CN111968024A - Self-adaptive image watermarking method - Google Patents
Self-adaptive image watermarking method Download PDFInfo
- Publication number
- CN111968024A CN111968024A CN202010723235.9A CN202010723235A CN111968024A CN 111968024 A CN111968024 A CN 111968024A CN 202010723235 A CN202010723235 A CN 202010723235A CN 111968024 A CN111968024 A CN 111968024A
- Authority
- CN
- China
- Prior art keywords
- watermark
- embedding
- image
- wavelet
- brightness
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 38
- 230000002441 reversible effect Effects 0.000 claims abstract description 16
- 230000008569 process Effects 0.000 claims abstract description 15
- 230000000007 visual effect Effects 0.000 claims abstract description 14
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 13
- 230000000873 masking effect Effects 0.000 claims abstract description 12
- 238000000605 extraction Methods 0.000 claims abstract description 9
- 238000001514 detection method Methods 0.000 claims abstract description 5
- 230000003044 adaptive effect Effects 0.000 claims description 19
- 238000007781 pre-processing Methods 0.000 claims description 9
- 238000012545 processing Methods 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 7
- 230000008447 perception Effects 0.000 claims description 6
- 230000035945 sensitivity Effects 0.000 claims description 6
- 230000009466 transformation Effects 0.000 claims description 5
- 230000010354 integration Effects 0.000 claims description 3
- 238000011160 research Methods 0.000 description 6
- 238000013461 design Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 238000011156 evaluation Methods 0.000 description 4
- 238000002474 experimental method Methods 0.000 description 3
- 238000001914 filtration Methods 0.000 description 3
- 239000000284 extract Substances 0.000 description 2
- 241000254158 Lampyridae Species 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000008094 contradictory effect Effects 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
- 230000000670 limiting effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 238000011056 performance test Methods 0.000 description 1
- 230000002829 reductive effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000016776 visual perception Effects 0.000 description 1
Images
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
- G06T1/0028—Adaptive watermarking, e.g. Human Visual System [HVS]-based watermarking
Abstract
The invention provides a self-adaptive image watermarking method, which comprises the following steps that step 1, DWT is adopted to carry out two-level wavelet decomposition on a carrier image; step 2, embedding the watermark into a low-frequency part LL2 after the secondary wavelet decomposition, and performing reverse color and Arnold scrambling pretreatment on the watermark image; step 3, according to the brightness masking characteristic of a human visual system, carrying out brightness sub-block division and classification on the carrier image and the watermark image, optimizing the embedding position and the embedding intensity factor of the watermark, and realizing the self-adaptive embedding of the watermark; and 4, performing reverse operation on the watermark embedding process to realize the detection and extraction of the watermark. The method has simple implementation process, optimizes the embedding position and the embedding strength factor of the watermark based on the brightness masking characteristic of a human visual system in the watermark embedding process, realizes the self-adaptive embedding of the embedding strength, the embedding position and the like of the watermark along with the different characteristics of each part of the carrier image, has good self-adaptive performance, and can better balance the robustness and invisibility of the system.
Description
Technical Field
The invention relates to the technical field of digital watermarking, in particular to a self-adaptive image watermarking method.
Background
With the rapid development of information technology and the popularization of communication media such as the internet, network digital media including pictures, audio, video and the like are widely applied, and copyright protection has become an important and urgent practical problem. Digital watermarking technology, as an effective method for protecting digital media copyright, has attracted great interest to people at home and abroad and has become a hot spot of research at home and abroad.
The research of the current digital watermarking technology is mostly performed around the factors of robustness, invisibility and the like of the watermark, however, the invisibility and the robustness of the watermark are two contradictory aspects. More watermark information is embedded into the carrier information, so that the performance of the watermark system for resisting attack is improved, the robustness is enhanced, and the invisibility is worsened; conversely, if the embedded watermark information is less, the invisibility is better, but when the watermark carrier data is slightly disturbed, the watermark information is difficult to extract, and the robustness becomes worse. How to balance this pair of contradictions is an important issue that must be faced and solved in designing digital watermarks. Around the problem, the research and application of the adaptive watermark technology is generated, and the purpose is to realize the adaptive embedding of the embedding strength, the embedding position and the like of the watermark along with the different characteristics of each part of the carrier signal, so as to solve the balance problem of invisibility and robustness of the digital watermark, realize the lower complexity of the algorithm as much as possible and enhance the practicability of the algorithm. A great deal of different adaptive watermarking algorithms exist so far, most of research is to improve invisibility of embedded watermarking information and enhance robustness against watermarking attack, however, digital watermarking still has difficulty in obtaining better balance between invisibility and robustness, complexity and practicability of watermarking algorithms need to be further optimized, and the method is also a main bottleneck encountered in the design and application process of current watermarking algorithms.
Disclosure of Invention
Around the balance problem of invisibility and robustness of digital watermarks, the invention designs a self-adaptive image watermarking method based on discrete wavelet transformation, optimizes the embedding position and the embedding intensity factor of the watermark according to the brightness masking characteristic of a human visual system, and realizes the self-adaptive embedding of the watermark.
The invention provides the following technical scheme: a self-adaptive image watermarking method mainly comprises the following steps:
and 4, performing reverse operation on the watermark embedding process to realize the detection and extraction of the watermark.
Further, the step 2 specifically includes:
carrying out block processing on the region of the carrier image embedded with the watermark, calculating the brightness average value of a surrounding block including an embedding position before embedding the watermark information, and then determining the size of the intensity factor of the embedded watermark according to the brightness feeling of human eyes and the parameter of a contrast sensitivity threshold;
the contrast sensitivity threshold is also called Weber, and the calculation method is as follows:
where B is the background luminance and Δ B is the luminance superimposed on the background;
carrying out reverse color preprocessing on the watermark image, which specifically comprises the following steps: if the average pixel value of the watermark image is larger than 127, modifying the pixel value of any pixel value x to be 255-x, otherwise, keeping the pixel value unchanged;
setting the embedding strength of the watermark as alpha, the maximum value of the pixel of the watermark after reverse color preprocessing as 127, and the average value of the pixel brightness of the carrier image area corresponding to the embedded watermark as A, when the watermark is embedded in a smooth area, the requirement of embedding the watermark in the smooth area is met
The maximum value of watermark embedding strength thus obtained is:
taking a basic embedding strength factor
The subjective brightness feeling S of human eyes to a certain sub-block is in direct proportion to the logarithm of the objective average brightness B of the background of the sub-block by the Weber-Fisher law, and the formula is expressed as follows:
S＝KlgB+K0 (4)
where K is a constant, is the relative intensity that gives rise to the subjective perception of difference, K0Is an integration constant, usually K0＝0；
To satisfy the invisibility of the embedded watermark, the constant K in the above formula is the basic embedding strength factor alpha calculated by the Weber ratio0Finally, the calculation formula of the embedding strength factor is determined as follows:
α1＝α0×lg(|B-128|+10) (5)
the watermark embedding strength range determined by the above formula is: alpha is alpha1∈[α0,2.14α0]。
Further, after the two-level wavelet decomposition is performed on the carrier image, N × N coefficients are randomly selected from the low-frequency wavelet coefficient LL2 to embed watermark information subjected to inverse color and Arnold scrambling preprocessing, where N × N is the size of the watermark image, and a specific embedding algorithm is as follows:
the method comprises the following steps: dividing the low frequency wavelet coefficients LL2 into appropriate sub-blocks;
step two: calculating the pixel brightness average value A of each sub-block, and calculating the most suitable embedding intensity alpha of each sub-block according to different brightness average values, wherein the formula is as follows:
step three: based on the brightness masking characteristic of a human visual system, dividing the average brightness of wavelet sub-blocks of a carrier image and the pixel value of a watermark image subjected to reverse color processing into four types of bright, dark, brighter and darker from a brightness angle, randomly embedding watermark information with a large pixel value of the watermark image into the wavelet sub-blocks with a bright background, and analogizing according to the rule to obtain the wavelet sub-blocks and an embedding mode after the brightness of the watermark information is classified;
step IV: embedding corresponding watermark information into each wavelet sub-block according to a known adaptive intensity factor, and adopting an embedding mode of an addition rule, wherein an embedding formula is as follows:
F'(x,y)＝F(x,y)+α·Wa(x,y) (7)
wherein alpha is an adaptive intensity factor, Wa(x, y) is information of the watermark at the (x, y) position after reverse color scrambling, F (x, y) is an original wavelet coefficient of the wavelet sub-block, and F' (x, y) is a wavelet coefficient after the watermark is embedded;
and performing inverse discrete wavelet transform on the wavelet coefficient of the image subjected to the operation to obtain the image embedded with the watermark.
Further, the step 4 specifically includes: firstly, respectively carrying out wavelet transformation twice on an original image and an image containing a watermark, and dividing a secondary low-frequency coefficient of LL2 into wavelet sub-blocks with the same size and quantity according to a division rule of an embedded watermark implementation process; by comparing the wavelet coefficient difference, the embedded watermark value information of each wavelet sub-block is calculated, and the original watermark information is obtained according to the inverse operation of the embedding formula
Then, rearranging the watermark information according to the classification rule during embedding; finally, according to the Arnold scrambling times pair
The original embedded watermark information is obtained by performing inverse Arnold scrambling followed by inverse color processing.
Compared with the prior art, the invention has the beneficial effects that:
the self-adaptive image watermarking method provided by the invention is simple in implementation process, optimizes the embedding position and the embedding strength factor of the watermark based on the brightness masking characteristic of a human visual system in the watermark embedding process, and realizes the self-adaptive embedding of the embedding strength, the embedding position and the like of the watermark along with the different characteristics of each part of the carrier image. Experimental results show that the method has good self-adaptive performance and can well balance the robustness and invisibility of the watermark.
Drawings
FIG. 1 is a diagram of embedding modes after wavelet sub-blocks and watermark information brightness classification;
FIG. 2 is a flow chart of an adaptive watermark embedding algorithm;
fig. 3 is a flow chart of a watermark extraction algorithm;
FIG. 4 shows the results of an adaptive performance test;
FIG. 5 shows an experimental watermark image and carrier image;
fig. 6 shows the robustness test results.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail below with reference to the accompanying drawings and embodiments. The embodiments described herein are only for explaining the technical solution of the present invention and are not limited to the present invention.
The invention designs a self-adaptive image watermarking method based on Discrete Wavelet Transform (DWT) around the balance problem of invisibility and robustness of digital watermarking, and designs and researches an embedding algorithm and an extraction algorithm. In the aspect of embedding the watermark, DWT is adopted to carry out two-level wavelet decomposition on the carrier image, the watermark image is subjected to inverse color and Arnold scrambling preprocessing, the carrier image and the watermark image are subjected to luminance sub-block division and classification according to the luminance masking characteristics of a human visual system, the embedding position and the embedding strength factor of the watermark are optimized, and the embedding strength, the embedding position and the like of the watermark are adaptively embedded along with the different characteristics of each part of the carrier image. The watermark embedding process is reversely operated, so that the detection and extraction of the watermark can be realized.
1 self-adaptive image watermark embedding algorithm
(1) Selection of embedding location
Different carrier images have different characteristics, and the outlines, textures and energy sizes of different parts of the carrier images are different. And the carrier image is subjected to wavelet decomposition to obtain a low-frequency approximate component LL, a high-frequency horizontal detail component LH, a high-frequency vertical detail component HL and a high-frequency diagonal detail component HH. The low frequency approximation component LL is an approximate sub-image of the original image, which can be decomposed continuously, and the other three high frequency detail components mainly represent edge detail information of the original image in horizontal, vertical and diagonal directions, respectively, such as contour and texture.
If the watermark is embedded into the high-frequency part of the carrier image subjected to wavelet transform, the watermark is not easily perceived by human eyes after being embedded according to the texture masking characteristic of a human visual system. However, the proportion of the high-frequency part information in the whole image data is generally very low, the high-frequency part information is easily interfered by external factors such as noise, filtering and geometric transformation, the stability is poor, the embedded watermark information is easily lost, and the robustness is not strong. Therefore, the high frequency parts (LH, HL, HH) are not subjected to watermark embedding. Since the wavelet coefficient of the low-frequency part LL2 after the secondary wavelet decomposition is relatively larger than that of other high-frequency parts of the same level, the part occupies most energy of the image, and the watermark is embedded in the low-frequency part LL2, so that information is not easy to lose, and the robustness is high. Thus, the present invention embeds the watermark in the low frequency portion LL2 after the two-level wavelet decomposition.
(2) Calculation of adaptive embedding strength
The low frequency part LL2 represents a smooth part containing information such as contours, which the human eye is very sensitive to, and the embedded watermark strength, if not processed, can easily degrade the quality of the original image and cause distortion. According to the characteristics of a human visual system, a proper watermark embedding intensity factor is selected, self-adaptive adjustment can be performed according to different image contents, and meanwhile, the condition that the embedded watermark intensity is lower than a critical invisible threshold of the human visual system HVS is met, so that the robustness and invisibility of the watermark are guaranteed.
Since the features such as the brightness and texture complexity of different regions of an image are different, the threshold invisible threshold is also different. In order to fully utilize the difference, the invention carries out block processing on the watermark embedding area of the carrier image, calculates the brightness average value of a surrounding block including the embedding position before embedding the watermark information, and then determines the magnitude of the watermark embedding intensity factor according to parameters such as human eye brightness perception, contrast sensitivity threshold and the like.
The contrast sensitivity threshold is also called Weber, and the calculation method is as follows:
where B is the background luminance and Δ B is the luminance superimposed on the background. For backgrounds of different brightness, the brightness variation thresholds that the human eye can perceive are also different. Studies have shown that, in general, k 0.018 is the limit value of the luminance change that can be recognized by the human eye. When the exposure is too strong or insufficient, the background is brighter or darker, and k may reach 0.05 or more, and the watermark embedding strength may be suitably increased.
Meanwhile, the magnitude of the watermark embedding strength is related to the average pixel value of the watermark. The higher the average pixel value of the watermark image is, the higher the visibility of the watermark image after being embedded into the carrier image is, and in order to meet the invisibility requirement, the lower the corresponding embedding strength is. In order to ensure that the watermark image with a high average pixel value also has embedding strength as large as possible, that is, to meet the robustness requirement, the watermark image needs to be subjected to inverse color preprocessing, specifically: and if the average pixel value of the watermark image is larger than 127, modifying the pixel value of any pixel value x to be 255-x, otherwise, keeping the pixel value unchanged.
Setting the embedding strength of the watermark as alpha, the maximum value of the pixel of the watermark after reverse color preprocessing as 127, and the average value of the pixel brightness of the carrier image area corresponding to the embedded watermark as A, when the watermark is embedded in a smooth area, the requirement of embedding the watermark in the smooth area is met
The maximum value of watermark embedding strength thus obtained is:
taking a basic embedding strength factor
From weber-fisher' S law, the subjective brightness perception S (psychological quantity) of a certain sub-block by the human eye is proportional to the logarithm of the objective average brightness B (physical quantity) of the background of the sub-block, and the formula is expressed as follows:
S＝KlgB+K0 (4)
where K is a constant, is the relative intensity that gives rise to the subjective perception of difference, K0Is an integration constant, usually K00. The larger or smaller B, i.e. the larger the absolute value of the difference of the sub-block average luminance and 128 this intermediate pixel value, the greater the intensity of the embedding. To satisfy the invisibility of the embedded watermark, the constant K in the above formula is the basic embedding strength factor alpha calculated by the Weber ratio0Finally, the calculation formula of the embedding strength factor is determined as follows:
α1＝α0×lg(|B-128|+10) (5)
the watermark embedding strength range determined by the above formula is: alpha is alpha1∈[α0,2.14α0]。
Watermark embedding in the above formula range leads to average brightness variation of each sub-block
(3) Embedding algorithm
After the carrier image is subjected to secondary wavelet decomposition, N × N (watermark image size) coefficients are randomly selected from low-frequency wavelet coefficients LL2 to embed watermark information subjected to inverse color and Arnold scrambling pretreatment, and the specific embedding algorithm is as follows:
the low frequency wavelet coefficients LL2 are divided into appropriate sub-blocks, such as 25 × 25 sub-blocks (if uniform division by square blocks cannot be guaranteed, appropriate rectangular block partitions may be selected).
Secondly, calculating the pixel brightness average value A of each sub-block, and calculating the most appropriate embedding intensity alpha of each sub-block according to different brightness average values, wherein the formula is as follows:
and thirdly, based on the brightness masking characteristic of the human visual system, if the brightness value of the watermark is larger, the sub-block with the background brightness closer to the extreme value is embedded. Accordingly, the average brightness of the wavelet sub-blocks of the carrier image and the pixel value of the watermark image after the inverse color processing are divided into four categories of bright, dark, brighter and darker from the brightness perspective, and the watermark information (bright) with a large pixel value of the watermark image is randomly embedded into the wavelet sub-blocks with bright background, so that the embedding mode shown in fig. 1 is obtained by analogy with the rule.
Embedding corresponding watermark information into each wavelet sub-block according to the known adaptive intensity factor, wherein the embedding mode of the addition criterion is adopted by the invention, and the embedding formula is as follows:
F'(x,y)＝F(x,y)+α·Wa(x,y) (7)
wherein alpha is an adaptive intensity factor, WaAnd (x, y) is information of the watermark at the (x, y) position after inverse color scrambling, F (x, y) is the original wavelet coefficient of the wavelet sub-block, and F' (x, y) is the wavelet coefficient after embedding the watermark.
The wavelet coefficient of the image after the above operations is subjected to inverse discrete wavelet transform to obtain the image with the embedded watermark, and the flow of the complete adaptive watermark embedding algorithm is shown in fig. 2.
2 image watermark extraction algorithm
The detection and extraction of the watermark is the reverse operation of the watermark embedding process. Firstly, wavelet transformation is carried out twice on an original image and an image containing a watermark, and wavelet sub-blocks with the same size and quantity are divided for a LL2 secondary low-frequency coefficient according to a division rule of an embedded watermark implementation process. By comparing the wavelet coefficient difference, the embedded watermark value information of each wavelet sub-block is calculated, and the original watermark information is obtained according to the inverse operation of the embedding formula
And then the watermark information is rearranged according to the classification rule during embedding. Finally, according to the key (Arnold scrambling times) pair
The original embedded watermark information is obtained by performing inverse Arnold scrambling followed by inverse color processing. A flow chart of a specific extraction algorithm is shown in fig. 3.
3 evaluation of System Performance
(1) Performance evaluation index
The performance of the digital watermarking system after being attacked is evaluated, and the performance can be qualitatively evaluated by using a human perception system or quantitatively evaluated by using a relevant standard. The paper adopts the following two indexes to evaluate and analyze the performance of the watermark system.
Peak Signal-to-noise ratio (PeakSignalNoiseRatio, PSNR)
Ii,jRepresenting the values of the pixels of the original image,
representing the pixel values of the image after embedding the watermark, M representing the number of rows of the image and N representing the number of columns of the image. Generally speaking, a larger PSNR value means that the distortion of the watermarked image compared to the original image is smaller, i.e. the quality of the watermarked image is better maintained. Typical PSNR values are generally 25-45 dB, and PSNR values obtained by different methods are generally different.
Normalized Correlation coefficient (Normalized Correlation, NC)
Wi,jIs the pixel value of the original watermark image,
is the extracted watermark image pixel value. The normalized coefficients are used to describe the similarity of the extracted watermark to the original watermark. For robust watermarks, generally the closer the NC value is to 1 the better, and for fragile watermarks the smaller the NC value is the better.
(2) Evaluation of adaptivity
In order to analyze the problem of self-adaptability of the algorithm to different watermarks embedded into different carrier images, the invention adopts the following two groups of experiments to embed the watermarks. The evaluation of the algorithm performance adopts peak signal-to-noise ratio (PSNR) and normalized correlation coefficient (NC), namely: when the PSNR is more than 20 and the NC is more than 0.7, the algorithm is effective; when the PSNR is more than 30, the watermark is excellent in transparency and has good invisibility; when NC is more than 0.9, the anti-attack performance is not wrong, and the watermark can be completely extracted.
And (4) embedding the same watermark into the same carrier image with different brightness. The invention obtains two carrier images with different exposure degrees by using MATLAB function imagjust to adjust the brightness of an original carrier image scene.jpg (2048 multiplied by 1448), and respectively embeds and extracts watermarks nculogo.png (244 multiplied by 244), and the experimental result is shown as the 1 st-3 rd lines in figure 4. The result shows that the same watermark is embedded into the same carrier image with different brightness, the PSNR values of the same watermark are basically unchanged, and the PSNR > 20 and NC > 0.7 are all satisfied, which shows that the algorithm can effectively perform the adaptive embedding of the watermark strength on the carrier images with different brightness.
And secondly, embedding different watermarks into the same carrier image. The invention uses three different watermark images nculogo.png (244 × 244, 5.74KB), badge.jpg (348 × 348, 29.5KB) and logo.GIF (122 × 108, 1.28KB), and embeds the watermark images into the same original carrier image scenry.jpg (2048 × 1448), and extracts the watermark respectively, and the experimental results are respectively shown in lines 1, 4-5 in fig. 4. The result shows that under the condition that the carrier image is not changed, the data quantity of the embedded watermark image is increased by a multiple of 4-5 times, but the PSNR value of the carrier image containing the watermark is not reduced by a multiple, namely, the embedding strength is adaptively adjusted to ensure the invisibility of the watermark along with the increase of the information quantity of the watermark. Meanwhile, the experimental data of the group meet PSNR & gt 20 and NC & gt 0.9, feasibility and effectiveness of the algorithm are shown, and adaptive embedding of watermark strength can be achieved. The visual perception also can know that the quality of the image embedded with the watermark is not greatly different from that of the original carrier image, and the existence of the watermark is difficult to be perceived by naked eyes.
(3) Robustness and invisibility assessment
In order to evaluate the robustness and invisibility of the algorithm, the algorithm is adopted to embed the watermark images nculogo.png (244 × 244), ncu _1.png (86 × 86) and logo.GIF (122 × 108) into the carrier images scene.jpg (2048 × 1448), trailer.jpg (1024 × 684) and office _5.jpg (960 × 600) respectively to obtain corresponding watermark-containing images, and the PSNR value of each watermark-containing image is calculated, which is specifically shown in fig. 5. Attack experiments are respectively carried out on each watermark-containing image by adopting attacks such as filtering, zooming, clipping, noise, rotation and the like, the watermark image is extracted from the attacked image, the NC value of the watermark image is calculated and compared with the document [1] (Behnam Kazemish, Mohsen Ebrahimi Moghaddam.A robust digital image watermarking technique using shifting motion transform and Firefly algorithmm [ J ]. Multimedia Tools and Applications,2017,76(20): 20499-20524.), and the experimental result is shown in FIG. 6.
As can be seen from fig. 5, the algorithm of the present invention embeds different watermark images into different carrier images, and the PSNR values of the obtained watermark-containing images are related to characteristic factors such as the sizes of the carrier images and the watermark images, and the PSNR values are all greater than 30 or close to 30, which indicates that the embedding of the watermark does not cause too great changes in the quality of the carrier images, and the watermark system has excellent transparency and good invisibility. Fig. 6 shows different types of attack experiments performed on each watermark-containing image, and the result shows that the size of the NC value (which reflects robustness) of the watermark image extracted after the watermark-containing image is attacked is related to the carrier image, the characteristics of the watermark image, the type of the attack, and other factors. From the attack type, the NC value of the watermark image extracted after the attack such as filtering, scaling, clipping (upper left corner 1/8), noise, rotation and the like is larger and meets NC & gt 0.9, which shows that the algorithm has good anti-attack performance and the watermark can be completely extracted. The NC value of the watermark image extracted after the attack of the clipping (center 1/4) is smaller, but is close to 0.9, and the NC > 0.7 is met, which indicates that the algorithm is effective. Compared with the document [1], the watermark image extracted by the algorithm after being attacked by the clipping (the center 1/4) has the NC value smaller than that of the document [1], and the watermark images extracted after other types of attacks have the NC values larger than that of the document [1], so that the robustness of the algorithm against the attacks is better than that of the document [1] on the whole. In sum, the watermarking system has good robustness against different types of attacks.
The invention designs and researches the self-adaptive image watermarking method, has simple realization process, optimizes the embedding position and the embedding strength factor of the watermark based on the brightness masking characteristic of a human visual system in the watermark embedding process, and realizes the self-adaptive embedding of the embedding strength, the embedding position and the like of the watermark along with the different characteristics of each part of the carrier image. Experimental results show that the method has good self-adaptive performance and can well balance the robustness and invisibility of the watermark.
The foregoing merely represents preferred embodiments of the invention, which are described in some detail and detail, and therefore should not be construed as limiting the scope of the invention. It should be noted that, for those skilled in the art, various changes, modifications and substitutions can be made without departing from the spirit of the present invention, and these are all within the scope of the present invention. Therefore, the protection scope of the present patent shall be subject to the appended claims.
Claims (4)
1. An adaptive image watermarking method, characterized by: the method mainly comprises the following steps:
step 1, carrying out two-level wavelet decomposition on a carrier image by adopting DWT;
step 2, embedding the watermark into a low-frequency part LL2 after the secondary wavelet decomposition, and performing reverse color and Arnold scrambling pretreatment on the watermark image;
step 3, according to the brightness masking characteristic of a human visual system, carrying out brightness sub-block division and classification on the carrier image and the watermark image, optimizing the embedding position and the embedding intensity factor of the watermark, and realizing the self-adaptive embedding of the watermark;
and 4, performing reverse operation on the watermark embedding process to realize the detection and extraction of the watermark.
2. An adaptive image watermarking method according to claim 1, wherein: the step 2 specifically comprises:
carrying out block processing on the region of the carrier image embedded with the watermark, calculating the brightness average value of a surrounding block including an embedding position before embedding the watermark information, and then determining the size of the intensity factor of the embedded watermark according to the brightness feeling of human eyes and the parameter of a contrast sensitivity threshold;
the contrast sensitivity threshold is also called Weber, and the calculation method is as follows:
where B is the background luminance and Δ B is the luminance superimposed on the background;
carrying out reverse color preprocessing on the watermark image, which specifically comprises the following steps: if the average pixel value of the watermark image is larger than 127, modifying the pixel value of any pixel value x to be 255-x, otherwise, keeping the pixel value unchanged;
setting the embedding strength of the watermark as alpha, the maximum value of the pixel of the watermark after reverse color preprocessing as 127, and the average value of the pixel brightness of the carrier image area corresponding to the embedded watermark as A, when the watermark is embedded in a smooth area, the requirement of embedding the watermark in the smooth area is met
The maximum value of watermark embedding strength thus obtained is:
taking a basic embedding strength factor
The subjective brightness feeling S of human eyes to a certain sub-block is in direct proportion to the logarithm of the objective average brightness B of the background of the sub-block by the Weber-Fisher law, and the formula is expressed as follows:
S＝K lg B+K0 (4)
where K is a constant, is the relative intensity that gives rise to the subjective perception of difference, K0Is an integration constant, usually K0＝0；
To satisfy the invisibility of the embedded watermark, the constant K in the above formula is the basic embedding strength factor alpha calculated by the Weber ratio0Finally, the calculation formula of the embedding strength factor is determined as follows:
α1＝α0×lg(|B-128|+10) (5)
the watermark embedding strength range determined by the above formula is: alpha is alpha1∈[α0,2.14α0]。
3. An adaptive image watermarking method according to claim 1, wherein: after the carrier image is subjected to secondary wavelet decomposition, N × N coefficients are randomly selected from low-frequency wavelet coefficients LL2 to be embedded into watermark information subjected to inverse color and Arnold scrambling preprocessing, wherein N × N is the size of the watermark image, and the specific embedding algorithm is as follows:
the method comprises the following steps: dividing the low frequency wavelet coefficients LL2 into appropriate sub-blocks;
step two: calculating the pixel brightness average value A of each sub-block, and calculating the most suitable embedding intensity alpha of each sub-block according to different brightness average values, wherein the formula is as follows:
step three: based on the brightness masking characteristic of a human visual system, dividing the average brightness of wavelet sub-blocks of a carrier image and the pixel value of a watermark image subjected to reverse color processing into four types of bright, dark, brighter and darker from a brightness angle, randomly embedding watermark information with a large pixel value of the watermark image into the wavelet sub-blocks with a bright background, and analogizing according to the rule to obtain the wavelet sub-blocks and an embedding mode after the brightness of the watermark information is classified;
step IV: embedding corresponding watermark information into each wavelet sub-block according to a known adaptive intensity factor, and adopting an embedding mode of an addition rule, wherein an embedding formula is as follows:
F'(x,y)＝F(x,y)+α·Wa(x,y) (7)
wherein alpha is an adaptive intensity factor, Wa(x, y) is information of the watermark at the (x, y) position after reverse color scrambling, F (x, y) is an original wavelet coefficient of the wavelet sub-block, and F' (x, y) is a wavelet coefficient after the watermark is embedded;
and performing inverse discrete wavelet transform on the wavelet coefficient of the image subjected to the operation to obtain the image embedded with the watermark.
4. An adaptive image watermarking method according to claim 1, wherein: the step 4 specifically includes: firstly, respectively carrying out wavelet transformation twice on an original image and an image containing a watermark, and dividing a secondary low-frequency coefficient of LL2 into wavelet sub-blocks with the same size and quantity according to a division rule of an embedded watermark implementation process; by comparing the wavelet coefficient difference, the embedded watermark value information of each wavelet sub-block is calculated, and the original watermark information is obtained according to the inverse operation of the embedding formula
Then, rearranging the watermark information according to the classification rule during embedding; finally, according to the Arnold scrambling times pair
Inverse Arnold scrambling followed by inverse colorThen, the original embedded watermark information is obtained.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010723235.9A CN111968024A (en) | 2020-07-24 | 2020-07-24 | Self-adaptive image watermarking method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010723235.9A CN111968024A (en) | 2020-07-24 | 2020-07-24 | Self-adaptive image watermarking method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111968024A true CN111968024A (en) | 2020-11-20 |
Family
ID=73362451
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010723235.9A Pending CN111968024A (en) | 2020-07-24 | 2020-07-24 | Self-adaptive image watermarking method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111968024A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112637440A (en) * | 2020-12-17 | 2021-04-09 | 合肥图溯信息技术有限公司 | Picture invisible watermark embedding method and system compatible with scanner equipment |
| CN113691885A (en) * | 2021-09-09 | 2021-11-23 | 深圳万兴软件有限公司 | Video watermark removing method and device, computer equipment and storage medium |
| CN113706359A (en) * | 2021-08-04 | 2021-11-26 | 武汉理工大学 | Thangka digital watermarking method based on visual perception model |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102142258A (en) * | 2011-03-31 | 2011-08-03 | 上海第二工业大学 | Wavelet transform and Arnold based adaptive gray-scale watermark embedded method |
| CN105335924A (en) * | 2015-11-18 | 2016-02-17 | 曲阜师范大学 | Wavelet domain color image watermark encryption algorithm based on differential evolution |
-
2020
- 2020-07-24 CN CN202010723235.9A patent/CN111968024A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102142258A (en) * | 2011-03-31 | 2011-08-03 | 上海第二工业大学 | Wavelet transform and Arnold based adaptive gray-scale watermark embedded method |
| CN105335924A (en) * | 2015-11-18 | 2016-02-17 | 曲阜师范大学 | Wavelet domain color image watermark encryption algorithm based on differential evolution |
Non-Patent Citations (1)
| Title |
|---|
| 刘庆: "基于小波变换的自适应图像水印算法研究", 《中国优秀硕士论文电子期刊网 信息科技辑》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112637440A (en) * | 2020-12-17 | 2021-04-09 | 合肥图溯信息技术有限公司 | Picture invisible watermark embedding method and system compatible with scanner equipment |
| CN113706359A (en) * | 2021-08-04 | 2021-11-26 | 武汉理工大学 | Thangka digital watermarking method based on visual perception model |
| CN113706359B (en) * | 2021-08-04 | 2024-05-07 | 武汉理工大学 | Tang-dynasty digital watermarking method based on visual perception model |
| CN113691885A (en) * | 2021-09-09 | 2021-11-23 | 深圳万兴软件有限公司 | Video watermark removing method and device, computer equipment and storage medium |
| CN113691885B (en) * | 2021-09-09 | 2024-01-30 | 深圳万兴软件有限公司 | Video watermark removal method and device, computer equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Huang et al. | A contrast-sensitive visible watermarking scheme | |
| CN111968024A (en) | Self-adaptive image watermarking method | |
| Parthasarathy et al. | An improved method of content based image watermarking | |
| CN108416727B (en) | Color image full-blind robust digital watermarking method based on self-embedding technology | |
| CN112700363A (en) | Self-adaptive visual watermark embedding method and device based on region selection | |
| Juarez-Sandoval et al. | Digital image ownership authentication via camouflaged unseen-visible watermarking | |
| Santoyo-Garcia et al. | Visible watermarking technique based on human visual system for single sensor digital cameras | |
| Dhaou et al. | An Efficient Group of Pictures Decomposition based Watermarking for Anaglyph 3D Video. | |
| Saikrishna et al. | An invisible logo watermarking using Arnold transform | |
| Qi et al. | An adaptive visible watermark embedding method based on region selection | |
| Al-Gindy et al. | Watermarking of colour images in the DCT domain using Y channel | |
| Rahmani et al. | A new robust watermarking scheme to increase image security | |
| Al-Gindy et al. | A novel blind Image watermarking technique for colour RGB images in the DCT domain using green channel | |
| Hu et al. | Image fusion based visible watermarking using dual-tree complex wavelet transform | |
| Dhaou et al. | An efficient anaglyph 3d video watermarking approach based on hybrid insertion | |
| CN113706359B (en) | Tang-dynasty digital watermarking method based on visual perception model | |
| Dhaou et al. | A Robust Anaglyph 3D Video Watermarking based on Multi-sprite Generation. | |
| Al-Otum et al. | Color image watermarking based on self-embedded color permissibility with preserved high image quality and enhanced robustness | |
| Abdallah et al. | SVD-based watermarking scheme in complex wavelet domain for color video | |
| Rewani et al. | Digital image watermarking: a survey | |
| Hosam et al. | A hybrid ROI-embedding based watermarking technique using DWT and DCT transforms | |
| Yuliani et al. | Copyright protection for color images based on transform domain and luminance component | |
| Yang et al. | High‐Quality and Robust Reversible Data Hiding by Coefficient Shifting Algorithm | |
| Kumar et al. | Digital image watermarking technique based on adaptive median filter and HL sub-band of two-stage DWT | |
| Chukwuchekwa et al. | Improved DCT based Digital Image Watermarking Technique using Particle Swarm Optimization Algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20201120 |
|
| RJ01 | Rejection of invention patent application after publication |