TW201112756A - A digital watermarking technique based on random halftoning - Google Patents

Abstract

Halftone images are used in many pressworks, for example, books, magazines, the documents output from the printers and so on. The images of presswork can be easily copied, misapplied and juggled, and these actions would seriously affect the intellectual property rights of the owners. This invention presents a new watermarking technique to protect the intellectual property rights of halftone images. The proposed method combines the image halftoning and digital watermarking; moreover, it does not need the original cover image and watermark for watermark extraction such that it is more convictive and robust for proving the ownerships of images. The experimental results show that the invention can effectively extract integral watermarks for human vision even the stego-images have been attacked by shaping, modifying or cropping processing. Hence it can confirm that the proposed watermarking technique is robust and can protect the intellectual property rights of halftone images effectively.

Description

201112756 IV. Designated representative map: (1) The representative representative figure of this case is: (1) The symbol of the chart is simple and sharp m:

The program of hiding the original grayscale image into the watermarking program has the chemical formula of the halftone image conversion program printed with the watermark. If there is a chemical formula in this case, please reveal the best indication of the invention. 6. Description of the invention: Technical field

The present invention relates to "information security", and proposes an effective digital watermarking technique for applying the halftone to the grayscale image of the printed matter in halftone: i. It is very wide-ranging, and the images in the books, magazines, and printers' output documents are all presented in a way. In this information explosion and printing technology is very skillful (four) generation, a variety of printed private books, magazines, newspapers, 'printer output files, etc. in the day t life in these prints, the text can not be posted _ by the image The way to convey 'there are images in a variety of prints often occupy the majority of A's layout. However, today's computer and printing technology is very sophisticated and widely used. The images on print 201112756 can be easily copied, stolen or sneaked by ordinary users. Her protection of the shadow material authority is a matter of financial difficulties.

In order to prevent copyrighted images from being damaged and stolen by illegal users, various = protection and its version _ technology and weaving. Digital Watermarking Technology One of them is a very effective and widely researched and used technology in recent years. If the digital watermark technology age is to hide the image or text of the Lai Daixin bribe into the image that needs to be forged, the purpose is to protect the lion. Digital floating watermarking technology can be mainly used in two types: visible financial watermarking technology and non-floating watermarking technology. The visual watermarking technology age is to directly put the image of the trademark image or text of Wei Gu New in the protected image under the premise of visually perceivable by the human age, so the image after the visual floating water (4) is processed. Any user can detect the trademark image or text of the person, which is the owner of the image. The visual watermarking technology has the disadvantage that the image will be destroyed (distorted) under human vision after the image is placed in the watermark, so if the image is applied = the field is not allowed to be destroyed by the image, For example, the map of the museum collection: the original, the image used for sale or the image will be lost after the destruction, the technology will not be applicable. Another watermarking technology is the invisible watermarking technology. This technology mainly uses the imagery technique (steganagraphy) to bring the business image or text of the training brand to the human under the premise of human perception. In the image of the tilted S image, any user can not easily detect the trademark image or text of the person who has been processed by such watermarking technology. Since the watermark of the Tibetans cannot be directly observed, the watermarking technique generally includes two parts: a watermark hiding program and a watermarking extracting program. In the screaming part of the floating water, it is necessary to require that the 201112756 image with the watermark already hidden and the original image cannot be perceived differently in human vision; in the watermarking take-out part, the protected image is subjected to malicious tampering and After the destruction, the human visually identifiable trademark image or text is taken out as a proof of copyright. This feature is called r强bust, which is used to evaluate the inferior watermarking technology. Important indicators. At present, digital watermarking technology can focus on grayscale, color, and halftone images, while images on printed matter are presented in halftone images. The invisible watermarking technique for protecting images on printed matter is the main research subject of the present invention. ^ The invention is in the field of "Information * Security" or may be subject to relevant professional advice or assistance from the "Republic of China Security Society". [Prior Art] There are many technologies used to protect the intellectual property rights of images. One of them is digital watermarking, which is one of the most commonly used techniques. Currently, the watermarking technology for halftone images can be divided into two main types. Two kinds, the kind of visible floating water, its ocean watermark can be directly detected from the image of the tilted image with the naked eye to achieve the purpose of copyright announcement; another technique is invisible watermark, its technology uses hidden #〇iidmg teehnique ^f^M卩 is hidden in the protected image, so its watermark can't be perceived by the naked eye', but it needs to be removed from the watermark via the watermark to restore the ownership of the wealthy image. Technology, this technology can protect the county and integrity of the image in the sense of j, and also increase the degree of malicious damage to the foreign watermark by the general user. The present invention also belongs to the fresh watermark technology. In the past five years, there has been a technology for watermarking in the past four years. The technology has been developed by MSFu et al., and a visual type of sewage printing technology applied to the half-color image (CWED, 〇> Rrehti (m.b_ mtemrni'kiiig Error· Difflision) 'This technique will first calculate a halftone image rniage to be protected) and the black and white floating water (four) image to be hidden between the _ _ number (_) coffee n coeffident Then, by the characteristics of the correlation coefficient, the black and white watermark image is hidden in the protected halftone image by the bit string mbitst job. Sc Pei et al. proposed a technique called PMEDF, which is a low-complexity watermarking technique that applies halftones. The image type is a name. A half-color female image produced by the digital diffusion gray-scale image half-tone conversion technique of error diffusion. This half-color conversion technology converts a grayscale image using a gammel-free core (kemel). For a halftone image, there are many types of cores. Scpd et al. use the foot core 2 and the value of 128 threshold (threshdd) for black and white watermark hiding. In addition, the above scholars have found that NBEDF (Noise Bal_d Em)i* Diffuse and k^df (Kemds Alt_ted e_) are used in the error diffusion method.

DiffiisKm) These two different cores are converted to digital halftone images and converted to halftone images. After the fast Fourier transform (FastF〇urierTna f surface, FFT), the distribution will be different. Therefore, in the stage of hiding the ocean watermark, sc% et al. convert the black and white floating watermark to be hidden into a bit string, and use the above two cores to represent the hidden color in the process of grayscale image halftone conversion. The entry is 〇 or! . In the process of taking out the watermark, the fast Fourier transform is used to determine the core used in the halftone conversion of each block in the halftone image, and then the bit element is sequentially extracted by the bit represented by the core, thereby restoring the bit element. Watermarking. In addition, SC Pei et al. proposed another watermarking technique for digitally tuned halftone image 201112756 (dithering halftone image), which uses an ordered dithering halftone. Window ) cuts a halftone image into several sub-images. Because of the sorted brightness threshold in the halftone window, halftone conversion using this method causes each in the halftone scene. The proportion of black pixels in the sub-image will be decremented. In the process of hiding the watermark, the black and white watermark image will be hidden in the bit string manner. The hidden way is to hide one bit in the two sub-images, and adjust the black pixel ratio in the sub-image to hide In place 0 or 1; the method of taking out the watermark also judges that the hidden bit is 〇 or 丨, in order to increase the toughness of the watermark, the technology must use the original in the process of taking out the watermark The watermark is ANDed with the bit 7C taken out to reduce the error caused by the image being attacked, but the watermark image taken out will still be distorted. The present invention will improve on this phenomenon. In the original floating water (four) like the miscellaneous state, in the possession of watermarking, it is subject to attack and light reduction and subtraction from the visual reversible watermark. [Summary of the Invention] The item is a kind of Tibetan watermarking in a half-color image, which is a Tibetan with halftone conversion and watermarking. In order to achieve the image halftone to reflect the Tibetan watermark at the same time, I (4) - recognize 2 image half-color _ change technology, because IH can be adjusted to have a number of miscellaneous adjustments ' so it helps to fine-tune halftone The converted, privately-owned R-day machine can be used as a watermark for copyright proofing: images on paper, magazines, etc. are all halftones. [The invention is suitable for the wisdom of images applied to protect printed matter. property. I 201112756 will introduce the processing flow of the watermarking technology based on RgjH proposed by the present invention, which respectively includes the RBIHi operation flow and the program for hiding and retrieving the watermark. In general, if you use a single threshold (thresh〇ld) directly, convert a grayscale image with a pixel value range of 0 to 255 into a binary image with a pixel value of only 255 and 255, which will result in the converted two. There is a visually significant difference between the meta-image and the original gray-scale image. Since the pixel value of the gray-scale image represents the brightness change of each pixel in the image, in order to be able to show a smooth brightness in the binary image-printing ship The _ degree changes, halftone image technology came into being. The halftone image still belongs to the binary image. It simulates the brightness change of the grayscale image by touching the black pixel of the image towel, that is, the frequency of the pixel whose pixel value is G. When the grayscale image is darker in a certain area, In the half-color women's towel (4), the black pixel density of the area is increased to reduce the brightness, and vice versa. In order to effectively adjust the density of black pixels in each area of the image, most of the image halftone conversion technology will cut a block of fixed size (10) ck) centering on the pixel being processed during the conversion process, each Use the block side • < half of the color changer will have a side-size kernel (kernel) whose core size is the size of the block. The private block contains the block (10). The weight of the pixel. These weights are used to calculate the gated value of the pixel binarization. When the value of a pixel in the original grayscale image is greater than the threshold, the value of the pixel in the half=tone image is 255. Otherwise, it is set to 〇. The core social group and its weight value have a very important influence on the quality of the image after halftone conversion, -: the smaller core can make the halftone shadow after the conversion ": wind 0 V, ' In addition, the core of most halftone conversion technologies is fixed, so for 201112756, the thresholds for pixels in the same position are the same, that is, the color image. Technical threats - do the ranks Produce a kind of semi-proposed τ- χ, σ people in order to combine halftone conversion and watermarking technology, neighboring W Γ 具 具 具 的 具 具 , , , , , , , , 调整 调整 调整 调整 调整 调整 调整 调整 调整 调整 调整 调整 调整 调整 调整 调整The threshold of binarization is used to control the black pixel density of the two blocks in each block. In this issue, we will have a Tibetan watermark

Once added seven series (bltstream comment 'Before the image halftone conversion will first ^ ^ knife ... - some blocks of the same size and do not overlap ' each block will produce a solid size phase _ core 'in each core There will be a corresponding pixel binarization when the 1 ke value b (/, y), these threshold values are randomly generated, due to the gate weight, large or magic, for the converted image will produce Large error, so the threshold value will be defined in a certain range _4 and .), where (10) and Μ are the upper and lower limits of the threshold (up b_mGW bGund). In the conversion of the miscellaneous, half-color change is performed for the two rotated blocks at the same time, at the same time, a watermark of a bit is hidden in the 'his hidden; ^ is to compare the two blocks in half The number of black pixels included in the tone conversion, when the Tibetan/watermark bit is to be G, the number of black pixels included in the first block after the half color change is greater than the number of black pixels included in the second block. When the bit to be hidden is 丨, the opposite is true. Of course, after the first halftone conversion, such a rule may not be satisfied, so when an unsatisfactory condition occurs, the kernel must be regenerated until the state of the two blocks satisfies the hidden watermark. The conditions required for the bit. In the method proposed by the present invention, two blocks can be hidden in a watermarking bit, so the size of the block used to cut the image can be determined according to the size of the watermark bit string to be hidden. When the watermark is larger, the smaller block can be used to cut the image to get more blocks. On 201112756 ί: This image halftone conversion (rbih) and the hidden watermarking 1 love color silly removal part is similar to the Tibetan program, can be judged by each pair of blocks, and then - take out The schematic diagram of the method of hiding the watermark Θ is shown in the figure--. [Embodiment]

The invention is a digital watermarking technology based on a random halftone method, and the two parts of the watermark are collected. The detailed implementation method of the present invention is as follows: 1. Image halftone conversion and algorithm for hiding watermarks 丨: input image to be protected (e_image) c and black and white floating to be Tibetan The watermark 'root, the width of the image C αν and the height and the width of the watermark ^v and the south degree determine the size of the block. The width and height of the block are obtained by equations (1) and (2) respectively. . BN = CN WNx2 BM =

CM WM (1)(2) Step 2 · Convert the black and white watermark into a string of length L and use the secret gold sink as the random ritual generator (PRNG) Seed, perform a random re-allocation of all the bits in the bit string to add security 201112756 sex step 3 = the main target of the first two Shangtaohuan areas, where /e{〇, U,.. .,z —1}. Step 4: With the same core, the value of the second core is determined by the equation (3), the number of the S block, (~) is represented by the core of the core center, 乂) The corresponding block uses the threshold value of the pixel ~ 〇, /) value of 6. 'Steps from, 7) = random(1), ω #, 办加(10), ό ^ {1, do (3): Binary the two blocks _ pixel secret ship with the corresponding nuclear power, by the equation such as street # _ 1 槛 threshold code W), and stored in , and ^ respectively. - Valued pixel values (4 '0, when 八力·, Λ <夂^, y); 255, others. Step 6: Calculate the number of pixels in Na and U/2, respectively, and the number of pixels in 5 and 2, that is, the number of black pixels. 1 Step 7: Read - % of the watermark bit to be hidden, will be shouted into the cage 4vc to get the item as "fake, when, repeat the steps otherwise" will get ^^ and 5%, 2 as a possession There is floating water: the corresponding block of the image. 'Color animal, when νι^ = 0 and 10,000%, i S ; TBt true, when wl = l to reverse Βΐ ΙΛ ΙΛ β β β β 其它 其它 其它 其它 其它 ' ' ' ' ' ' ' 5) 201112756 Step 8· Repeat the steps + Step 9: Turn left and κ2 to 8 until the watermark bit is hidden. And, ιηΓ7卩 half-color image to complete image halftone conversion and sub-watermark hiding Procedure: Step 10: Depends on the size of the money and the size of the watermark mixWM.

After the use of the shadow color conversion and material printing hidden technology: f complete _ watermark, the owner of the conversion when saving the size of the print, please \_ must be the image of the second, the watermark reduction processing algorithm Want to use the secret key key used in this (4) Fangcheng material water (4) watermark hiding program and floating two =. In this fortune, the material of the watermark is also listed below. Step 1: The half-color portrait of the private money watermark S, the secret key level used when the image surface is hidden by the watermark and the floating water WNxWM 〇 size Step 2: The width of the image block and the height can be used to image and watermark The size is substituted into equations (1) and (2). Cutting the watermarked image into a number of blocks 5, /b, which are not overlapping and of the same size and Vx5M; /g {0,1,2,...,Z-1}, do e {1,2} . Step 3: Read the first two unprocessed blocks in order, and use % as the main target for current processing. ’... Step 4: Calculate the number of pixels with an internal pixel value of 〇 and 50,000 'U, that is, the number of black pixels. Taking ββ'ζ ι and substituting 12 (6) 201112756 (6), the watermark position of the Tibetan is taken as BB1'幺ββ, ., 1, when ΒΒ''>ββ< If there are 6 unprocessed blocks in the bit/sequence, repeat steps 3 through 5. Step Step 6: Execute through the secret watermark 2, save the floating position - Γ There is a bit set (4) to the original floating water

The corresponding position in the column to generate the restored watermark bit. Step 7: The watermark bitstream of the coffee is broken, and the floating structure restoration process is completed. The results of the continuous test and the analysis of the watermarking technique in the process are like the processing of the Tibetan Linyin when the half-color touchover is carried out. In the experiment, we will evaluate

The iB watermarking technology proposed by the invention is capable of resisting attacks when a plurality of different images are attacked, and the damage attacks used include: adding other words, adding other pictures, cutting and sharpening, and the like. 〃 In the test, 'we use the similarity measure (SM) similarity measure and the bit error ratio (BER) to determine the completeness of the watermark restoration. The definitions are as follows: Shown: / Similarity measure (SM): commonly used to evaluate the similarity between two black and white (binary) images or bit strings, here I use this indicator to present the original watermark y and restore float The similarity of the watermark, the similarity is between -1 and 1. When the similarity between the two is higher, the value is larger; otherwise, the smaller. The formula is as follows: 13 201112756 su(iv^y, L-ϊ ΣχιχΊ 1=0 /=〇V /=〇X/,Λ:',€ {l,-l}. (7) The second middle L is the watermark bit The length, if ice, and < is the position within y and y, when W/(M//) is 0, x, (A) is set to 1; otherwise, X/(Ο is set to -1. Bit Error Rate Indicator (BER): commonly used to evaluate the error rate (%) between two binary images or a bit string 歹j. Here, we use this indicator to present the difference between the original watermark^ and the restored float (4). Degree, the error rate is between Q% and touch %, the higher the error between the two is, the larger the error rate is; the smaller the error is. The formula is as follows: BER(W, W') = \- , r〇, ; 2 —» y, ye{0,l}〇(8) _二中乙 is the length of the watermark bit, if % and W, / is the position in 妒 and W ^ A is 〇 when (6) Set to Q; otherwise, less is a small complement. Like Lenn month; ^ test in the first use - Zhang has 512x512 pixel grayscale shadow silly ('蔺田, 〇 shirt image, and digital watermark is a The black and white image of Zhang 9〇χ9〇 is shown in the figure.) The halftone image of the Tibetan watermark is shown in Figure 2 (8): In the halftone image with the watermark, the restored t-image is exactly the same and H, after the attack, the method proposed by the moon is still in the watermarked image. In this experiment, I added it: II! 印印++-tone image experiments of various attacks, including sub-ports into his eight pictures, cropping and sharpening attack experiments. 201112756 l Image sharpening attack experiment and analysis one in this In the experiment, the image is sharpened by the person (as shown in Figure 3 (8)). The image of the forest-printed image is clearly visible in the human visual image (as shown in Figure 9 (9)). Tear to 〇65, the difference was 17 7%.

2. Experiments and analysis of images added to other text or image attacks. Beta may be added to some text or other images by the general user due to the image on the brush. 'This experiment will study the method proposed in this book (4) to resist the addition of other texts, or The ability of image attack 'Our people carry out the experiment of τ 五 五 · · · · 在 τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ (3) In the image of Yinxian Printing, ^ΦΝΤΙΤ, LAI" and "And,, text, (5) in the image with m water and a small shell-like image, (5) in the image of Cailiu Liu The images of the five cases in this test are shown in Figure 4 (4) to Figure 8 (4). The images (as shown in Figure 4 (8) to Figure 8 (b)) are still visible in human vision. Watermarking image. 3. Experiment and analysis of image cropping attack. This experiment is performed by the following == (the ability of the _ method to resist the cutting attack, I mean four things ===== The lower half of the image of the watermark is 'two-thirds of the image money printed. _. The real H (1) shop The area with floating water is hidden in human vision. 15 201112756 The real money from the various destruction attacks listed above, the watermarking technology proposed by the company can have a considerable degree of resistance to the above-mentioned attacks. 'Therefore, the restored watermark image still has enough discernibility in human vision. The tear and swollen values of the watermark images restored by the above experiments are listed in Table 1. 4. Different images are subject to various attacks. Experiments and comparative analysis~ In order to prove that the present invention can be used to image various images with different scenes, in this experiment ten, in addition to Lena, we used _, Ding You (4), • Ρ-Γ and Barbara The image is used to perform various attacks as described above, and whether the quality of the watermarked image can be maintained at a similar level after each image is attacked by the marriage, and the image is restored to the W and the biliary curve of the watermark image after various attacks. The figures are shown in Figure 14 and Figure 15 respectively. It can be found from the figure that 'each image is very similar whether it is W or job, so the experiment can prove the invention. The watermarking technology cuts the eye #stable and can be used in various images of different scenes. 4. Security analysis • In the aspect of security analysis, 'whether the watermark of the person riding on the scale+ is easily deleted by others, the present invention The proposed watermarking technology uses the secret gold record to confuse the address of each bit of the original watermark bit string to change the order of hiding when hiding the watermark. In addition, the size of the watermark is It also affects the size of the block. Therefore, as long as the level is large enough, and the size of the sink and the watermark is affected by the imagery, the watermark in the image will not be easily Deletion and destruction, in order to ensure the safety of the present invention. 16 201112756 [Simple description of the diagram] Table-: This is the hair and interest of each image. Watermarking Table 2: The Bab〇on image used in the present invention is for tearing and smashing of printed images. Ten pairs of floating water restored by various attacks Table 3: The SM and MR of the tree image Tiffany f 对. The smear/pre-water table 4: The present invention makes the Poppers view the SM and the swelling of various printed images. ^ (4) Son Water Table 5: ^ The hair and shape Ba_a image is for SM and BER of various silk original print images. &amp; Figure 1: Grayscale imagery, conversion, watermarking, and taking out the flow chart. _· This is the original watermark image of Lena reading the original material. (8) and (9) are not yet attacked. It retrieves the watermark image. ° Figure: This is a halftone image of the image and the print image '8' and (b) of the various types of material used by Lena County in the present invention, and the watermark image obtained by the watermark image sharpening. Figure 5: Figure 4: The present invention makes the Lena image in each of the observations and the restored watermark image, and (8) and (9) each add a watermark image to the NTIT" halftone image and its retrieval. The watermark image is the image of the Lena image used in the present invention after various attacks and the restored watermark image, and (8) and (b) respectively add "ΝΤΓΓ" and "LAI, half of the text for the watermark image. Tone image and its retrieval 17 201112756 image. Figure 6: The image of the Lena image used in the present invention after various attacks and the original image (4) and (8) have their own floating image added to the ^^^ Α ", and the butterfly image of the text and its retrieval Figure 7. The image of the invention after the face attack of Lena f彡 and its original watermark image, (4) and (b) each of which is a half-color image after capturing a plurality of shell-like red images with a watermark image. Back to the watermark image. Figure \. The Lena scene used in the present invention is suspected of various images after the attack and its watermark image '(8) and (b) each has a watermark image to recognize the halftone of the outer frame Image and its retrieved watermark image. Figure IX. The image of the Lena f彡 image of the present invention after various attacks and the restored watermark image '(8) and (9) are each for cutting the watermark image. The half-color image of the four-part area and its retrieved watermark image 0 y Figure 10. The Lena image used in the present invention is used in various images of the attack and the fingerprint images of the images ('8) and (9) To crop a halftone image of a region with a watermark image and a cross + (4) Watermarking image. This is the image of the thief-like Lena image in various kinds of images and the original forest image, (4) and (b) each of the silk has a watermark image - a half cross (four) half Tone image and its retrieved watermark image 0, v Figure 10 - The image of Lena image used in the present invention after various attacks and the restored ocean watermark image, (8) and (b) The halftone image of the next sub-region of the watermark image and the watermark image 201112756 of the retrieved image. Figure 13: The watermark image restored by Lena shadow y used by the present invention, (8) and its r - Halftone === Figure 14^ This is a comparison of all the halftones produced by the watermark images restored by the images of the various images. The ugly comparison curve of the watermark image restored by the image after various attacks. 1 Original grayscale image 2 _ — 4 3 Program for hiding watermark 5 Halftone image with ocean watermark 6 Black and white floating Watermark image fetching watermark program

Claims (1)

201112756 VII. The scope of application for patents: 1. A planting based on the randomized gray-scale digital image halftone method = ^ Watermark hiding technology, including Tibetan and retrieving the digital watermark two large λ ^ tune shirts like: Bruise, which is characterized by the digital watermark hiding method: Step 1: Selecting the white watermark image for the grayscale original image to be protected; ~J information black step. : Performing the image halftone conversion of the present invention The Tibetan algorithm with watermarking uses the halftone of the watermarked image; the silk preserves the secret key and the size of the watermark milk Vx milk^·; "Step.: The county (4) that will be produced The consumer image can be published, sold and delivered on the printed matter; the method of digital watermark retrieval: Step 1: Loading the watermark The image; the secret gold size WxfFAZ; Yu You and the step of watermarking, the algorithm to obtain the possession of the watermark based on the application of the old (4) - using the random nuclear order financial image like half-color printing technology, halftone In the image, the method of hiding the black and white watermark is used to preserve the image with the watermark, so that after the processing of the sub-watermark Tibetan is completed, a copyright protection of one image is generated. Watermark 20 201112756 The algorithm for hiding and retrieving the watermark is as follows: [Image Halftone Conversion and Watermarking Encryption Algorithm] [Input] Image to be protected (cover image) C and float to be hidden Watermark 酽; [Output] Halftone image with watermarked v. 1. Enter the image to be protected and the black and white watermark to be hidden. According to the width CW of the image C and the height CM and the width of the watermark, and the height gray determines the size of the image block, and the width and height of the block are respectively determined by the equation (丨) and (2) Acquire; νStep 2. Convert the black and white watermark into a bit string of length ζ, and use the secret key 6ΆΓ as a random random number generator (pSeud〇ran(j〇m number generator ' PRNG) Seed, perform PRNG (the network randomly reassigns the positions of all the bits in the bit string to increase security; Step 3. Read the first two unprocessed blocks in sequence and ] and 2, as The main target of the current processing, 仏 {〇, l, Wi}; Step 4: randomly generate two cores (four) & the same size as the block, the values W, y) contained in the two cores can be obtained by equation (3), Where $ is the arbitrarily set parameter, the block number is used, (/, y_) is the coordinate position in the core, and the W in the core is the threshold value used for the pixels in the body %. ; ' ' Two blocks (10) pixel secret, y) binarized using the corresponding core, Equation (4) obtains the binarized pixel value stored in B, 1 and 3^2; Step 6: f is not calculated, and the number of pixels with an internal pixel value of 0 及 and 氏, 2, 疋The number of black pixels; a step 7: read - a county to be hidden ", will be W, on behalf of the material (5) to obtain 201112756 items 'when the item is "false" 'repeated steps 4 to 7 Otherwise, the corresponding block will be obtained as a halftone image with a watermark hidden; Steps. Repeat steps 3 to 8 until the watermark bit is hidden; Image 51, the program of image halftone conversion and watermark hiding is completed; Step 10: The author or owner of the protected image saves the secret key owed and the size of the watermark WNxfVAf; [Watermark restoration algorithm] [Input] There is a halftone of the watermark, the secret key used when the image is hidden by the watermark 5: and the size of the watermark _&gt;&lt;_; [output] restored watermark bit string 妒| Step 1: Enter a halftone image with a watermark t The size of the secret key and the size of the watermark used in the watermarking are given. Step 2: The width and height of the block can be used to enlarge the image and the watermark (1) and (2) In the middle, the cut has a watermark image ^ is the block of the stack and the size is the same as Mxw, where /e{〇i 夕不重^6(1,2} ; , , step 3. The main targets of the first two unprocessed areas are read sequentially; heart and heart are taken as the current step 4: respectively calculate the number of pixels with % and intracardiac pixel values, ie the number of black pixels, The code "and the view, 2 generations of people '2' are hidden in the watermarking bit 1). Step 5 ·· The watermarking bit &lt; the stored watermark bit is stored in the temporary storage watermark bit. Block, repeat step 3 to step 5. . If still 22 201112756 Step 6: Execute PRNGp fire with the assistance of secret transcript SAT, rearrange all the bit positions in the temporary watermark bit string. Up to the corresponding position in the original watermark bit string to generate a restored watermark bit string ^; Step 7: Loss The restored watermark bit string 妒 ', complete the watermark restore program;