WO2021103676A1 - 一种基于整数小波变换的自适应可逆信息隐藏方法 - Google Patents

一种基于整数小波变换的自适应可逆信息隐藏方法 Download PDF

Info

Publication number
WO2021103676A1
WO2021103676A1 PCT/CN2020/110276 CN2020110276W WO2021103676A1 WO 2021103676 A1 WO2021103676 A1 WO 2021103676A1 CN 2020110276 W CN2020110276 W CN 2020110276W WO 2021103676 A1 WO2021103676 A1 WO 2021103676A1
Authority
WO
WIPO (PCT)
Prior art keywords
watermark
component
embedded
block
information
Prior art date
Application number
PCT/CN2020/110276
Other languages
English (en)
French (fr)
Inventor
刘连山
孟令壮
Original Assignee
山东科技大学
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 山东科技大学 filed Critical 山东科技大学
Publication of WO2021103676A1 publication Critical patent/WO2021103676A1/zh

Links

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
    • G06T1/005—Robust watermarking, e.g. average attack or collusion attack resistant
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2201/00—General purpose image data processing
    • G06T2201/005—Image watermarking
    • G06T2201/0203—Image watermarking whereby the image with embedded watermark is reverted to the original condition before embedding, e.g. lossless, distortion-free or invertible watermarking

Definitions

  • the invention belongs to the field of digital watermarking and information hiding, and specifically relates to an adaptive reversible information hiding method based on integer wavelet transform.
  • Reversible watermarking is an important branch of information hiding technology. This is of great significance for the copyright protection of digital carrier data (such as audio, video, and image) and the prevention of tampering and forgery. Different from traditional watermarks, the complete reversibility and concealment of reversible watermarks play an important role in certain sensitive fields (such as medical images, military satellite images and legal evidence verification). If there is an error between the original image and the watermarked image, it will cause irreparable loss.
  • Watermark embedding method Is an algorithm based on discrete wavelet transform and interleaving prediction, which has a good visual effect, but because the discrete wavelet transform produces decimals, the algorithm cannot be completely reversible.
  • the present invention proposes an adaptive reversible information hiding method based on integer wavelet transform, which is reasonable in design, overcomes the shortcomings of the prior art, and has good effects.
  • An adaptive and reversible information hiding method based on integer wavelet transform including watermark embedding steps, watermark extraction and image restoration steps;
  • the watermark embedding steps are as follows:
  • Step S11 Set the original image I as an M ⁇ N grayscale image, and divide the original image I into a watermark embedding area I 1 and an additional information embedding area I 2 ; if the length of the original image I is greater than 24 pixels, the additional information embedding area Take only one row in I 2 , otherwise it will increase adaptively;
  • Step S13 Perform integer wavelet transform on the watermark embedding area I 1 and divide it into four components A, H, V, and D;
  • Step S14 Decompose component A and component D into 2 ⁇ 2 non-overlapping sub-blocks, which are a 1 , a 2 , a 3 and a 4 in a clockwise direction from the upper left corner;
  • Step S15 Scan all blocks of component A, calculate the threshold T according to the mean square error ⁇ of each block and the size of the watermark information w, and then convert the threshold T, the length and width of the watermark into 8-bit binary, and connect them in sequence to form a binary Additional information b, replacing the additional information b with the lowest bit of the first 24 pixels of the embedded area I 2 with the additional information;
  • Step S16 Scan each block of component A and component S, judge whether it can be embedded according to a certain block in component A, and embed the watermark information into the block corresponding to component D; the specific method is: according to component A The mean square error ⁇ of a block and the threshold T, judge whether the block is an embedded block; if ⁇ T, the block can be embedded smoothly. At this time, the watermark information of the corresponding block in component D is embedded, and the following operations are performed, otherwise the judgment is as follows An image block; after scanning, the embedded component D′ is obtained;
  • Step S162 According to the difference e 1 , the watermark information w i is embedded into a 1 to obtain a 1 ′, as shown in formula (1):
  • a 1 ′ a 2 +2 ⁇ e 1 +w i (1);
  • Step S164 According to the difference e 2 , the watermark information w i is embedded into a 3 to obtain a 3 ′, as shown in formula (2):
  • a 3 ′ a 4 +2 ⁇ e 2 +w i (2);
  • Step S17 image reconstruction
  • Step S18 Connect the modified watermark embedding area I′ 1 and the additional information embedding area I′ 2 to obtain a watermarked image I w ;
  • the watermark extraction and image restoration steps are as follows:
  • Step S21 Divide the watermark embedded image I w into a watermark embedding area I′ 1 and an additional information embedding area I′ 2 , and extract the lowest bits of the first 24 pixels of the additional information embedding area I′ 2 to obtain the embedding Time threshold T, the length and width of the watermark;
  • Step S22 Perform integer wavelet transformation on the watermark embedding area I′ 1 to obtain four components of A, H, V and D′;
  • Step S23 Decompose component A and component D'into 2 ⁇ 2 non-overlapping blocks, which are a 1 , a 1 , a 3 and a 4 clockwise from the upper left corner, scan each block and perform the following operations:
  • Step S231 Scan each block of component A and component D', judge whether information is embedded according to the block in component A, and extract the information from the corresponding block in component D'; the specific method is: according to the block in component A Mean square error ⁇ and threshold T to determine whether the block is an embedded block: If ⁇ T, it means that the block is embedded with information, then perform the following operations on the corresponding block in component D′ to extract the information, otherwise, determine the next block; scan After completion, get the restored component D;
  • Step S234 Restore the original part of the carrier according to formula (3):
  • Step S237 Restore the original components of the carrier image according to formula (4):
  • Step S24 Image reconstruction, performing integer wavelet reconstruction on the components A, H, V and the restored component D, and obtaining the watermark embedding area I 1 of the original image according to the inverse transformation;
  • Step S25 Replace the first 24 bits of the additional information embedded area I′ 2 with the last 24 bits of the extracted embedded information b;
  • Step S26 Connect the restored watermark embedding area I 1 and the modified additional information embedding area I 2 to obtain the original carrier image I.
  • the method for calculating the threshold T is as follows: first, set the threshold T to 1. Since each image block is embedded with 2bit data, it is adaptive when the number of image blocks with ⁇ T is just greater than the watermark size. Set the minimum threshold, otherwise T plus 1 to re-judge.
  • the invention discloses an adaptive reversible information hiding method based on integer wavelet transform, which is aimed at the problems of cumbersome steps and poor effect of reversible watermarking algorithm, and at the same time increases the embedding capacity of the watermark, and simplifies the algorithm steps while maintaining the visual effect of the image.
  • embedding there is no need to perform complex preprocessing, and when extracting the watermark, you do not need to know the watermark size and other information; compared with the traditional wavelet transform, the watermark embedding is carried out on the basis of the integer wavelet transform lifting transform, which can realize the complete extraction and extraction of the watermark.
  • Reversible restoration of the original image adaptively set the threshold according to the mean square error of the approximate component (low-frequency part) in the wavelet coefficients and the size of the watermark; at the same time, the noise in the diagonal component (high-frequency part) is not easy to be noticed by the human eye.
  • Embedding the watermark in the middle can increase the PSNR value and improve the visual effect of the watermarked image; the threshold and watermark information are also embedded in the image, which can achieve complete blind extraction; this application proposes to embed in the diagonal component according to the smoothness of the approximate component
  • the watermarking method is based on whether a certain image block in the approximate component is smooth (whether the mean square error of the block is less than the threshold), if it is smooth, the watermark is embedded in the corresponding diagonal component, otherwise the next block is judged; the integrity of the watermark is used to determine Determine whether the ciphertext is damaged, and finally restore the original image without distortion; on the basis of ensuring complete reversibility and embedding capacity, the method of the present invention has high PSNR of the image after embedding the watermark, high watermark invisibility, simple and easy experiment, and has better performance. Good visual effects have practical value.
  • Figure 1 is a schematic diagram of image decomposition.
  • Figure 2 is a schematic diagram of additional information of the present invention.
  • Fig. 3 is a schematic diagram of a block image of the present invention.
  • Figure 4 is a flowchart of watermark embedding.
  • Figure 5 is a flow chart of watermark extraction.
  • An adaptive and reversible information hiding method based on integer wavelet transform including watermark embedding steps, watermark extraction and image restoration steps;
  • Step S11 Set the original image I as an M ⁇ N grayscale image, and divide the original image I into a watermark embedding area I 1 and an additional information embedding area I 2 ; as shown in Figure 1; if the length of the original image I is greater than 24 pixels , Then only one line is taken in the additional information embedding area I 2 , otherwise it is adaptively increased;
  • Step S13 Perform integer wavelet transform on the watermark embedding area I 1 and divide it into four components A, H, V, and D;
  • Step S14 Decompose component A and component D into 2 ⁇ 2 non-overlapping sub-blocks, which are a 1 , a 2 , a 3 and a 4 in a clockwise direction from the upper left corner; as shown in Figure 3;
  • Step S15 Scan all the blocks of component A, and calculate the threshold T according to the mean square error ⁇ of each block and the size of the watermark information w.
  • the calculation method is: first set the threshold T to 1. Because this method embeds 2bit data for each image block, When the number of image blocks with ⁇ T is just greater than the size of the watermark, it is the minimum threshold set adaptively, otherwise T plus 1 is re-judged; then the threshold T, the length and width of the watermark are converted to 8-bit binary, and sequentially Connecting to form the binary additional information b, as shown in Figure 2; replacing the additional information b with the lowest bit of the first 24 pixels of the embedded area I 2;
  • Step S16 Scan each block of component A and component D, judge whether it can be embedded according to a certain block in component A, and embed the watermark information into the block corresponding to component D; the specific method is: according to component A The mean square error ⁇ of a block and the threshold T, judge whether the block is an embedded block; if ⁇ T, the block can be embedded smoothly. At this time, the watermark information of the corresponding block in component D is embedded, and the following operations are performed, otherwise the judgment is as follows An image block; after scanning, the embedded component D′ is obtained;
  • Step S162 According to the difference e 1 , the watermark information w i is embedded into a 1 to obtain a 1 ′, as shown in formula (1):
  • a 1 ′ a 2 +2 ⁇ e 1 +w i (1);
  • Step S164 According to the difference e 2 , the watermark information w i is embedded into a 3 to obtain a 3 ′, as shown in formula (2):
  • a 3 ′ a 4 +2 ⁇ e 2 +w i (2);
  • Step S17 image reconstruction
  • Step S18 Connect the modified watermark embedding area I′ 1 and the additional information embedding area I′ 2 to obtain a watermarked image I w ;
  • Step S21 Divide the watermark embedded image I w into a watermark embedding area I′ 1 and an additional information embedding area I′ 2 , and extract the lowest bits of the first 24 pixels of the additional information embedding area I′ 2 to obtain the embedding Time threshold T, the length and width of the watermark;
  • Step S22 Perform integer wavelet transformation on the watermark embedding area I′ 1 to obtain four components of A, H, V and D′;
  • Step S23 Decompose component A and component D'into 2 ⁇ 2 non-overlapping blocks, which are a 1 , a 2 , a 3 and a 4 clockwise from the upper left corner, scan each block and perform the following operations:
  • Step S231 Scan each block of component A and component D', judge whether information is embedded according to the block in component A, and extract the information from the corresponding block in component D'; the specific method is: according to the block in component A Mean square error ⁇ and threshold T to determine whether the block is an embedded block: If ⁇ T, it means that the block is embedded with information, then perform the following operations on the corresponding block in component D′ to extract the information, otherwise, determine the next block; scan After completion, get the restored component D;
  • Step S234 Restore the original part of the carrier according to formula (3):
  • Step S237 Restore the original components of the carrier image according to formula (4):
  • Step S24 Image reconstruction, performing integer wavelet reconstruction on the components A, H, V and the restored component D, and obtaining the watermark embedding area I 1 of the original image according to the inverse transformation;
  • Step S25 Replace the first 24 bits of the additional information embedded area I′ 2 with the last 24 bits of the extracted embedded information b;
  • Step S26 Connect the restored watermark embedding area I 1 and the modified additional information embedding area I 2 to obtain the original carrier image I.
  • the original image is divided into the watermark embedding area and the additional information embedding area according to the method shown in Figure 1.
  • the calculation method is: first set the threshold T to 1. Since this method embeds 2bit data for each image block, when ⁇ T When the number of image blocks is just greater than the watermark size, it is the minimum threshold set adaptively, otherwise T plus 1 is re-judged.
  • the threshold value T and the length and width of the watermark are sequentially composed of binary information according to Figure 2, replacing the lowest bit of the first 24 pixels of the additional information embedding area.
  • a 1 ′ a 2 +2 ⁇ e 1 +w i (1)
  • the image After scanning the approximate component A, the image is restored, and the components A, H, V and the restored component D are reconstructed by integer wavelet.
  • the watermark embedding area I 1 of the original image is obtained; the additional information is embedded in the area
  • the first 24 bits of I′ 2 are replaced with the last 24 bits of the extracted embedded information b; I 1 and I 2 are connected to obtain the original carrier image I.
  • the method proposed by the present invention based on integer wavelet transform is relatively simple to implement for grayscale images, but overall the watermark embedding capacity is relatively small, and the watermark embedding and extraction are simple and easy to understand.
  • PSNR Peak Signal to Noise Ratio (peak signal-to-noise ratio) values are all above 50. The impact on the pixel value of the image is minimal, and after the watermark is proposed, the image can be restored without distortion, and the reversible watermark can be realized.
  • This method can be evaluated from both threshold and PSNR: the threshold reflects the smoothness of the image itself, the higher the threshold, the lower the smoothness of the image itself, and the higher the perception of the human eye after the image with low smoothness is embedded; PSNR is used A more extensive objective evaluation index for image quality, the higher the PSNR value, the smaller the image distortion.
  • PSNR PSNR
  • Table 1 lists the PSNR and threshold values of six images of Lena, Baboon, Boat, Couple, Man, and Peppers when they are embedded with a 64 ⁇ 64 watermark.

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Editing Of Facsimile Originals (AREA)
  • Compression Or Coding Systems Of Tv Signals (AREA)
  • Image Processing (AREA)

Abstract

一种基于整数小波变换的自适应可逆信息隐藏方法，属于数字水印和信息隐藏领域。该方法首先将载体图像分为两个区域：水印嵌入区域和附加信息嵌入区域；水印嵌入区域进行整数小波变换，利用水印大小和小波系数的均方差自适应地设置阈值；根据近似分量确定某图像块是否嵌入，然后将水印信息嵌入对应的对角分量中；附加信息嵌入区域用于嵌入附加信息，包括阈值、水印的长度和宽度。该方法能够实现自适应地嵌入水印，并能做到完全的盲提取；不仅可以实现可逆信息隐藏，而且具有较好的视觉效果，失真度小，嵌入水印后图像的PSNR高，水印不可见性较好，具有实用价值。

Description

一种基于整数小波变换的自适应可逆信息隐藏方法 技术领域
本发明属于数字水印和信息隐藏领域，具体涉及一种基于整数小波变换的自适应可逆信息隐藏方法。
背景技术
随着人们安全意识和知识产权意识的提高，人们对自身信息的安全度也越来越重视，信息隐藏和数字水印技术的使用也越来越多。可逆水印是信息隐藏技术的重要分支。这对于数字载体数据(例如音频，视频和图像)的版权保护，防止篡改和伪造等方面具有重要意义。与传统水印不同，可逆水印的完全可逆性和隐蔽性在某些敏感领域(例如医学图像，军事卫星图像和法律证据验证)中发挥着重要作用。如果原始图像与提取水印后的图像之间存在误差，将导致无法弥补的损失。
文献“Xiong L,Xu Z,Shi Y Q.An integer wavelet transform based scheme for reversible data hiding in encrypted images”(International Conference on Information Science&Technology.IEEE,2012)提出了一种自回归的自适应可逆数据隐藏方法，该算法为不同的图像自适应地设置一个阈值，用来将图像划分平滑区域和纹理区域，通过最小二乘法，得到像素值的预测，但由于其采用了空间域嵌入，使得其鲁棒性不好。文献“Li F,Mao Q,Chang C C.Reversible data hiding scheme based on the Haar discrete wavelet transform and interleaving prediction method”(Multimedia Tools and Applications,2017,77:(8)1-20.)的水印嵌入方法，是基于离散小波变换和交织预测的算法，具有较好的视觉效果，但由于离散小波变换产生了小数而使得该算法不能完全可逆。
发明内容
针对现有技术中存在的上述技术问题，本发明提出了一种基于整数小波变换的自适应可逆信息隐藏方法，设计合理，克服了现有技术的不足，具有良好的效果。
为了实现上述目的，本发明采用如下技术方案：
一种基于整数小波变换的自适应可逆信息隐藏方法，包括水印嵌入步骤以及水印提取和图像恢复步骤；
其中，水印嵌入步骤如下：
步骤S11：设原图像I为M×N的灰度图像，将原图像I划分成水印嵌入区域I 1和附加信息嵌入区域I 2；如果原图像I的长大于24像素，则附加信息嵌入区域I 2中只取一行，否则自适应增加；
步骤S12：提取附加信息嵌入区域I 2的前24位像素的最低位lsb(i),i＝1,2,…,24，将其 连接到水印信息之后，形成含有最低有效位LSB的水印信息w i；
步骤S13：对水印嵌入区域I 1进行整数小波变换，分为A、H、V、D四个分量；
步骤S14：将分量A和分量D分解为2×2的非重叠子块，从左上角按顺时针方向依次为a 1、a 2、a 3和a 4；
步骤S15：扫描分量A的全部块，根据每块的均方差σ和水印信息w的大小计算阈值T，然后将阈值T、水印的长度和宽度分别转换为8位二进制，并依次连接组成二进制的附加信息b，将附加信息b替换附加信息嵌入区域I 2的前24位像素的最低位；
步骤S16：扫描分量A和分量S的每个块，根据分量A中某块判断是否可嵌入，并将水印信息嵌入到该块在分量D所对应的块中；具体方法为：根据分量A中某块的均方差σ和阈值T，判断该块是否为嵌入块；如果σ＜T，说明该块平滑能够嵌入，此时对分量D中相应的块嵌入水印信息，执行以下操作，否则判断下一个图像块；扫描完成后，得到嵌入后的分量D′；
步骤S161：计算a 1和a 2的差值e 1：e 1＝a 1-a 2；
步骤S162：根据差值e 1，将水印信息w i嵌入到a 1中，得到a 1′，如公式(1)所示：
a 1′＝a 2+2×e 1+w i       (1)；
步骤S163：计算a 3和a 4的差值e 2：e 2＝a 3-a 4；
步骤S164：根据差值e 2，将水印信息w i嵌入到a 3中，得到a 3′，如公式(2)所示：
a 3′＝a 4+2×e 2+w i      (2)；
步骤S17：图像重构；
对分量A、H、V和嵌入后的分量D′进行整数小波重构，得到嵌入后的水印嵌入区域I′ 1；
步骤S18：连接修改后的水印嵌入区域I′ 1和附加信息嵌入区域I′ 2，得到嵌入水印的图像I w；
水印提取和图像恢复步骤如下：
步骤S21：将已嵌入水印信息的图像I w划分为水印嵌入区域I′ 1和附加信息嵌入区域I′ 2，并提取附加信息嵌入区域I′ 2的前24位像素的最低位，从而获得嵌入时的阈值T、水印的长度和宽度；
步骤S22：对水印嵌入区域I′ 1进行整数小波变换，得到A、H、V和D′的四个分量；
步骤S23：将分量A和分量D′分解为2×2的不重叠块，从左上角顺时针依次为a 1、a 1、a 3和a 4，对每个块进行扫描并执行以下操作：
步骤S231：扫描分量A和分量D′的每个块，根据分量A中的块判断是否嵌入了信息，并从分量D′中对应的块提取信息；具体方法为：根据分量A中某块的均方差σ和阈值T，判断该块是否为嵌入块：如果σ＜T，说明该块嵌入了信息，则对分量D′中对应的块执行以下操作提取信息，否则，判断下一个块；扫描完成后，得到恢复后的分量D；
步骤S232：计算a′ 1和a 2的差值e′ 1：e′ 1＝a′ 1-a 2；
步骤S233：根据预测误差e′ 1，得到嵌入信息b＝LSB(e′ 1)；
步骤S234：根据公式(3)恢复载体的原始部分：
Figure PCTCN2020110276-appb-000001
其中，
Figure PCTCN2020110276-appb-000002
表示向下取整；
步骤S235：计算＝a′ 3和a 4的差值e′ 2：e′ 2＝a′ 3-a 4；
步骤S236：根据预测误差e′ 2，得到嵌入信息b＝LSB(e′ 2)；
步骤S237：根据公式(4)恢复载体图像的原始分量：
Figure PCTCN2020110276-appb-000003
步骤S24：图像重构，将分量A，H，V和恢复后的分量D进行整数小波重构，根据逆变换得到原始图像的水印嵌入区域I 1；
步骤S25：将附加信息嵌入区域I′ 2的前24位替换为所提取到的嵌入信息b的后24位；
步骤S26：连接恢复后的水印嵌入区域I 1和修改后的附加信息嵌入区域I 2，得到原始的载体图像I。
优选地，在步骤S15中，计算阈值T方法如下：首先置阈值T为1，由于每个图像块嵌入2bit数据，因此当σ＜T的图像块的个数刚好大于水印大小时即为自适应设置的最小阈值，否则T加1重新判断。
本发明所带来的有益技术效果：
本发明公开了一种基于整数小波变换的自适应可逆信息隐藏方法，针对可逆水印算法步骤繁琐，效果差等问题，同时提高水印的嵌入容量，在保持图像视觉效果的前提下，简化了算法步骤，嵌入时，不需要进行复杂的预处理，提取水印时，不需要知道水印大小等信息；对比传统的小波变换，在整数小波变换提升变换的基础上进行水印嵌入，能够实现水印的完全提取和原图像的可逆恢复；根据小波系数中近似分量(低频部分)的均方差与水印大小自适应地设置阈值；同时，对角分量(高频部分)中的噪声不易被人眼察觉，在此分量中嵌入水印能够提高PSNR值，使含水印图像的视觉效果提高；阈值和水印信息也嵌入到图像中，能够实现完全的盲提取；本申请提出了根据近似分量的平滑度在对角分量中嵌入水印的方法，根据近似分量中某图像块是否平滑(该块的均方差是否小于阈值)，若平滑，则在对应的对角分量中嵌入水印，否则继续判断下一块；通过水印的完整度来判断密文是否遭受破坏，最终可以无失真恢复原图像；本发明方法在确保完全可逆和嵌入容量的基础上，嵌入水印后图像的PSNR高，水印不可见性高，实验简单易行，具有较好的视觉效果，具有实用价值。
附图说明
图1为图像分解示意图。
图2为本发明的附加信息示意图。
图3为本发明的分块图像示意图。
图4为水印嵌入流程图。
图5为水印提取流程图。
具体实施方式
下面结合附图以及具体实施方式对本发明作进一步详细说明：
一种基于整数小波变换的自适应可逆信息隐藏方法，包括水印嵌入步骤以及水印提取和图像恢复步骤；
其中，水印嵌入(其流程如图4所示)步骤如下：
步骤S11：设原图像I为M×N的灰度图像，将原图像I划分成水印嵌入区域I 1和附加信息嵌入区域I 2；如图1所示；如果原图像I的长大于24像素，则附加信息嵌入区域I 2中只取一行，否则自适应增加；
步骤S12：提取附加信息嵌入区域I 2的前24位像素的最低位lsb(i),i＝1,2,…,24，将其连接到水印信息之后，形成含有最低有效位LSB的水印信息w i；
步骤S13：对水印嵌入区域I 1进行整数小波变换，分为A、H、V、D四个分量；
步骤S14：将分量A和分量D分解为2×2的非重叠子块，从左上角按顺时针方向依次为a 1、a 2、a 3和a 4；如图3所示；
步骤S15：扫描分量A的全部块，根据每块的均方差σ和水印信息w的大小计算阈值T，计算方法为：首先置阈值T为1，由于本方法每个图像块嵌入2bit数据，因此当σ＜T的图像块的个数刚好大于水印大小时即为自适应设置的最小阈值，否则T加1重新判断；然后将阈值T、水印的长度和宽度分别转换为8位二进制，并依次连接组成二进制的附加信息b，如图2所示；将附加信息b替换附加信息嵌入区域I 2的前24位像素的最低位；
步骤S16：扫描分量A和分量D的每个块，根据分量A中某块判断是否可嵌入，并将水印信息嵌入到该块在分量D所对应的块中；具体方法为：根据分量A中某块的均方差σ和阈值T，判断该块是否为嵌入块；如果σ＜T，说明该块平滑能够嵌入，此时对分量D中相应的块嵌入水印信息，执行以下操作，否则判断下一个图像块；扫描完成后，得到嵌入后的分量D′；
步骤S161：计算a 1和a 2的差值e 1：e 1＝a 1-a 2；
步骤S162：根据差值e 1，将水印信息w i嵌入到a 1中，得到a 1′，如公式(1)所示：
a 1′＝a 2+2×e 1+w i      (1)；
步骤S163：计算a 3和a 4的差值e 2：e 2＝a 3-a 4；
步骤S164：根据差值e 2，将水印信息w i嵌入到a 3中，得到a 3′，如公式(2)所示：
a 3′＝a 4+2×e 2+w i     (2)；
步骤S17：图像重构；
对分量A、H、V和嵌入后的分量D′进行整数小波重构，得到嵌入后的水印嵌入区域I′ 1；
步骤S18：连接修改后的水印嵌入区域I′ 1和附加信息嵌入区域I′ 2，得到嵌入水印的图像I w；
水印提取和图像恢复(其流程如图5所示)步骤如下：
步骤S21：将已嵌入水印信息的图像I w划分为水印嵌入区域I′ 1和附加信息嵌入区域I′ 2，并提取附加信息嵌入区域I′ 2的前24位像素的最低位，从而获得嵌入时的阈值T、水印的长度和宽度；
步骤S22：对水印嵌入区域I′ 1进行整数小波变换，得到A、H、V和D′的四个分量；
步骤S23：将分量A和分量D′分解为2×2的不重叠块，从左上角顺时针依次为a 1、a 2、a 3和a 4，对每个块进行扫描并执行以下操作：
步骤S231：扫描分量A和分量D′的每个块，根据分量A中的块判断是否嵌入了信息，并从分量D′中对应的块提取信息；具体方法为：根据分量A中某块的均方差σ和阈值T，判断该块是否为嵌入块：如果σ＜T，说明该块嵌入了信息，则对分量D′中对应的块执行以下操作提取信息，否则，判断下一个块；扫描完成后，得到恢复后的分量D；
步骤S232：计算a′ 1和a 2的差值e′ 1：e′ 1＝a′ 1-a 2；
步骤S233：根据预测误差e′ 1，得到嵌入信息b＝LSB(e′ 1)；
步骤S234：根据公式(3)恢复载体的原始部分：
Figure PCTCN2020110276-appb-000004
其中，
Figure PCTCN2020110276-appb-000005
表示向下取整；
步骤S235：计算＝a′ 3和a 4的差值e′ 2：e′ 2＝a′ 3-a 4；
步骤S236：根据预测误差e′ 2，得到嵌入信息b＝LSB(e′ 2)；
步骤S237：根据公式(4)恢复载体图像的原始分量：
Figure PCTCN2020110276-appb-000006
步骤S24：图像重构，将分量A，H，V和恢复后的分量D进行整数小波重构，根据逆变换得到原始图像的水印嵌入区域I 1；
步骤S25：将附加信息嵌入区域I′ 2的前24位替换为所提取到的嵌入信息b的后24位；
步骤S26：连接恢复后的水印嵌入区域I 1和修改后的附加信息嵌入区域I 2，得到原始的载 体图像I。
具体实施例如下：
水印嵌入，其流程如图4所示
设图像大小为M×N，其灰度值为I(i,j)(0≤I(i,j)≤255)，将水印图像转换为水印序列w，本申请选取的是64×64的二值水印图片，转化为1×4096的水印序列，嵌入水印后的图像用I w表示，如图4所示：
水印嵌入具体过程如下：
首先对原图像按照图1方式划分为水印嵌入区域和附加信息嵌入区域。
然后对水印嵌入区域进行整数小波变换，对变换后的近似分量和对角分量按照图3进行分块。扫描分量A的全部块，根据各块的均方差σ和水印序列大小计算出阈值T，计算方法为：首先置阈值T为1，由于本方法每个图像块嵌入2bit数据，因此当σ＜T的图像块的个数刚好大于水印大小时即为自适应设置的最小阈值，否则T加1重新判断。将阈值T、水印的长和宽按照图2依次组成二进制信息，替换掉附加信息嵌入区域前24位像素的最低位。
扫描近似分量A和对角分量D的每个图像块，对于分量A从第一个分块开始，计算均方差σ判断是否可以嵌入，若σ＜T，则该块平滑可嵌入，进行如下操作，否则判断下一块。
计算a 1和a 2的差值e 1:e 1＝a 1-a 2；将水印信息w根据a 2嵌入到a 1，公式如下：
a 1′＝a 2+2×e 1+w i     (1)
同样，计算a 3和a 4的差值e 2＝a 3-a 4，并将水印信息嵌入到a 3中，公式如下：
a 3′＝a 4+2×e 2+w i    (2)
扫描完近似分量A后，对分量A、H、V和嵌入后的分量D′进行整数小波重构，并将其与修改后的附加信息嵌入区域连接，得到嵌入水印的图像I w；
水印提取和图像恢复，其流程如图5所示，
首先进行水印提取，对嵌入水印后的图像进行处理，分块方式仍是按照图1的方式进行划分处理，划分为水印嵌入区域I′ 1和附加信息嵌入区域I′ 2，并提取附加信息嵌入区域的前24位像素的最低位，从而得到阈值T，水印长度和宽度。
对水印嵌入区域I′ 1进行整数小波变换，得到A、H、V和D′的四个分量，将近似分量A和对角分量D′按照图3分解为2×2的不重叠图像块，根据分量A中该块的均方误差σ和提取到的阈值T确定该块是否为可嵌入块：如果σ＜T，则对分量D′中的对应块进行以下操作，否则，判断下一个图像块；
计算a′ 1和a 2的差值e′ 1：e′ 1＝a′ 1-a 2，根据预测误差e′ 1提取嵌入信息：b＝LSB(e′ 1)，通过以下公式恢复载体的原始部分(
Figure PCTCN2020110276-appb-000007
表示向下取整，下同)：
Figure PCTCN2020110276-appb-000008
类似地，计算出差值e′ 2＝a′ 3-a 4，得到嵌入信息b＝LSB(e′ 2)，然后恢复载体图像的原始分量：
Figure PCTCN2020110276-appb-000009
扫描完近似分量A后，再对图像进行恢复，将分量A，H，V和恢复后的分量D进行整数小波重构，根据逆变换得到原始图像的水印嵌入区域I 1；将附加信息嵌入区域I′ 2的前24位替换为所提取到的嵌入信息b的后24位；连接I 1和I 2，得到原始的载体图像I。
由于可逆水印算法通常比较复杂，本发明提出的基于整数小波变换的，对于灰度图像而言，实现比较简单，但是总体上水印的嵌入容量比较少，水印的嵌入和提取简单易懂，PSNR(Peak Signal to Noise Ratio，峰值信噪比)值都在50以上。对图像的像素值影响极小，并且提出水印后，可以无失真恢复图像，实现可逆水印。
本方法可以从阈值和PSNR两方面进行评价：阈值所体现的是图像本身平滑度，阈值越高，说明图像本身平滑度低，平滑度低的图像嵌入后人眼感知度越高；PSNR是使用较为广泛的图像质量客观评价指标，PSNR值越高，说明图像失真越小。
PSNR的计算方式如下：
Figure PCTCN2020110276-appb-000010
其中：
Figure PCTCN2020110276-appb-000011
表1列出了Lena、Baboon、Boat、Couple、Man和Peppers六幅图像嵌入64×64水印时的PSNR和阈值。
表1六幅图像的PSNR和阈值对比
Figure PCTCN2020110276-appb-000012
当然，上述说明并非是对本发明的限制，本发明也并不仅限于上述举例，本技术领域的技术人员在本发明的实质范围内所做出的变化、改型、添加或替换，也应属于本发明的保护范围。

Claims (2)

  1. 一种基于整数小波变换的自适应可逆信息隐藏方法，其特征在于：包括水印嵌入步骤以及水印提取和图像恢复步骤；
    其中，水印嵌入步骤如下：
    步骤S11：设原图像I为M×N的灰度图像，将原图像I划分成水印嵌入区域I 1和附加信息嵌入区域I 2；如果原图像I的长大于24像素，则附加信息嵌入区域I 2中只取一行，否则自适应增加；
    步骤S12：提取附加信息嵌入区域I 2的前24位像素的最低位lsb(i),i＝1,2,…,24，将其连接到水印信息之后，形成含有最低有效位LSB的水印信息w i；
    步骤S13：对水印嵌入区域I 1进行整数小波变换，分为A、H、V、D四个分量；
    步骤S14：将分量A和分量D分解为2×2的非重叠子块，从左上角按顺时针方向依次为a 1、a 2、a 3和a 4；
    步骤S15：扫描分量A的全部块，根据每块的均方差σ和水印信息w的大小计算阈值T，然后将阈值T、水印的长度和宽度分别转换为8位二进制，并依次连接组成二进制的附加信息b，将附加信息b替换附加信息嵌入区域I 2的前24位像素的最低位；
    步骤S16：扫描分量A和分量D的每个块，根据分量A中某块判断是否可嵌入，并将水印信息嵌入到该块在分量D所对应的块中；具体方法为：根据分量A中某块的均方差σ和阈值T，判断该块是否为嵌入块；如果σ＜T，说明该块平滑能够嵌入，此时对分量D中相应的块嵌入水印信息，执行以下操作，否则判断下一个图像块；扫描完成后，得到嵌入后的分量D′；
    步骤S161：计算a 1和a 2的差值e 1：e 1＝a 1-a 2；
    步骤S162：根据差值e 1，将水印信息w i嵌入到a 1中，得到a 1′，如公式(1)所示：
    a 1′＝a 2+2×e 1+w i  (1)；
    步骤S163：计算a 3和a 4的差值e 2：e 2＝a 3-a 4；
    步骤S164：根据差值e 2，将水印信息w i嵌入到a 3中，得到a 3′，如公式(2)所示：
    a 3′＝a 4+2×e 2+w i  (2)；
    步骤S17：图像重构；
    对分量A、H、V和嵌入后的分量D′进行整数小波重构，得到嵌入后的水印嵌入区域I′ 1；
    步骤S18：连接修改后的水印嵌入区域I′ 1和附加信息嵌入区域I′ 2，得到嵌入水印的图像I w；
    水印提取和图像恢复步骤如下：
    步骤S21：将已嵌入水印信息的图像I w划分为水印嵌入区域I′ 1和附加信息嵌入区域I′ 2，并提取附加信息嵌入区域I′ 2的前24位像素的最低位，从而获得嵌入时的阈值T、水印的长度和宽度；
    步骤S22：对水印嵌入区域I′ 1进行整数小波变换，得到A、H、V和D′的四个分量；
    步骤S23：将分量A和分量D′分解为2×2的不重叠块，从左上角顺时针依次为a 1、a 2、a 3和a 4，对每个块进行扫描并执行以下操作：
    步骤S231：扫描分量A和分量D′的每个块，根据分量A中的块判断是否嵌入了信息，并从分量D′中对应的块提取信息；具体方法为：根据分量A中某块的均方差σ和阈值T，判断该块是否为嵌入块：如果σ＜T，说明该块嵌入了信息，则对分量D′中对应的块执行以下操作提取信息，否则，判断下一个块；扫描完成后，得到恢复后的分量D；
    步骤S232：计算a′ 1和a 2的差值e′ 1：e′ 1＝a′ 1-a 2；
    步骤S233：根据预测误差e′ 1，得到嵌入信息b＝LSB(e′ 1)；
    步骤S234：根据公式(3)恢复载体的原始部分：
    Figure PCTCN2020110276-appb-100001
    其中，
    Figure PCTCN2020110276-appb-100002
    表示向下取整；
    步骤S235：计算＝a′ 3和a 4的差值e′ 2：e′ 2＝a′ 3-a 4；
    步骤S236：根据预测误差e′ 2，得到嵌入信息b＝LSB(e′ 2)；
    步骤S237：根据公式(4)恢复载体图像的原始分量：
    Figure PCTCN2020110276-appb-100003
    步骤S24：图像重构，将分量A，H，V和恢复后的分量D进行整数小波重构，根据逆变换得到原始图像的水印嵌入区域I 1；
    步骤S25：将附加信息嵌入区域I′ 2的前24位替换为所提取到的嵌入信息b的后24位；
    步骤S26：连接恢复后的水印嵌入区域I 1和修改后的附加信息嵌入区域I 2，得到原始的载体图像I。
  2. 根据权利要求1所述的基于整数小波变换的自适应可逆信息隐藏方法，其特征在于：在步骤S15中，计算阈值T方法如下：首先置阈值T为1，由于每个图像块嵌入2bit数据，因此当σ＜T的图像块的个数刚好大于水印大小时即为自适应设置的最小阈值，否则T加1重新判断。
PCT/CN2020/110276 2019-11-29 2020-08-20 一种基于整数小波变换的自适应可逆信息隐藏方法 WO2021103676A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201911196081.6A CN110910299B (zh) 2019-11-29 2019-11-29 一种基于整数小波变换的自适应可逆信息隐藏方法
CN201911196081.6 2019-11-29

Publications (1)

Publication Number Publication Date
WO2021103676A1 true WO2021103676A1 (zh) 2021-06-03

Family

ID=69820436

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/110276 WO2021103676A1 (zh) 2019-11-29 2020-08-20 一种基于整数小波变换的自适应可逆信息隐藏方法

Country Status (2)

Country Link
CN (1) CN110910299B (zh)
WO (1) WO2021103676A1 (zh)

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110910299B (zh) * 2019-11-29 2021-01-05 山东科技大学 一种基于整数小波变换的自适应可逆信息隐藏方法
CN113205445B (zh) * 2021-04-16 2023-10-17 北京大学 一种基于数据压缩的可逆明水印方法和装置
CN114493969A (zh) * 2022-01-24 2022-05-13 西安闻泰信息技术有限公司 基于dwt的数字水印方法、系统、电子设备及存储介质
CN115393151A (zh) * 2022-08-19 2022-11-25 淮阴工学院 基于iwt和预测差值直方图平移的可逆水印算法
CN115631084B (zh) * 2022-10-14 2024-06-21 淮阴工学院 基于iwt和插值扩展的可逆水印方法
CN116308980B (zh) * 2023-02-28 2024-06-21 淮阴工学院 基于块最优动态选择的可逆信息隐藏方法

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2006044802A2 (en) * 2004-10-20 2006-04-27 New Jersey Institute Of Technology System and method for lossless data hiding using the integer wavelet transform
US20080002851A1 (en) * 2006-05-08 2008-01-03 Shi Yun Q Data hiding with wavelet transform histogram shifting
CN105741224A (zh) * 2016-01-28 2016-07-06 广东工业大学 基于pvo和自适应块分割的可逆水印算法
CN106228501A (zh) * 2016-07-13 2016-12-14 西安电子科技大学 一种基于整数变换的自适应可逆数字水印方法
CN110458747A (zh) * 2019-07-29 2019-11-15 淮阴工学院 基于不变特征的鲁棒可逆图像水印算法
CN110910299A (zh) * 2019-11-29 2020-03-24 山东科技大学 一种基于整数小波变换的自适应可逆信息隐藏方法

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106485640A (zh) * 2016-08-25 2017-03-08 广东工业大学 一种基于多层次ipvo的可逆水印计算方法

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2006044802A2 (en) * 2004-10-20 2006-04-27 New Jersey Institute Of Technology System and method for lossless data hiding using the integer wavelet transform
US20080002851A1 (en) * 2006-05-08 2008-01-03 Shi Yun Q Data hiding with wavelet transform histogram shifting
CN105741224A (zh) * 2016-01-28 2016-07-06 广东工业大学 基于pvo和自适应块分割的可逆水印算法
CN106228501A (zh) * 2016-07-13 2016-12-14 西安电子科技大学 一种基于整数变换的自适应可逆数字水印方法
CN110458747A (zh) * 2019-07-29 2019-11-15 淮阴工学院 基于不变特征的鲁棒可逆图像水印算法
CN110910299A (zh) * 2019-11-29 2020-03-24 山东科技大学 一种基于整数小波变换的自适应可逆信息隐藏方法

Also Published As

Publication number Publication date
CN110910299B (zh) 2021-01-05
CN110910299A (zh) 2020-03-24

Similar Documents

Publication Publication Date Title
WO2021103676A1 (zh) 一种基于整数小波变换的自适应可逆信息隐藏方法
Ernawan et al. A robust image watermarking technique with an optimal DCT-psychovisual threshold
Bhatnagar et al. A new robust reference watermarking scheme based on DWT-SVD
CN102103738B (zh) 数字图像篡改内容可恢复的变容量水印生成与认证方法
CN110232650B (zh) 一种彩色图像水印嵌入方法、检测方法及系统
Meng et al. An adaptive reversible watermarking in IWT domain
Chakraborty et al. Reversible color image watermarking using trigonometric functions
US7356158B2 (en) Methods and apparatus for lossless data hiding
Ernawan et al. Image watermarking based on integer wavelet transform-singular value decomposition with variance pixels
CN105512999A (zh) 一种双变换的彩色图像全息水印方法
Song et al. Recent advances and classification of watermarking techniques in digital images
CN115330582B (zh) 基于单向极值预测误差扩展的可逆水印方法
CN115908095A (zh) 一种基于分层注意力特征融合的鲁棒图像水印方法及系统
Loukhaoukha et al. Hybrid watermarking algorithm based on SVD and lifting wavelet transform for ownership verification
CN114037593B (zh) 基于逆向直方图平移的可逆图像水印算法
Shi et al. A Blind Watermarking Technique for Color Image based on SVD with Circulation.
Kejgir et al. Lifting wavelet transform with singular value decomposition for robust digital image watermarking
Bedi et al. 2L-DWTS—Steganography technique based on second level DWT
CN109493270B (zh) 一种基于slt-dm的水印图像还原方法
Qiang et al. Color image self-embedding and watermarking based on DWT
CN111242830B (zh) 基于极谐变换的图像可逆鲁棒数据隐藏方法
Kadu et al. Secure data hiding using robust firefly algorithm
Hu et al. A blind watermarking algorithm for color image based on wavelet transform and Fourier transform
Maity et al. Reversible image watermarking using modified difference expansion
Zheng et al. Histogram based watermarking algorithm resist to geometric attacks

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20892904

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20892904

Country of ref document: EP

Kind code of ref document: A1