CN113434828A - Intelligent terminal image divulgence tracking and copyright authentication method based on digital watermarking - Google Patents

Abstract

The invention discloses an intelligent terminal image divulgence tracking and copyright authentication method based on digital watermarking. The invention can realize copyright authentication and leakage tracing of the image under a complex network environment by repeatedly embedding the watermark information with small capacity and the locator in the texture complex area of the image for many times; JPEG compression attack, cutting attack and noise attack can be effectively resisted; under JPEG compression with quantization factor of more than 70%, under cutting attack of 1/4, or under noise attack with signal-to-noise ratio of more than 50, the watermark extraction accuracy rate reaches more than 80%; when embedding, the system synchronously embeds the digital watermark information without delay and pause in sense.

Description

Intelligent terminal image divulgence tracking and copyright authentication method based on digital watermarking

Technical Field

The invention relates to the technical field of information security and digital evidence obtaining, in particular to an intelligent terminal image divulgence tracking and copyright authentication method based on digital watermarking.

Background

Optical information hiding is an ever-new discipline [1 ]. Ancient Chinese Tibetan poetry is simple, original and popular information hiding, and plays an important role in the battle of two military. With the development of digital technology, information hiding is endowed with new meanings, and secret information is hidden in a host signal by mainly utilizing insensitivity of human sense organs and redundancy existing in the digital signal, and the effect and the use value of the host signal are not influenced. 28/8/2020, the national ministry of commerce and the ministry of science and technology jointly issue announcements to add new information hiding and discovering technologies and information disguising technologies into the contents of the adjustment of the technical catalog of export restriction of Chinese prohibited export.

In order to realize secure transmission of secret messages in a network, information hiding techniques typically hide secret information in a digital multimedia bunker such as text, images, and video for covert communication. Compared with the traditional cryptography, the information hiding technology can send the information to the opposite side in a stealing mode, so that the doubts of attackers are not easy to cause, namely, the secret communication event can be hidden. In conventional information hiding studies, researchers have made a slight modification to the multimedia bunker (carrier) mainly by exploiting its redundancy, so that secret information is embedded in the carrier in an "invisible" manner.

In the digital evidence obtaining field, one of branches of information hiding is adopted, and the digital watermark has less information quantity, can be embedded into a carrier object under the condition of not influencing the quality of the carrier, and can realize image leakage tracking and copyright authentication. The main digital watermarking methods at present can be roughly classified into a spatial domain concealment method, a frequency domain (transform domain) concealment method, and an image adaptive digital watermarking method based on an STC frame.

The image digital watermarking method of the airspace comprises the following steps: the spatial domain method directly superimposes the information to be embedded in the digital image, namely directly modifies the pixel value of the carrier image to embed the digital watermark, and the whole process is carried out in the spatial domain of the image. Among them, the most typical one is based on LSB (least significant bit) of image, which was first proposed by Trikel et al [2] in 1993. Early LSB methods were embedding digital watermarks by replacing the least significant bits of the image pixel values directly with secret information [3] or by modifying the LSB match [4 ]. Some information hiding tools based on the LSB method such as "mantelseg", "White Noise Storm", and "StegoDos" were also produced. Thereafter, in order to ensure the quality of the carrier image after embedding the watermark, many researchers have proposed some improved spatial domain information hiding methods on the basis of LSB, such as Gray Level Modification (GLM) [5], Pixel Indicator Technique (PIT) [6], Pixel Value difference method (PVD) [7], Pixel Pair Matching (PPM) [8], Predictive coding [9], Histogram Shifting [10], and so on.

The image information hiding method of the frequency domain comprises the following steps: the spatial domain algorithm mainly embeds the digital watermark into an unimportant area, so that secret information is likely to be lost due to compression, filtering and the like in the carrier transmission process. The spatial domain method has high performance in terms of embedding capacity, but is weak against statistical attacks. Therefore, the scholars propose to embed the watermark in the frequency domain (transform domain), i.e. to first convert the image from the spatial domain to the transform domain and then to implement the watermark embedding. For example, the secret information is embedded in the Transform domain such as the discrete Cosine Transform (dct) (discrete Cosine Transform) domain [11], the Wavelet Transform (WT) domain [12], the Fourier Transform (FT) domain [13], and the like. In theory, secret information can be embedded in such a transform domain as long as a transform can be performed on the image.

The image self-adaptive digital watermarking method based on the STC frame comprises the following steps: in order to ensure that the carrier needs to be modified as little as possible while ensuring watermark embedding. Filler and Fridrich et al [14] propose an image adaptation information embedding framework based on STC (motion-Trellis codes) coding. STC defines a general form of distortion function, and minimizes the distortion function value by a coding technique when embedding information. The adaptive embedding method is enabled to embed the watermark while taking into account the cost of modifying the carrier image, thereby minimizing the distortion function designed for capturing statistical feature anomalies. The self-adaptive information hiding method utilizes the conditions that human eyes or a steganalysis method have different sensibility to the modification behaviors of different content characteristic areas of the image according to the characteristics of the carrier, and adjusts the mode of embedding the secret information according to the specific characteristics of different carriers or different areas of the carrier, so that the distortion is minimized as much as possible to ensure the image quality. The method mainly embeds information as much as possible in an area with relatively complex image texture, and embeds little or no information in an area with relatively smooth texture. A representative method is a HUGO method proposed in document [15], which measures distortion cost by the change of a feature vector before and after embedding secret information, and embeds the secret information in the complex texture and edge regions of a carrier image. The WOW method proposed by the document [16] adopts a directional filter to evaluate distortion, reduces embedded modification at smooth edges of an image, and has improved resistance to SRM [17] detection compared with the HUGO method. The S-UNIWARD method proposed in the document [18] optimizes the distortion function adopted in the document [16], and the method performance is basically similar to that of the WOW method. The HILL method proposed in document [19] adopts high-pass and low-pass filters to diffuse the distortion cost of a texture complex region to its neighborhood, and improves the distortion function defined in WOW, so that secret information is more intensively embedded into the texture complex region which is difficult to detect. Similarly, document [20] proposes a channel-dependent load partitioning strategy based on amplifying the channel modification probability, thereby adaptively allocating the embedding capacity between RGB channels. Document [21] proposes a hidden writing method based on Gaussian Markov Random Fields (GMRF), which designs a loss function by interaction between local elements. The commonality of these methods is that the embedding cost for each pixel in the image is determined by an artificially designed distortion function, and then the secret information is embedded in a manner that minimizes the sum of the embedding costs. Since the manual design of distortion functions has limitations, a recent group of researchers have used neural networks to automatically learn the distortion functions [22] [23] [24 ].

Although the existing method can realize the embedding and correct extraction of the watermark, certain defects still exist in the aspect of robustness, and corresponding watermark information is difficult to be correctly extracted for an image which is maliciously attacked or edited. Therefore, for copyright authentication and disclosure tracking, which require that watermark information must be completely and correctly extracted, the requirements still cannot be completely met.

[1] Kurak C., McHugh J. A Cautionary Note on Image Downgrading, In: Proceedings of the 8th IEEE Annual Computer Security Applications Conference, 1992, 153-159.

[2] Van Schyndel R. G., Tirkel A. Z., Osborne C. F. A Digital Watermark. In: Proceedings of the IEEE International Conference on Image Processing, Austin, TX, USA, 1994, 2(2):86 - 90.

[3] Bender W. R., Gruhl D., Morimoto N., Lu A. Techniques for data hiding. IBM Systems Journal, 1996, 35(3.4):313-336.

[4] Mielikainen J. LSB matching revisited. IEEE Signal Processing Letters, 2006, 13(5):285-287.

[5] Potdar V. M., Chang E. Grey level modification steganography for secret communication, In: 2nd IEEE International Conference on Industrial Informatics, 2004: 223-228.

[6] Muhammad, K., Sajjad M., Mehmood I., Rho S., Baik S. W. A novel magic LSB substitution method (M-LSB-SM) using multi-level encryption and achromatic component of an image. Multimedia Tools and Applications, 2016, 75(22): 14867-14893.

[7] Wu D. C., Tsai W. H. A steganographic method for images by pixel-value differencing. Pattern Recognition Letters, 2003, 24(9-10): 1613-1626.

[8] Zhou W., Zhang W., Yu N. A new rule for cost reassignment in adaptive steganography. IEEE Transactions on Information Forensics and Security, 2017, 12(11): 2654-2667.

[9] Yu Y. H., Chang C. C., Hu Y. C. Hiding secret data in images via predictive coding. Pattern Recognition, 2005, 38(5): 691-705.

[10] Li X., Li B., Yang B., Zeng T. General framework to histogram-shifting-based reversible data hiding. IEEE Transactions on Image Processing, 2013, 22(6): 2181-2191.

[11] Saidi M., Hermassi H., Rhouma R., Belghith S. A new adaptive image steganography scheme based on DCT and chaotic map. Multimedia Tools and Applications, 2017, 76(11): 13493-13510.

[12] Kumar V., Kumar D. A modified DWT-based image steganography technique. Multimedia Tools and Applications, 2018, 77(11): 13279-13308.

[13] Chen W. Y. Color image steganography scheme using DFT, SPIHT codec, and modified differential phase-shift keying techniques. Applied Mathematics & Computation, 2008, 196(1):40-54.

[14] Filler T., Judas, J., Fridrich, J. Minimizing additive distortion in steganography using syndrome-trellis codes. IEEE Transactions on Information Forensics and Security, 2011, 6(3): 920-935.

[15] Pevný T., Filler T., Bas P. Using high-dimensional image models to perform highly undetectable steganography. In: International Workshop on Information Hiding. Springer, Berlin, Heidelberg, 2010: 161-177.

[16] Holub V., Fridrich, J. Designing steganographic distortion using directional filters. In: 2012 IEEE International workshop on information forensics and security (WIFS), 2012: 234-239.

[17] Fridrich J., Kodovsky J. Rich models for steganalysis of digital images. IEEE Transactions on Information Forensics and Security, 2012, 7(3): 868-882.

[18] Holub V., Fridrich, J. Digital image steganography using universal distortion. In: Proceedings of the first ACM workshop on Information hiding and multimedia security, ACM, 2013: 59-68.

[19] Li B., Wang M., Huang J., Li X. A new cost function for spatial image steganography. In: 2014 IEEE International Conference on Image Processing (ICIP) IEEE, 2014: 4206-4210.

[20] Liao X., Yu Y., Li B., Li Z., Qin Z. A new payload partition strategy in color image steganography. IEEE Transactions on Circuits and Systems for Video Technology, 2019, 30(3): 685-696.

[21] Su W., Ni J., Hu X., Fridrich J. Image Steganography with Symmetric Embedding using Gaussian Markov Random Field Model. IEEE Transactions on Circuits and Systems for Video Technology, 2020.

[22] Tang W., Tan S., Li B., Huang J. Automatic steganographic distortion learning using a generative adversarial network. IEEE Signal Processing Letters, 2017, 24(10): 1547-1551.

[23] Yang J., Ruan D., Huang J., Kang X., Shi Y.-Q. An embedding cost learning framework using GAN. IEEE Transactions on Information Forensics and Security, 2019, 15: 839-851.

[24] W. Tang, B. Li, M. Barni, J. Li, and J. Huang, "An Automatic Cost Learning Framework for Image Steganography Using Deep Reinforcement Learning," IEEE Transactions on Information Forensics and Security, 2020, 16: 952-967.

Disclosure of Invention

The invention discloses an intelligent terminal image divulgence tracking and copyright authentication method based on digital watermarking, aiming at copyright authentication and divulgence tracking for completely and correctly extracting watermark information.

The technical scheme of the invention is as follows: an intelligent terminal image divulgence tracking and copyright authentication method based on digital watermarking comprises watermark embedding and watermark extracting;

the watermark embedding steps are as follows:

step A1, selecting an embedded area: selecting a plurality of texture complex areas by using a sliding window with a specific size, and directly embedding a mark code in the areas;

step A2, generating image watermark information: generating a unique identity identification watermark according to the intelligent terminal and the owner information thereof;

step A3, watermark embedding: embedding watermark information of the image in the selected area by a DCT method;

the watermark extraction steps are as follows:

b1, selecting an image to be detected;

b2, watermark extraction: determining the position of the watermark by detecting the mark code, and trying to extract the watermark in the selected area by a DCT (discrete cosine transformation) method;

step B3, image restoration: if the watermark cannot be extracted, detecting whether the image is attacked or not, and performing targeted restoration on the image according to the attack type;

step B4. executes step B2 and step B3 in a loop until watermark information is successfully extracted or until all attack types are tried to be repaired;

step B5. outputs a watermark image: and converting the extracted watermark binary sequence into a binary watermark image.

Preferably, step a1 is specifically: calculating the texture complexity of each region by using a sliding window with a specific size, simultaneously considering the spatial distribution of the regions, selecting 5 regions as embedded regions, and embedding mark codes at four corners of the embedded regions, wherein the mark codes are fixed pixel difference combinations.

Preferably, step a2 is specifically:

a2.1, acquiring a unique identifier of the intelligent terminal;

step A2.2, acquiring storage time of a photographed image or an image;

and step A2.3, converting the character string information obtained in the step 2.1 and the step 2.2 into image watermark information.

Preferably, step a3 is specifically:

a3.1, partitioning a Y channel of a carrier image;

step A3.2, reading each carrier image block and determining the watermark value to be embedded in each block;

step A3.3. if the watermark value needing to be embedded is 1, the numerical value of the embedded point is larger than the numerical value of the reference point, and if the watermark value needing to be embedded is 0, the numerical value of the embedded point is smaller than the numerical value of the reference point;

step A3.4. if the condition of the step 3.3 is not met, modifying the numerical value of the embedded point by adopting a mode of gradually adding or subtracting 1, and repeating the step 3.3 until the watermark is completely embedded;

step A3.5, carrying out IDCT transformation on the image block of the carrier, and combining the images of the three channels to obtain an image containing the watermark;

and A3.6, storing the image containing the watermark or replacing the original carrier image.

Preferably, step B3 is specifically:

b3.1, forming a list of attack types and testing one by one;

b3.2, performing targeted repair on different types of attacks;

b3.3, continuously calculating the similarity of the marker codes of the Y channels in the repairing process, stopping when the similarity of the marker codes is greater than a threshold value, and otherwise, finishing all repairing possibilities;

step B3.4. if the similarity of the marker code is still smaller than the threshold value after the execution of the step B3.3 is finished, other types of attacks are selected for repairing, the step B3.2 and the step B3.3 are executed in a circulating way, and the operation is stopped when the similarity of the marker code is larger than the threshold value; if the similarity of the marker codes is still smaller than the threshold value after the restoration by all the methods, the image to be detected does not contain the watermark information.

Compared with the traditional method, the method has the following advantages: the invention can realize copyright authentication and leakage tracing of the image under a complex network environment by repeatedly embedding the watermark information with small capacity and the locator in the texture complex area of the image for many times; JPEG compression attack, cutting attack and noise attack can be effectively resisted; under JPEG compression with quantization factor of more than 70%, under cutting attack of 1/4, or under noise attack with signal-to-noise ratio of more than 50, the watermark extraction accuracy rate reaches more than 80%; when embedding, the system synchronously embeds the digital watermark information without delay and pause in sense.

Drawings

Fig. 1 is a flowchart of watermark generation according to an embodiment of the present invention.

Fig. 2 is a flowchart of watermark embedding in the embodiment of the present invention.

FIG. 3 is a flowchart illustrating image restoration according to an embodiment of the present invention.

Fig. 4 is a flowchart of watermark extraction according to an embodiment of the present invention.

FIG. 5 is a basic functional flow of an embedding part in an embodiment of the present invention.

Fig. 6 is a basic function flow of the extractor in the embodiment of the present invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.

The invention realizes the complete extraction of the watermark information under the complex network environment by repeatedly embedding the watermark information with small capacity and the locator in the texture complex area of the image for many times. The main working principle is that according to the unique attribute information of the intelligent terminal or the personal identity information of the owner of the intelligent terminal, the digital watermark for identifying the identity of the owner of the digital terminal is generated by combining the photographing time, and when the digital terminal stores a new image, the watermark is synchronously embedded into the stored image so as to identify the identity information of the owner of the intelligent terminal. When a secret leakage tracking task is realized, the method is mainly applied to intelligent terminals of related personnel of a secret-related unit; when the copyright protection task is realized, the method can be applied to intelligent terminals used by any needed units and individuals.

The embedded segment of the embodiment is pre-installed in a public mobile phone of a security unit in the form of system plug-in or system software to realize divulgence tracking. If the image copyright authentication is needed, the method can be installed by itself. In the embodiment, a plurality of texture complex areas are selected by using a fixed-size sliding window (the window size depends on the watermark information amount), a mark code is directly embedded in a spatial domain, then a unique identity mark watermark is generated according to the intelligent terminal and the owner information of the intelligent terminal, and then repeated embedding is carried out in the selected areas by a DCT (discrete cosine transformation) method.

The following specifically introduces the Android mobile phone as the platform of the intelligent terminal in this embodiment. The intelligent terminal of the embodiment mainly comprises the following modules.

An embedded region selection module: in order to enable watermark information to be correctly extracted in a complex network environment, information is mainly embedded in a texture complex area of an image. Calculating the texture complexity of each region through a sliding window with a specific size, simultaneously considering the spatial distribution of the regions, selecting 5 regions as embedded regions, and embedding markers in four corners of the embedded regions, wherein the markers are fixed pixel difference combinations, for example, the difference of adjacent pixels is a fixed value.

Watermark information generation module: the watermark information mainly comprises a unique identifier of the intelligent terminal and image storage time, and takes an Android system mobile phone as an example, and the specific flow is shown in fig. 1. When the watermark is generated, the IMEI number or Android ID of the mobile phone and the time for photographing or image storage are mainly used. It should be noted that, for a system with an Android version of 10.0 or more, the third-party developer cannot obtain the IMEI number of the mobile phone, so the Android ID is used. On the premise of not flashing the mobile phone, the Android ID is unique, so that identity information of a mobile phone owner can be confirmed. In addition, the time information can also verify the divulgence behavior and the copyright protection behavior. The process of generating the watermark is actually to convert the mobile phone identification code and the time information into an image without background, so that the purpose is that the image has better redundancy and can still identify the information after being attacked.

A watermark embedding module: the module is mainly used for embedding watermark information into a region marked by a marker, embedding watermark information of an image into a DCT (discrete cosine transformation) domain of the image, wherein the embedded watermark image is a binary image, namely, not 1, namely 0. The specific embedding flow is shown in fig. 2. The system mainly embeds watermark information into DCT coefficients of an image Y channel. When embedding, firstly, the image of the marked area is partitioned, and then the watermark value needing to be embedded in each block is determined. When the watermark value needing to be embedded is 1, the numerical value of the embedded point is required to be ensured to be larger than the numerical value of the reference point, if the watermark value needing to be embedded is 0, the numerical value of the embedded point is required to be ensured to be smaller than the numerical value of the reference point, and if the watermark value needing to be embedded is not satisfied, the watermark value is modified by gradually adding or subtracting 1; when all watermark information is completely embedded, the image blocks are transformed through IDCT, and then the images of the three channels are combined to obtain an image containing the watermark; it should be noted that if the image is taken by the software, the image is directly saved; if the image is generated or stored by other software, the image generated by the software and containing the watermark replaces the original image. Watermark information contained in the image can be extracted through management terminal extraction software, so as to achieve the purposes of copyright protection and divulgence tracking.

An image restoration module: when the image is transmitted through a complex network, some attacks may be received, and therefore, the image may need to be repaired first in the process of extracting the watermark. The embodiment mainly detects assumed attacks such as rotation, scaling, clipping, noise and the like, and repairs the image. The specific flow is shown in fig. 3, and the attack types are firstly formed into a list and tested one by one. When the rotation attack is detected, the image is rotated for a circle by taking 5 degrees as a step length, the similarity of the marker code is continuously calculated in the process, and the rotation attack is stopped when the similarity is greater than a threshold value, otherwise, all rotation possibilities are executed; and other attack repairing methods are similar, and if the similarity of all the methods is still smaller than the threshold value after repairing, the image to be detected is considered not to contain watermark information.

A watermark extraction module: the present embodiment first determines the location of the watermark by detecting the marker code. The specific watermark extraction process is shown in fig. 4, after the area where the watermark is located is determined by the mark code, firstly extracting the Y channel of the image area to be detected and blocking, then performing DCT transformation on the blocked image, and then comparing the DCT coefficients, wherein if the DCT coefficient is larger than the reference value, the watermark information is 1, otherwise, the watermark information is 0; if the marking code threshold value is smaller than a set value when the watermark area is determined, two conditions exist at the moment, namely, the image is attacked (zooming, noise, rotating, cutting and the like), and the marking code can be correctly extracted when the similarity of the marking code is larger than the threshold value through an image restoration method; secondly, the image does not contain the watermark, and all attack repairs need to be tried. And after all the watermark information is extracted, converting the watermark information into a binary watermark image so as to identify the identity of the owner.

In this embodiment, the basic functions and processes of the embedding part and the extracting part are as follows.

At the embedding end, as shown in fig. 5, a user identification watermark carrying an identification code is generated mainly according to an IMEI number (Android version is greater than or equal to 10.0) or an Android ID (Android version is less than 10.0) of a mobile phone and image storage time, then the watermark is embedded before image storage (for a picture shot by using APP) or the watermark is embedded in an image stored by other application software, and then an original image is replaced, so as to provide evidence for divulgence tracking or copyright protection. According to the method, a corresponding plug-in program or APP program, such as a digital forensic camera, can be developed. Taking a digital evidence obtaining camera as an example, when shooting, the system synchronously embeds digital watermark information, and the shooting is not delayed and stopped in sense. When the divulgence tracking task is realized, the method is mainly applied to official mobile phones of related personnel of a secret-related unit; when the copyright protection task is realized, the method can be applied to Android mobile phones used by any needed units and individuals.

At the extraction end, as shown in fig. 6, it mainly performs divulgence tracing or judges the identity of the copyright owner by detecting whether the image to be detected contains digital watermark information representing its identity. The watermark information mainly comprises an IMEI number (the Android version is more than or equal to 10.0) or an Android ID (the Android version is less than 10.0) of the mobile phone and image storage time.

The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims (5)

1. An intelligent terminal image divulgence tracking and copyright authentication method based on digital watermarking is characterized by comprising two parts of content, namely watermark embedding and watermark extracting;

the watermark embedding steps are as follows:

step A1, selecting an embedded area: selecting a plurality of texture complex areas by using a sliding window with a specific size, and directly embedding a mark code in the areas;

step A2, generating image watermark information: generating a unique identity identification watermark according to the intelligent terminal and the owner information thereof;

step A3, watermark embedding: embedding watermark information of the image in the selected area by a DCT method;

the watermark extraction steps are as follows:

b1, selecting an image to be detected;

b2, watermark extraction: determining the position of the watermark by detecting the mark code, and trying to extract the watermark in the selected area by a DCT (discrete cosine transformation) method;

step B3, image restoration: if the watermark cannot be extracted, detecting whether the image is attacked or not, and performing targeted restoration on the image according to the attack type;

step B4. executes step B2 and step B3 in a loop until watermark information is successfully extracted or until all attack types are tried to be repaired;

step B5. outputs a watermark image: and converting the extracted watermark binary sequence into a binary watermark image.

2. The intelligent terminal image divulgence tracking and copyright authentication method based on digital watermarking as claimed in claim 1, wherein the step a1 specifically comprises: calculating the texture complexity of each region by using a sliding window with a specific size, simultaneously considering the spatial distribution of the regions, selecting 5 regions as embedded regions, and embedding mark codes at four corners of the embedded regions, wherein the mark codes are fixed pixel difference combinations.

3. The intelligent terminal image divulgence tracking and copyright authentication method based on digital watermarking as claimed in claim 1, wherein the step a2 specifically comprises:

a2.1, acquiring a unique identifier of the intelligent terminal;

step A2.2, acquiring storage time of a photographed image or an image;

and step A2.3, converting the character string information obtained in the step 2.1 and the step 2.2 into image watermark information.

4. The intelligent terminal image divulgence tracking and copyright authentication method based on digital watermarking as claimed in claim 1, wherein the step a3 specifically comprises:

a3.1, partitioning a Y channel of a carrier image;

step A3.2, reading each carrier image block and determining the watermark value to be embedded in each block;

step A3.3. if the watermark value needing to be embedded is 1, the numerical value of the embedded point is larger than the numerical value of the reference point, and if the watermark value needing to be embedded is 0, the numerical value of the embedded point is smaller than the numerical value of the reference point;

step A3.4. if the condition of the step 3.3 is not met, modifying the numerical value of the embedded point by adopting a mode of gradually adding or subtracting 1, and repeating the step 3.3 until the watermark is completely embedded;

step A3.5, carrying out IDCT transformation on the image block of the carrier, and combining the images of the three channels to obtain an image containing the watermark;

and A3.6, storing the image containing the watermark or replacing the original carrier image.

5. The intelligent terminal image divulgence tracking and copyright authentication method based on digital watermarking as claimed in claim 1, wherein step B3 specifically comprises:

b3.1, forming a list of attack types and testing one by one;

b3.2, performing targeted repair on different types of attacks;

b3.3, continuously calculating the similarity of the marker codes of the Y channels in the repairing process, stopping when the similarity of the marker codes is greater than a threshold value, and otherwise, finishing all repairing possibilities;

step B3.4. if the similarity of the marker code is still smaller than the threshold value after the execution of the step B3.3 is finished, other types of attacks are selected for repairing, the step B3.2 and the step B3.3 are executed in a circulating way, and the operation is stopped when the similarity of the marker code is larger than the threshold value; if the similarity of the marker codes is still smaller than the threshold value after the restoration by all the methods, the image to be detected does not contain the watermark information.

Images