A self-recovery watermarking method based on integer DWT

Abstract

This invention discloses a self-recovering watermarking method based on integer DWT, belonging to the field of digital watermarking technology. The method includes: performing integer DWT on the original image to obtain low-frequency and high-frequency subband coefficients; obtaining watermarked low-frequency coefficients based on the low-frequency coefficients and the watermark; integerizing these low-frequency coefficients to obtain integerized low-frequency coefficients and first auxiliary information; combining the high-frequency subband coefficients to obtain a first reconstructed image, and then overflowing to obtain a second reconstructed image and second auxiliary information; reversibly embedding the two types of auxiliary information into the second reconstructed image to obtain the watermarked image. The watermarked image is reversibly extracted. If the auxiliary information can be extracted, the image is recovered by combining it with the high-frequency subband coefficients; otherwise, the watermarked image is transformed and quantized, and then combined with the transformed high-frequency subband coefficients to recover the image. This invention balances reversibility and robustness, achieving lossless recovery without attack and still allowing watermark extraction and image recovery even during attack.

Description

Technical Field

[0001] This invention belongs to the field of digital watermarking technology, and particularly relates to a self-recovering watermarking method based on integer DWT. Background Technology

[0002] With the rapid development of internet and digital multimedia technologies, digital images are increasingly widely used in information dissemination, data storage, and network sharing. However, digital images are vulnerable to illegal copying, alteration, or malicious attacks during dissemination, threatening the copyright and integrity of digital content. Therefore, effectively protecting the copyright and integrity of digital images has become an important research issue in the field of information security.

[0003] Digital watermarking is an important information hiding technology that embeds invisible information into digital images to achieve functions such as copyright protection, content authentication, and tamper detection. Depending on the application requirements, digital watermarking algorithms can generally be divided into two categories: robust watermarking and reversible watermarking.

[0004] Robust watermarking algorithms can extract watermark information even after an image has been subjected to attacks such as compression, noise, and cropping, thus having significant application value in the field of digital rights protection. However, most robust watermarking algorithms cause irreversible distortion to the original image after embedding the watermark, making it impossible to recover the original image content.

[0005] Reversible watermarking algorithms can completely recover the original image after extracting the watermark information, thus having significant application value in fields with high requirements for image integrity, such as medical imaging, legal evidence collection, and military imaging. However, traditional reversible watermarking algorithms often struggle to correctly extract watermark information and recover damaged image content when images are attacked.

[0006] In practical applications, such as digital copyright protection, judicial evidence collection, and medical image protection, it is not only necessary to extract watermark information, but also to recover some image content after the image has been tampered with or attacked. Therefore, how to design a watermarking method that can both achieve image recovery and maintain the extractability of the watermark when the image is attacked has become an important research problem in the field of digital watermarking. Summary of the Invention

[0007] To address the aforementioned technical problems, this invention proposes a self-recovery watermarking method based on integer DWT, thereby resolving the issues present in the prior art.

[0008] To achieve the above objectives, this invention provides a self-recovery watermarking method based on integer DWT, comprising:

[0009] Integer discrete wavelet transform is performed on the original image to obtain low-frequency coefficients and high-frequency subband coefficients;

[0010] Based on the low-frequency coefficients and watermark information, the watermarked low-frequency coefficients are obtained.

[0011] The low-frequency coefficients containing the watermark are integerized to obtain the integerized low-frequency coefficients and the first auxiliary information.

[0012] The first reconstructed image is obtained based on the integerized low-frequency coefficients and the high-frequency sub-band coefficients;

[0013] A second reconstructed image and second auxiliary information are obtained based on the first reconstructed image;

[0014] The first auxiliary information and the second auxiliary information are reversibly embedded into the second reconstructed image to obtain a watermarked image;

[0015] The watermarked image is reversibly extracted to determine whether auxiliary information can be successfully extracted. If successful, image restoration is performed based on the extracted auxiliary information and the high-frequency subband coefficients. If unsuccessful, the watermarked image is transformed and quantized sequentially, and then the high-frequency subband coefficients obtained from the transformation are used to restore the image.

[0016] Optionally, the process of obtaining the watermarked low-frequency coefficients based on the low-frequency coefficients and watermark information includes:

[0017] The watermark information is embedded into the low-frequency coefficients using a quantization index modulation method to obtain quantized low-frequency coefficients; the low-frequency coefficients and the quantized low-frequency coefficients are weighted and summed according to a preset embedding strength parameter to obtain the watermarked low-frequency coefficients.

[0018] Optionally, the process of integerizing the low-frequency coefficients of the watermark includes:

[0019] The watermarked low-frequency coefficients are rounded down to obtain the integerized low-frequency coefficients; the difference between the watermarked low-frequency coefficients and the integerized low-frequency coefficients is taken as the fractional part; the fractional part is processed into binary to obtain the first auxiliary information.

[0020] Optionally, the process of obtaining the second reconstructed image and the second auxiliary information based on the first reconstructed image includes:

[0021] In the first reconstructed image, pixels with values ​​less than 0 are set to 0, and pixels with values ​​greater than 255 are set to 255 to obtain the second reconstructed image; the positions and correction information of the overflowing pixels are extracted as the second auxiliary information.

[0022] Optionally, the process of obtaining the watermarked image includes:

[0023] The merged first and second auxiliary information are embedded into the second reconstructed image using a reversible information hiding algorithm to obtain the watermarked image.

[0024] Optionally, if successful, the image restoration process based on the extracted auxiliary information and the high-frequency subband coefficients includes:

[0025] The watermarked image is reversibly extracted to obtain the extracted second reconstructed image, the extracted first auxiliary information, and the extracted second auxiliary information;

[0026] Based on the extracted second auxiliary information, the extracted second reconstructed image is subjected to reverse overflow processing to obtain the restored first watermarked image;

[0027] The restored first watermarked image is subjected to integer discrete wavelet transform to obtain the restored integer low-frequency coefficients;

[0028] Based on the extracted first auxiliary information, the fractional part of the recovered integer low-frequency coefficients is recovered to obtain the watermarked low-frequency coefficients.

[0029] The recovered watermarked low-frequency coefficients are subjected to quantization index modulation decision to extract watermark information and recover the original low-frequency coefficients;

[0030] An integer discrete wavelet inverse transform is performed based on the original low-frequency coefficients and the high-frequency subband coefficients to obtain a lossless restored image.

[0031] Optionally, if the process fails, the image restoration process, which involves sequentially transforming and quantizing the watermarked image and then combining the high-frequency subband coefficients obtained from the transformation, includes:

[0032] The watermarked image is subjected to integer discrete wavelet transform to obtain the attacked low-frequency coefficients and the transformed high-frequency subband coefficients; the attacked low-frequency coefficients are subjected to quantization index modulation decision to extract the watermark information and recover the original low-frequency coefficients; the original low-frequency coefficients and the transformed high-frequency subband coefficients are used to perform inverse integer discrete wavelet transform to obtain the approximate recovered image.

[0033] Optionally, the integer discrete wavelet transform is a multi-level integer discrete wavelet transform; the integer discrete wavelet transform uses Haar wavelets.

[0034] The present invention provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described method.

[0035] The present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method.

[0036] Compared with the prior art, the present invention has the following advantages and technical effects:

[0037] 1) Achieve lossless recovery under attack-free conditions;

[0038] This invention quantizes and embeds low-frequency coefficients, recording decimal and overflow information. During watermark extraction, it utilizes auxiliary information to accurately reconstruct the image. Under unattack-free conditions, it achieves accurate watermark extraction and lossless restoration of the carrier image, avoiding the permanent distortion of the original image caused by traditional robust watermarking algorithms.

[0039] 2) Possesses self-recovery capabilities under attack conditions;

[0040] This invention introduces a low-frequency coefficient difference compensation mechanism during the embedding process, enabling the watermark information to contain some of the original image structure information. When the image is subjected to attacks such as noise or compression, even if the auxiliary information cannot be completely extracted, the low-frequency coefficients can still be estimated and recovered based on the watermark information, thereby achieving an approximate reconstruction of the carrier image and improving the usability of the image.

[0041] 3) Improve the robustness of watermark extraction;

[0042] This invention embeds the watermark into the low-frequency subband of an integer DWT and uses the Quantization Index Modulation (QIM) method for embedding, so that the watermark information can still be stably extracted when subjected to common signal processing attacks (including but not limited to JPEG compression, additive noise, etc.), thereby meeting the requirements of digital rights authentication.

[0043] 4) Balancing reversibility and robustness;

[0044] This invention combines robust watermark embedding with reversible information hiding to achieve complete reversible recovery under attack-free conditions, while still ensuring watermark extractability and a certain degree of image restoration capability under attack conditions, effectively overcoming the problem that traditional reversible watermarks and robust watermarks are difficult to achieve simultaneously.

[0045] 5) The algorithm has a simple structure and is easy to implement;

[0046] This invention is based on integer discrete wavelet transform and quantization modulation. It does not rely on complex models or training processes, has low computational complexity, is easy to implement in engineering, and can be widely used in fields such as digital copyright protection, tamper detection, and secure image transmission. Attached Figure Description

[0047] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0048] Figure 1 This is a schematic diagram of the self-recovery watermark embedding algorithm based on integer DWT according to an embodiment of the present invention;

[0049] Figure 2 This is a schematic diagram illustrating the process of effectively extracting watermarks and losslessly restoring carrier images under unattacked conditions, according to an embodiment of the present invention.

[0050] Figure 3 This is a schematic diagram illustrating the process of effectively extracting watermarks and restoring carrier images under attack conditions according to an embodiment of the present invention;

[0051] Figure 4 The PSNR diagrams of the watermarked image, the attacked image, and the restored image under JPEG attack are shown in the embodiments of the present invention.

[0052] Figure 5 The PSNR diagrams of the watermarked image, the attacked image, and the restored image under JPEG2000 attack are shown in this embodiment of the invention.

[0053] Figure 6 This is a PSNR diagram of a watermarked image, an attacked image, and a recovered image under an AWGN attack, according to an embodiment of the present invention. Detailed Implementation

[0054] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0055] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0056] Example 1

[0057] This embodiment provides a self-recovering watermarking method based on integer DWT, including a watermark embedding process, a process of effectively extracting watermark information and losslessly recovering the carrier image when not under attack, and a process of effectively extracting watermark information and losslessly recovering the carrier image when under attack.

[0058] The watermark embedding process is as follows Figure 1 As shown, the specific steps are as follows:

[0059] 1) For the original image Perform L-level integer DWT to obtain the L-th level low-frequency subband coefficients. and the high-frequency subband coefficients of each layer ,in These represent the wavelet horizontal high-frequency subband coefficients, vertical high-frequency subband coefficients, and diagonal high-frequency subband coefficients of the l-th layer, respectively.

[0060] 2) Remove watermark The watermarked low-frequency coefficients are embedded into the low-frequency coefficients A using quantization index modulation (QIM) to obtain the initial watermarked low-frequency coefficients. Then the difference between these two coefficients is added to the total coefficients in a certain proportion. In the process, the final low-frequency coefficients containing the watermark are obtained. .

[0061] 3) Add watermarked low-frequency coefficients Rounding down yields the first auxiliary information Aux1 and the integerized low-frequency coefficients. Subsequently The first reconstructed image is obtained by performing multi-level integer IDWT on the high-frequency subband coefficients. .

[0062] 4) For the first reconstructed image Overflow processing is performed to obtain the second auxiliary information Aux2 and the second reconstructed image. .

[0063] 5) The first auxiliary information Aux1 and the second auxiliary information Aux2 are reversibly embedded into the second reconstructed image. Obtain watermarked image .

[0064] When not under attack, the process of effectively extracting watermark information and losslessly restoring the carrier image is as follows: Figure 2 As shown, the extracted or recovered parameters correspond one-to-one with the parameters obtained in the embedding stage, and their values ​​are consistent under normal circumstances. The specific steps are as follows:

[0065] 6) For the transmitted watermarked images Perform reversible extraction to obtain auxiliary information. Image after extracting auxiliary information and auxiliary information .

[0066] 7) Utilize auxiliary information right Perform reverse overflow processing to obtain the image. .

[0067] 8) For the image Perform multi-level integer DWT based on auxiliary information. To obtain the low-frequency coefficients, add decimals. .

[0068] 9) For low-frequency coefficients QIM extraction and self-recovery were performed to obtain low-frequency coefficients. and watermark .

[0069] 10) Based on the low-frequency coefficient Multi-level integer IDWT is performed with the high-frequency subband coefficients to obtain the recovered carrier image. .

[0070] In the event of an attack, the process of effectively extracting the watermark and restoring the carrier image is as follows: Figure 3 As shown, the specific steps are as follows:

[0071] 11) For noisy images (attacked images) Perform multi-layer integer DWT to obtain the attacked low-frequency coefficients. and high-frequency subband coefficients.

[0072] 12) Low-frequency coefficients under attack QIM extraction and self-recovery were performed to obtain the recovered low-frequency coefficients. and watermark .

[0073] 13) Based on the low-frequency coefficient By performing multi-level integer IDWT on the high-frequency subband coefficients, an approximate reconstructed image is obtained. .

[0074] Step 1) specifically refers to:

[0075] 101. Perform multi-level integer discrete wavelet decomposition on the original image. Let the original image be I. Perform L-level integer DWT on it to obtain the low-frequency coefficients. and high-frequency subband coefficient It is represented as follows:

[0076] .

[0077] Step 2) specifically refers to:

[0078] 201. Construct a quantization index modulation model and set a watermark. ,in, Let represent the k-th watermark bit, and K represent the watermark length. Define the quantization function. in: Indicates when embedded bits The corresponding quantization result at that time , Represents the original low-frequency coefficients. To quantize the step size, This is the initial offset. .

[0079] 202. Quantize and embed watermarks for low-frequency coefficients, using the following formula:

[0080] .

[0081] 203. Embedding intensity modulation: Introducing the embedding intensity parameter α, the coefficients are obtained. The specific formula is as follows:

[0082] .

[0083] in, , The larger, The closer ,otherwise The closer .

[0084] Step 3) specifically involves:

[0085] 301. Integerize low-frequency coefficients by rounding down the embedded coefficients: , This is for rounding down.

[0086] 302. Extracting the decimal part Constructing auxiliary information ,in , This indicates that a decimal integer is converted into a binary representation of length n, with zeros padded to the high bits if the number of bits is less than n.

[0087] 303. Reconstructing the image using integer IDWT involves inverse transforming the low-frequency coefficients with the high-frequency subband coefficients. The specific formula is as follows:

[0088] .

[0089] Step 4) specifically involves:

[0090] 401. For the reconstructed image Overflow handling is performed, which involves constraining the range of pixel values. The specific formula is as follows:

[0091] ;

[0092] 402. Record the overflow location and correction information to construct auxiliary information:

[0093] .

[0094] Step 5) specifically involves:

[0095] 501. Combine the two parts of auxiliary information:

[0096] .

[0097] 502. Combine the auxiliary information Reversible embedding into an image middle:

[0098] .

[0099] in, It is a reversible information hiding algorithm that can completely recover the embedded information under attack-free conditions.

[0100] Step 6) specifically involves:

[0101] 601. Reversible data extraction: Extracting the received watermarked image. As input, it is extracted using a reversible information extraction algorithm:

[0102] ;

[0103] in, It is the image after extracting auxiliary information. It is the extracted auxiliary information.

[0104] 602. Auxiliary information separation: The auxiliary information is split into two parts:

[0105]

[0106] in, It is low-frequency coefficient decimal information. It is an overflow message.

[0107] Step 7) specifically involves:

[0108] 701. Based on auxiliary information Perform reverse overflow processing on the image:

[0109] .

[0110] Step 8) specifically involves:

[0111] 801. Regarding the image Perform multi-level integer DWT:

[0112] .

[0113] 802. Based on auxiliary information Restore the decimal part:

[0114] ;

[0115] The recovered low-frequency coefficients are obtained:

[0116] .

[0117] Step 9) specifically involves:

[0118] 901. Based on quantization index modulation, make a decision on each low-frequency coefficient and extract the watermark information:

[0119] ;

[0120] .

[0121] in, Represents the coefficient to be detected and the quantizer The distance between corresponding quantization results, when If the value is 0, the watermark bit is determined to be 1; otherwise, it is determined to be 1.

[0122] 902. Restore the original low-frequency coefficients based on the judgment result:

[0123] .

[0124] Step 10) specifically involves:

[0125] 1001. Perform an inverse transform between the recovered low-frequency coefficients and the high-frequency subband coefficients to obtain the recovered carrier image (lossless restoration image):

[0126] .

[0127] Step 11) specifically involves:

[0128] 1101. Because the watermarked image has been attacked, the reversible information extraction algorithm can no longer extract auxiliary information. Therefore, the watermarked image after the attack is directly processed. Perform a multi-level integer DWT, where the superscript 'a' represents the data after the attack:

[0129] .

[0130] Step 12) specifically involves:

[0131] 1201. Based on quantization index modulation, make a decision on each low-frequency coefficient and extract the watermark information:

[0132] ;

[0133] ;

[0134] in, The coefficients to be detected and the quantizer after being attacked The distance between corresponding quantification results.

[0135] 1202. Restore the low-frequency coefficients according to the judgment:

[0136] .

[0137] Step 13) specifically involves:

[0138] 1301. Perform an inverse transform between the recovered low-frequency coefficients and the high-frequency subband coefficients to obtain the recovered carrier image:

[0139] .

[0140] This embodiment provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above method.

[0141] This embodiment provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method.

[0142] The invention will be further illustrated below with specific examples.

[0143] (1) Experimental subjects and parameter settings;

[0144] In this embodiment, standard grayscale images Barbara, Elaine, Lake, Peppers and 50 randomly selected images from the BOSSBase image dataset were chosen as experimental subjects, with a uniform image size of 512×512.

[0145] The watermark information uses a 128-bit binary sequence. The integer discrete wavelet transform uses Haar wavelets, with a decomposition level of 5. The quantization step size in quantization index modulation is... Set to 16, initial offset The embedding strength parameter α is set to 0. It is used to adjust the weight between the quantization result and the original coefficient. The larger the value of α, the closer the embedding result is to the quantization value, thereby improving robustness, but correspondingly reducing the self-recovery capability. In this example, α is set to 0.75.

[0146] (2) Watermark embedding example;

[0147] The watermark embedding process includes the following steps:

[0148] a) Perform a 5-level integer discrete wavelet transform on the original image to obtain the low-frequency coefficients. and high-frequency subband coefficient ;

[0149] b) Watermark information is embedded in the low-frequency coefficients using a quantization index modulation method, and the quantization result is adjusted by the embedding intensity parameter α. Perform interpolation compensation to obtain the low-frequency coefficients after embedding the watermark. The specific formula is as follows:

[0150] ;

[0151] ;

[0152] .

[0153] c) For low-frequency coefficients Integerization is performed to obtain and extract The decimal part is used as auxiliary information, and the specific formula is as follows:

[0154] ;

[0155] ;

[0156] .

[0157] d) Perform integer inverse wavelet transform on the processed low-frequency coefficients and high-frequency subband coefficients to obtain the reconstructed image. :

[0158] .

[0159] e) Reconstructing the image Pixel value range constraints are performed, and overflow positions and correction information are recorded as auxiliary information. The specific formula is as follows:

[0160] ;

[0161] .

[0162] f) Embed the aforementioned auxiliary information into the image using a reversible information hiding method to obtain the final watermarked image. The specific formula is as follows:

[0163] ;

[0164] .

[0165] (3) Example of extraction and recovery under unattacked conditions;

[0166] The process of watermark extraction and image restoration under unattacked conditions includes the following steps:

[0167] a) The received watermarked image Reversible information extraction is performed to obtain auxiliary information and restore the image. The specific formula is as follows:

[0168] ;

[0169] .

[0170] b) Based on auxiliary information The specific formula for inverting overflow in an image is as follows:

[0171] .

[0172] c) For the image Perform integer discrete wavelet transform and recover the fractional part of the low-frequency coefficients based on auxiliary information. The specific formula is as follows:

[0173] ;

[0174] ;

[0175] .

[0176] d) For low-frequency coefficients Quantitative judgment is performed to extract watermark information and restore the original low-frequency coefficients. The specific formula is as follows:

[0177] ;

[0178] ;

[0179] .

[0180] e) Perform an inverse integer wavelet transform on the recovered low-frequency coefficients and high-frequency subband coefficients to obtain the recovered image. The specific formula is as follows:

[0181] .

[0182] (4) Performance verification under low-intensity attack conditions;

[0183] The verification process under low-intensity attack conditions includes the following steps:

[0184] a) Images containing watermarks JPEG compression (quality factor Q=90), JPEG2000 compression (compression ratio C=3), and additive white Gaussian noise attack (variance δ=0.0001) were applied respectively.

[0185] b) Perform multi-level integer discrete wavelet transform on the attacked image to obtain low-frequency coefficients and high-frequency sub-band coefficients;

[0186] c) Quantify and determine the low-frequency coefficients to extract watermark information;

[0187] d) Recover the low-frequency coefficients based on the quantization decision results, and perform integer inverse wavelet transform on the recovered low-frequency coefficients and high-frequency subband coefficients to obtain the recovered image;

[0188] e) Calculate the peak signal-to-noise ratio (PSNR) for the watermarked image, the attacked image, and the restored image. Perform statistical analysis on 50 images selected from the BOSSBase dataset to obtain PSNR variation curves under different attack conditions.

[0189] (5) Robustness verification under high-intensity attacks;

[0190] The verification process under high-intensity attack conditions includes the following steps:

[0191] a) Apply JPEG compression, JPEG2000 compression, and additive Gaussian noise attacks of varying intensities to the watermarked image;

[0192] b) Extract watermark information under various attack conditions and calculate the bit error rate (BER) of the watermark.

[0193] c) Statistical analysis of the bit error rate under different attack intensities, and presentation in tabular form;

[0194] d) Analyze the stability of watermark extraction based on the changes in bit error rate.

[0195] Experimental results are as follows Figures 4-6 As shown in Tables 1 to 3, the results are analyzed as follows:

[0196] (1) Image restoration performance analysis under low-intensity attacks;

[0197] Depend on Figures 4-6It can be seen that under JPEG compression (Q=90), JPEG2000 compression (C=3), and low-intensity Gaussian noise attack (δ=0.0001), the PSNR of the attacked image decreases significantly; the PSNR of the image restored by the method of this invention is higher than that of the attacked image; the PSNR of the restored image is close to the quality level of the watermarked image. This indicates that the present invention can effectively restore image structural information under low-intensity attack conditions and has good self-recovery capability.

[0198] (2) Robustness analysis under JPEG attack;

[0199] As shown in Table 1, when the quality factor Q ≥ 20, the bit error rate (BER) of each test image is close to 0; even under strong compression (Q = 10), the BER remains at a low level (less than 2%). This indicates that the method of the present invention has strong robustness to JPEG compression and can stably extract watermark information under conventional compression conditions.

[0200] Table 1

[0201]

[0202] (3) Robustness analysis under JPEG2000 attack;

[0203] As shown in Table 2, under low compression ratio conditions (e.g., C≤30), the watermark can be extracted with virtually no errors. As the compression ratio increases, the bit error rate gradually rises, but remains within an acceptable range overall. This indicates that the method of this invention still possesses a certain degree of robustness when facing stronger compression attacks.

[0204] Table 2

[0205]

[0206] (4) Robustness analysis under Gaussian noise attack;

[0207] As shown in Table 3, the bit error rate is low under low noise intensity (e.g., δ≤0.01); as the noise intensity increases, the bit error rate gradually rises, but the main watermark information can still be extracted. This indicates that the method of the present invention has a certain anti-interference ability against Gaussian noise interference.

[0208] Table 3

[0209]

[0210] Based on the above experimental results, it can be concluded that the present invention achieves fully reversible restoration under no-attack conditions; achieves image self-restoration and watermark extraction under low-intensity attack conditions; and maintains watermark robustness under high-intensity attack conditions. Therefore, the present invention achieves a good balance between image restoration capability and watermark robustness.

[0211] The above are merely preferred embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A self-recovery watermarking method based on integer DWT, characterized in that, Includes the following steps: Integer discrete wavelet transform is performed on the original image to obtain low-frequency coefficients and high-frequency subband coefficients; Based on the low-frequency coefficients and watermark information, the watermarked low-frequency coefficients are obtained. The low-frequency coefficients containing the watermark are integerized to obtain integerized low-frequency coefficients and first auxiliary information; The process of integerizing the low-frequency coefficients of the watermark includes: The low-frequency coefficients containing the watermark are rounded down to obtain the integerized low-frequency coefficients; the difference between the low-frequency coefficients containing the watermark and the integerized low-frequency coefficients is taken as the fractional part; the fractional part is processed into binary to obtain the first auxiliary information. The first reconstructed image is obtained based on the integerized low-frequency coefficients and the high-frequency sub-band coefficients; A second reconstructed image and second auxiliary information are obtained based on the first reconstructed image; The process of obtaining the second reconstructed image and the second auxiliary information based on the first reconstructed image includes: In the first reconstructed image, pixels with values ​​less than 0 are set to 0, and pixels with values ​​greater than 255 are set to 255 to obtain the second reconstructed image; the positions and correction information of the overflowing pixels are extracted as the second auxiliary information. The first auxiliary information and the second auxiliary information are reversibly embedded into the second reconstructed image to obtain a watermarked image; The watermarked image is reversibly extracted to determine whether auxiliary information can be successfully extracted. If successful, image restoration is performed based on the extracted auxiliary information and the high-frequency subband coefficients. If unsuccessful, the watermarked image is transformed and quantized sequentially, and then the high-frequency subband coefficients obtained from the transformation are used to restore the image.

2. The self-recovery watermarking method based on integer DWT according to claim 1, characterized in that, The process of obtaining the watermarked low-frequency coefficients based on the low-frequency coefficients and watermark information includes: The watermark information is embedded into the low-frequency coefficients using a quantization index modulation method to obtain quantized low-frequency coefficients; the low-frequency coefficients and the quantized low-frequency coefficients are weighted and summed according to a preset embedding strength parameter to obtain the watermarked low-frequency coefficients.

3. The self-recovery watermarking method based on integer DWT according to claim 1, characterized in that, The process of obtaining the watermarked image includes: The merged first and second auxiliary information are embedded into the second reconstructed image using a reversible information hiding algorithm to obtain the watermarked image.

4. The self-recovery watermarking method based on integer DWT according to claim 1, characterized in that, If successful, the image restoration process based on the extracted auxiliary information and the high-frequency subband coefficients includes: The watermarked image is reversibly extracted to obtain the extracted first auxiliary information, the second reconstructed image after extracting the auxiliary information, and the extracted second auxiliary information. Based on the extracted second auxiliary information, the second reconstructed image after extracting the auxiliary information is subjected to inverse overflow processing to obtain the restored first reconstructed image; The first reconstructed image after restoration is subjected to integer discrete wavelet transform to recover the integerized low-frequency coefficients and high-frequency subband coefficients. The integer low-frequency coefficients are supplemented with decimals based on the extracted first auxiliary information to obtain the recovered watermarked low-frequency coefficients. The recovered low-frequency coefficients containing the watermark are subjected to quantization index modulation decision, the watermark information is extracted and the original low-frequency coefficients are recovered; An integer discrete wavelet inverse transform is performed based on the original low-frequency coefficients and the high-frequency subband coefficients to obtain a lossless restored image.

5. The self-recovery watermarking method based on integer DWT according to claim 1, characterized in that, If the process fails, the image restoration process, which involves sequentially transforming and quantizing the watermarked image and then combining the high-frequency subband coefficients obtained from the transformation, includes: The watermarked image is subjected to integer discrete wavelet transform to obtain the attacked low-frequency coefficients and the transformed high-frequency subband coefficients; the attacked low-frequency coefficients are subjected to quantization index modulation decision to extract the watermark information and recover the original low-frequency coefficients; the original low-frequency coefficients and the transformed high-frequency subband coefficients are used to perform inverse integer discrete wavelet transform to obtain the approximate recovered image.

6. The self-recovery watermarking method based on integer DWT according to claim 1, characterized in that, The integer discrete wavelet transform is a multi-level integer discrete wavelet transform; the integer discrete wavelet transform uses Haar wavelets.

7. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the steps of the method according to any one of claims 1-6.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the steps of the method described in any one of claims 1-6.

Images