watermarking scheme using just noticeable difference | |
| Suthaharan et al. | Perceptually tuned robust watermarking scheme for digital images | |
| Hu et al. | Incorporating Watson's perceptual model into patchwork watermarking for digital images | |
| Taherinia et al. | Blind dewatermarking method based on wavelet transform | |
| Peungpanich et al. | An improving method for image watermarking using image averaging and tuned pixels prediction | |
| Zhou et al. | The research for tamper forensics on MPEG-2 video based on compressed sensing |
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 | ||
| ASS | Succession or assignment of patent right |
Owner name: BEIJING FANGZHENG APABI TECHNOLOGY CO., LTD.; PEK Free format text: FORMER OWNER: BEIJING FOUNDER ELECTRONICS CO., LTD.; PEKING UNIVERSITY Effective date: 20080425 |
|
| C41 | Transfer of patent application or patent right or utility model | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20080425 Address after: Twelve floor, central tower, No. 52, North Fourth Ring Road, Haidian District, Beijing, zip code: 100085 Co-patentee after: Peking University Patentee after: FOUNDER APABI TECHNOLOGY Ltd. Address before: Beijing Haidian District Haidian District five street 9 founder building zip code: 100085 Co-patentee before: Peking University Patentee before: BEIJING FOUNDER ELECTRONICS Co.,Ltd. |
|
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20080326 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |