Robust steganography method and device for images with high quality factor and large size

Abstract

The application relates to the technical field of digital media processing, in particular to a robust steganography method and device for an image with a high quality factor and a large size, wherein the method comprises the following steps: decoding a carrier image to obtain first spatial domain pixel values of the carrier image; performing pre-scaling processing on a spatial domain image corresponding to the carrier image based on channel characteristics of a target network platform and obtaining discrete cosine transform (DCT) coefficients; embedding target secret information into the DCT coefficients to obtain initial carrier-cipher DCT coefficients, adjusting unstable coefficients of the initial carrier-cipher DCT coefficients, and obtaining final carrier-cipher DCT coefficients; transforming the final carrier-cipher DCT coefficients into a spatial domain to obtain second spatial domain pixel values, and modifying the first spatial domain pixel values according to the second spatial domain pixel values until an air domain image corresponding to the modified first spatial domain pixel values satisfies a first preset scaling condition; and compressing the air domain image corresponding to the modified first spatial domain pixel values to generate a final carrier-cipher image satisfying a preset size.

Description

Technical Field

[0001] This application relates to the field of digital media processing technology, and in particular to a robust steganography method and apparatus for images with high quality factor and large size. Background Technology

[0002] In related technologies, early image steganography algorithms assumed the communication channel was lossless, neglecting potential lossy processing such as scaling and compression within the channel. They focused on adaptively selecting the information embedding location based on the image content to improve the statistical detection resistance of the embedded image. Therefore, robust steganography techniques gradually gained attention from researchers in the steganography field to address the problem of covert communication in lossy channels. However, later research on robust image steganography algorithms often only considered single lossy processing such as scaling or recompression. Anti-scaling steganography algorithms typically embed and modify data in the spatial domain, usually assuming the carrier file format is lossless and lacking robustness to JPEG compression, thus severely limiting their application. Furthermore, anti-JPEG recompression schemes require adjusting the file size below the communication channel threshold to avoid triggering scaling operations, which is behaviorally unsafe.

[0003] However, the image steganography algorithms in related technologies do not take into account the lossy processing present in the channel. The generated secret images often cannot correctly extract the secret messages, or they only consider single lossy processing such as scaling or recompression. The monitoring party can easily observe the quality factor and size of the user-uploaded image, and thus distinguish communication behavior from public behavior. The scope of use is relatively limited, and it may lead to a series of behavioral security problems, which urgently need to be solved. Summary of the Invention

[0004] This application provides a robust steganography method, apparatus, electronic device, and storage medium for images with high quality factor and large size. It addresses the problems in related technologies where image steganography algorithms do not consider lossy processing present in the channel, resulting in generated images that often fail to correctly extract secret messages, or only consider single lossy processing such as scaling or recompression. This allows monitoring parties to easily observe the quality factor and size of user-uploaded images, thereby distinguishing communication behavior from general public behavior. This limited applicability may lead to a series of behavioral security issues.

[0005] The first aspect of this application provides a robust steganography method for images with high quality factors and large sizes, comprising the following steps: decoding a carrier image to obtain a first spatial pixel value of the carrier image; pre-scaling the spatial image corresponding to the carrier image based on the channel characteristics of the target network platform to obtain a pre-scaled spatial image; and obtaining the discrete cosine transform (DCT) coefficients of the pre-scaled spatial image; embedding target secret information into the DCT coefficients to obtain initial secret DCT coefficients; adjusting the instability coefficients in the initial secret DCT coefficients to obtain final secret DCT coefficients; transforming the final secret DCT coefficients to the spatial domain to obtain corresponding second spatial pixel values; modifying the first spatial pixel value according to the second spatial pixel value until the spatial image corresponding to the modified first spatial pixel value satisfies a first preset scaling condition; and compressing the spatial image corresponding to the modified first spatial pixel value according to the quality factor of the carrier image to generate a final secret image that meets a preset size.

[0006] Optionally, in one embodiment of this application, embedding the target secret information into the DCT coefficients to obtain initial DCT coefficients with hidden information includes: performing steganalytic quantization on the DCT coefficients to generate a steganalytic carrier; embedding the target secret information into the steganalytic carrier to obtain a hidden information sequence; and modifying the DCT coefficients according to the carrier sequence of the steganalytic carrier and the hidden information sequence to obtain the initial DCT coefficients with hidden information.

[0007] Optionally, in one embodiment of this application, before embedding the target secret information into the DCT coefficients, the method further includes: obtaining quantization step size information of the coefficient block composed of the DCT coefficients; and determining the target position where the target secret information is embedded into the DCT coefficients based on the quantization step size information.

[0008] Optionally, in one embodiment of this application, adjusting the instability coefficients in the initial DCT coefficients to obtain the final DCT coefficients includes: performing inverse discrete cosine transform, spatial rounding, and truncation on the initial DCT coefficients to obtain corresponding first simulated spatial pixel values; modifying the first simulated spatial pixel values ​​based on the first simulated spatial pixel values ​​to obtain second simulated spatial pixel values; obtaining a pre-transmission intermediate image that meets a second preset scaling condition based on the second simulated spatial pixel values; and performing simulated channel processing on the pre-transmission intermediate image to adjust the instability coefficients in the initial DCT coefficients to obtain the final DCT coefficients.

[0009] Optionally, in one embodiment of this application, modifying the first spatial pixel value according to the second spatial pixel value includes: calculating the modification cost of each pixel in the spatial image using a target distortion function; calculating the interpolation block corresponding to each element in the second spatial pixel value on the spatial image to obtain the weight value of each element in the interpolation block; sorting the weight values ​​according to the modification cost to generate a modification sequence; and modifying the first spatial pixel value according to the second spatial pixel value based on the modification sequence.

[0010] A second aspect of this application provides a robust steganography apparatus for images with high quality factors and large sizes, comprising: a first processing module for decoding a carrier image to obtain a first spatial pixel value of the carrier image, and pre-scaling the spatial image corresponding to the carrier image based on the channel characteristics of the target network platform to obtain a pre-scaled spatial image, and acquiring the discrete cosine transform (DCT) coefficients of the pre-scaled spatial image; a second processing module for embedding target secret information into the DCT coefficients to obtain initial secret DCT coefficients, adjusting the instability coefficients in the initial secret DCT coefficients to obtain final secret DCT coefficients; and a steganography module for transforming the final secret DCT coefficients to the spatial domain to obtain corresponding second spatial pixel values, modifying the first spatial pixel values ​​according to the second spatial pixel values ​​until the spatial image corresponding to the modified first spatial pixel values ​​meets a first preset scaling condition, and compressing the spatial image corresponding to the modified first spatial pixel values ​​according to the quality factor of the carrier image to generate a final secret image that meets a preset size.

[0011] Optionally, in one embodiment of this application, the second processing module includes: a first processing unit, configured to perform steganalytic quantization processing on the DCT coefficients to generate a steganalytic carrier; an embedding unit, configured to embed the target secret information into the steganalytic carrier to obtain a steganalytic sequence; and a first modification unit, configured to modify the DCT coefficients according to the carrier sequence of the steganalytic carrier and the steganalytic sequence to obtain the initial steganalytic DCT coefficients.

[0012] Optionally, in one embodiment of this application, it further includes: an acquisition module, which acquires quantization step size information of a coefficient block composed of the DCT coefficients before embedding the target secret information into the DCT coefficients; and a determination module, which determines the target position where the target secret information is embedded into the DCT coefficients based on the quantization step size information.

[0013] Optionally, in one embodiment of this application, the second processing module includes: a second processing unit, configured to perform inverse discrete cosine transform processing, spatial rounding processing, and truncation processing on the initial DCT coefficients to obtain corresponding first simulated spatial pixel values; a second modification unit, configured to modify the first simulated spatial pixel values ​​based on the first simulated spatial pixel values ​​to obtain second simulated spatial pixel values; a generation unit, configured to obtain a pre-transmission intermediate image that satisfies a second preset scaling condition based on the second simulated spatial pixel values; and an adjustment unit, configured to perform simulated channel processing on the pre-transmission intermediate image to adjust the instability coefficients in the initial DCT coefficients to obtain final DCT coefficients.

[0014] Optionally, in one embodiment of this application, the steganography module includes: a first calculation unit, configured to calculate the modification cost of each pixel in the spatial domain image using a target distortion function; a second calculation unit, configured to calculate the interpolation block corresponding to each element in the second spatial domain pixel value on the spatial domain image, and obtain the weight value of each element in the interpolation block; a sorting unit, configured to sort the weight values ​​according to the modification cost to generate a modification sequence; and a third modification unit, configured to modify the first spatial domain pixel value according to the second spatial domain pixel value based on the modification sequence.

[0015] A third aspect of this application provides an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the robust steganography method for images with high quality factor and large size as described in the above embodiments.

[0016] A fourth aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the robust steganography method described above for images with high quality factors and large sizes.

[0017] A fifth aspect of this application provides a computer program product, including a computer program that, when executed, is used to implement the robust steganography method described above for images with high quality factor and large size.

[0018] This application embodiment uses a pre-scaled image of the original carrier image as a medium, embeds secret messages based on a dithering modulation algorithm combined with an adaptive steganography framework, adjusts unstable DCT coefficients through simulated channel processing, and finally generates a large-size, high-quality steganography image through inverse interpolation. Thus, the generated steganography image has the same quality factor and size as the original carrier image, enabling covert communication and resisting scaling and JPEG recompression attacks in lossy social media platforms, correctly extracting the secret information from the steganography image. This solves the problems of related image steganography algorithms that do not consider lossy processing in the channel, often resulting in steganography images that cannot correctly extract secret messages, or those that only consider single lossy processing such as scaling or recompression. Monitoring parties can easily observe the quality factor and size of user-uploaded images, thus distinguishing communication behavior from general public behavior, limiting the scope of application and potentially leading to a series of behavioral security issues.

[0019] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0020] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0021] Figure 1 This is a schematic diagram of a social platform channel processing procedure according to an embodiment of this application;

[0022] Figure 2 This is a schematic diagram of the framework of a robust steganography method for images with high quality factor and large size according to an embodiment of this application.

[0023] Figure 3 This is a flowchart illustrating a robust steganography method for images with high quality factor and large size, according to an embodiment of this application.

[0024] Figure 4 This is a schematic diagram of fine jitter modulation according to an embodiment of this application;

[0025] Figure 5 This is a schematic diagram of the structure of a robust steganography device for images with high quality factor and large size provided according to an embodiment of this application;

[0026] Figure 6 This is a schematic diagram of the structure of an electronic device provided according to an embodiment of this application.

[0027] Figure label:

[0028] 10- Robust steganography device for high-quality factor and large-size images: 100- First processing module, 200- Second processing module and 300- Steganography module; 601- Memory, 602- Processor and 603- Communication interface. Detailed Implementation

[0029] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0030] The following describes a robust steganography method and apparatus for images with high quality factor and large size, based on embodiments of this application, with reference to the accompanying drawings. Regarding the image steganography algorithms mentioned in the background art, which do not consider lossy processing present in the channel, the generated steganographic images often fail to correctly extract the secret message, or only consider single lossy processing such as scaling or recompression. Monitors can easily observe the quality factor and size of user-uploaded images, thus distinguishing communication behavior from general public behavior. This limitation in application may lead to a series of behavioral security problems. This application provides a robust steganography method for images with high quality factor and large size. In this method, a pre-scaled image of the original carrier image is used as a medium. A dithering modulation algorithm combined with an adaptive steganography framework is used to embed the secret message. The unstable DCT coefficients are adjusted through simulated channel processing. Finally, inverse interpolation is used to generate a large-size, high-quality steganographic image. Thus, the generated steganographic image has the same quality factor and size as the original carrier image, enabling covert communication and resisting scaling and JPEG recompression attacks in lossy social media platform channels, correctly extracting the secret information from the steganographic image. This solves the problem that image steganography algorithms in related technologies do not consider lossy processing in the channel, and the generated secret images often cannot correctly extract secret messages, or only consider single lossy processing such as scaling or recompression. The monitoring party can easily observe the quality factor and size of the user-uploaded image, and thus distinguish communication behavior from public behavior. This has a relatively limited scope of application and may lead to a series of behavioral security problems.

[0031] Before explaining the robust steganography method for images with high quality factor and large size in the embodiments of this application, the system framework of the robust steganography method for images with high quality factor and large size in the embodiments of this application will be explained first.

[0032] Steganography is a technique that conceals communication activities beneath normal behavior, thus achieving covert communication. Steganography embeds secret information into a carrier sample in an imperceptible manner, creating a coded sample, and then transmits the coded sample to achieve covert transmission of the secret information. Because images themselves contain rich information and the human eye is insensitive to subtle changes, coupled with the widespread popularity of image sharing on social networking platforms, images have become an excellent carrier for covert communication on these platforms. With the increasing number of users on social networking platforms, massive amounts of data are uploaded to these platforms every day. Due to considerations of network communication speed and data storage capacity, image files often undergo lossy operations such as scaling and compression during the upload process to social networking platforms to reduce data volume. These lossy processing methods usually do not affect the content information represented by the original data, but they can cause irreversible damage to the image data. This means that the secret information in the coded image may be destroyed and cannot be correctly extracted.

[0033] In real life, people often share photos taken with their mobile phones on social networking platforms. These images are usually JPEG images with large file sizes and high quality factors. After being uploaded to social networking platforms, these images are typically scaled down to a smaller file size and recompressed into JPEG with a lower quality factor, such as... Figure 1 The diagram shown is a schematic diagram of the social platform channel processing process according to an embodiment of this application.

[0034] Steganography, in essence, aims to make communication indistinguishable from general public behavior. However, ordinary users upload high-quality, large-format photos taken with their mobile phones, while steganographic users upload low-quality or small-format images. Therefore, surveillance can easily observe the quality and size of uploaded images. Comparatively, steganography on large, high-quality images and uploading them is a more secure communication method. Therefore, it is necessary to design robust steganography algorithms resistant to scaling and JPEG recompression for high-quality, large-format images.

[0035] To address the issue that social platforms scale and JPEG recompress images with large size and high quality factor, and that related robust steganography algorithms cannot resist both attacks simultaneously, this application proposes a robust steganography algorithm resistant to scaling and JPEG recompression for images with high quality factor and large size. This aims to solve the problem of covert communication when images with high quality factor and large size are used on social platforms with scaling and JPEG recompression.

[0036] Figure 2 This is a schematic diagram illustrating the framework of a robust steganography method for images with high quality factor and large size, according to one embodiment of this application. Figure 2As shown, the framework mainly comprises four modules: preprocessing, message embedding based on dithering modulation, coefficient adjustment based on analog channel processing, and secret image generation. For ease of explanation, uppercase letters are primarily used to represent matrices and vectors, while lowercase letters are used to represent the corresponding elements within the matrices or vectors. This application uses a pre-scaled image as a medium, embeds secret messages based on a dithering modulation algorithm combined with an adaptive steganography framework, adjusts unstable DCT coefficients through analog channel processing, and finally generates a large-size, high-quality secret image through an inverse interpolation algorithm. This achieves a robust steganography algorithm resistant to scaling and JPEG recompression for images with high-quality factors and large sizes.

[0037] Specifically, Figure 3 A flowchart illustrating a robust steganography method for images with high quality factor and large size, provided as an embodiment of this application.

[0038] like Figure 3 As shown, this robust steganography method for images with high quality factors and large sizes includes the following steps:

[0039] In step S301, the carrier image is decoded to obtain the first spatial pixel value of the carrier image, and based on the channel characteristics of the target network platform, the spatial image corresponding to the carrier image is pre-scaled to obtain the pre-scaled spatial image, and the discrete cosine transform (DCT) coefficients of the pre-scaled spatial image are obtained.

[0040] It is understandable that the carrier image here refers to the original image that will carry secret information. The spatial image here can be understood as a spatial distribution representation of the carrier image. Each pixel in the spatial image corresponds to a point in the carrier image in space, and the spatial pixel value reflects the brightness or color information at that point. In the spatial domain, image processing usually directly affects these pixel values.

[0041] In some embodiments, in order to facilitate image manipulation, this application can perform decoding operations on the carrier image to obtain the first spatial pixel value, and can process the spatial image of the carrier image with the same scaling algorithm and scaling factor according to the channel characteristics of the target network platform to obtain a pre-scaled image, and obtain the DCT (Discrete Cosine Transform) coefficient values ​​corresponding to the pre-scaled image.

[0042] Here, the target network platform can be understood as the network platform selected by the communication user when transmitting images, such as some social networking platforms. Channel characteristics can be understood as the influences and changes experienced by the signal during transmission through the target network platform's channel during network communication.

[0043] Furthermore, the carrier image decoding operations in the embodiments of this application include, but are not limited to, entropy decoding processing, inverse quantization processing, spatial rounding processing, and truncation processing.

[0044] Step S302: Embed the target secret information into the DCT coefficients to obtain the initial secret DCT coefficients, and adjust the instability coefficients in the initial secret DCT coefficients to obtain the final secret DCT coefficients.

[0045] In other embodiments, in addition to preprocessing the carrier image, this application also requires processing the target secret information loaded into the carrier image. Specifically, the target secret information can be embedded into the obtained DCT coefficients to obtain the initial carrier DCT coefficients.

[0046] Furthermore, considering that there may be some unstable coefficients in the initial encrypted DCT coefficients, in order to ensure the security of the final transmitted image, the embodiments of this application can also adjust the unstable coefficients in the initial encrypted image, thereby obtaining the final encrypted DCT coefficients required by the embodiments of this application.

[0047] For example, the DCT coefficients obtained in step S301 can be denoted as follows: Next, this application can use a dither-based message embedding module to embed the target secret message into the DCT coefficients of the prescaled image. The initial density DCT coefficients were obtained. Finally, the coefficient adjustment module based on analog channel processing is used to adjust the... The instability coefficients in the data are further adjusted to obtain the final DCT coefficients required for the embodiments of this application.

[0048] The process will now be explained further.

[0049] Optionally, in one embodiment of this application, embedding the target secret information into DCT coefficients to obtain initial secret-carrying DCT coefficients includes: performing steganalytic quantization on the DCT coefficients to generate a steganalytic carrier; embedding the target secret information into the steganalytic carrier to obtain a secret-carrying sequence; and modifying the DCT coefficients according to the carrier sequence and the secret-carrying sequence of the steganalytic carrier to obtain the initial secret-carrying DCT coefficients.

[0050] Based on the descriptions of other embodiments, it will be understood that this application can embed target secret information into DCT coefficients, thereby obtaining initial secret-carrying DCT coefficients. For example, a dither-modulated message embedding module can be used to embed the target secret message into the DCT coefficients of a pre-scaled image. The initial density DCT coefficients were obtained. .

[0051] The message embedding module based on jitter modulation in this application embodiment mainly includes, but is not limited to, three steps: steg quantization, STC encoding, and jitter modulation.

[0052] Specifically, in this embodiment, the DCT coefficients can first be stegatically quantized to generate a stegtext carrier, and then the target secret information can be embedded into the stegtext carrier to obtain a secret sequence. Finally, the DCT coefficients can be modified based on the carrier sequence and the secret sequence of the stegtext carrier to obtain the initial secret DCT coefficients.

[0053] For example, since the loss caused by quantization in JPEG recompression is strongly correlated with the quantization step size, the quantization table used in JPEG recompression can be used with the target network platform's channel. The quantization step size is adaptively quantized for the coefficients at corresponding positions in the 8x8 DCT block. The resulting robust steganalytic vector is then used... X This process can be represented as follows:

[0054] (1)

[0055] Based on the JPEG image scaling and recompression process, it can be known that the steganographic carrier is equal to the QDCT coefficients of the image obtained after channel scaling and recompression of the original carrier image.

[0056] Obtain the steganographic carrier Subsequently, embodiments of this application can use the distortion cost function to calculate the stegtext carrier. Cost of modifying elements Then, use ternary STC embedding to embed the secret message. Adaptive embedding into robust steganographic carriers In the process, the encrypted sequence is obtained. The process can be represented as follows:

[0057] (2)

[0058] Among them, the use of the STC framework in robust steganography carriers When performing adaptive embedding, The modification range for each element can be set to +1, -1, or 0, etc.

[0059] Finally, based on the vector sequence To the encrypted sequence The changes affect the DCT coefficients of the prescaled spatial domain image of the carrier image. Make modifications, and denote the modified DCT coefficients as the initial DCT coefficients. The goal of the modification is to make Zhongyu The corresponding element, after steganographic quantization, can be equal to That is, it satisfies the constraints shown in the formula:

[0060] (3)

[0061] Furthermore, to ensure the security and imperceptibility of the encrypted image, embodiments of this application may also employ a fine dithering modulation scheme. elements in Revise. Figure 4 This is a schematic diagram of fine jitter modulation according to an embodiment of this application. Figure 4 As shown, embodiments of this application can be based on the distance from the boundary. The length is The interval is divided into , Two sub-intervals, the one closer to the boundary and farther from the midpoint is... An interval, also known as a non-robust interval; the interval that is farther from the boundary and closer to the midpoint is... The interval, also known as the robust interval. In jitter modulation, the embodiments of this application may not directly use... Instead of modifying to the midpoint of the interval, modify it to the robust interval. Mid-range The nearest point. That is, modifications only affect the nearest point. Jitter modulation is applied to positions where a "+1" or "-1" change has occurred, while positions that remain unchanged after STC encoding are left unmodified. It's important to note that even... Even after channel processing, the elements in the interval may not necessarily satisfy the constraints shown in formula (3), and the robustness needs to be further enhanced through subsequent coefficient adjustments.

[0062] It should be noted that, in the embodiments of this application, before performing STC encoding, it is necessary to calculate... The modification cost of each element is considered, and the position with the minimum total distortion is adaptively selected for information embedding. In this application embodiment, the distortion function of the present invention can be designed by combining the modification amplitude of jitter modulation with existing distortion functions. Specifically, existing distortion calculation methods such as JUNIWARD (a digital steganography technique) can be used first on a robust carrier. Calculate the initial distortion for each element's forward and backward modifications and denote them as follows: , Then calculate according to... Figure 4 The modulation method shown is modified forward and backward to obtain the modified DCT coefficients, which are denoted as follows: 、 The final distortion is defined as the product of the jitter modulation modification magnitude and the initial distortion, which can be expressed as follows:

[0063] (4)

[0064] The following section provides a further explanation of the process for selecting the embedding location in the embodiments of this application.

[0065] Optionally, in one embodiment of this application, before embedding the target secret information into the DCT coefficients, the method further includes: obtaining quantization step size information of the coefficient block composed of DCT coefficients; and determining the target position where the target secret information is embedded into the DCT coefficients based on the quantization step size information.

[0066] In actual implementation, before performing STC encoding, this application needs to calculate the modification cost of each element in the steganography carrier, and then adaptively select the position with the minimum total distortion for information embedding.

[0067] Specifically, in this embodiment, a coefficient block composed of DCT coefficients obtained from the spatial domain image of the original carrier image can be obtained first, and then the quantization step size information of the coefficient block can be obtained, thereby determining the target position of the target secret information embedded in the DCT coefficients based on the quantization step size information.

[0068] For example, this application can select the embedding domain based on the quantization step size corresponding to the elements in the 8*8 DCT coefficient block.

[0069] Given that the JPEG compression-resistant steganography method based on dithering modulation has a larger quantization step size corresponding to high-frequency and mid-frequency coefficients, and is more stable before and after JPEG compression, i.e., information embedding at mid-frequency and high-frequency of the image has better robustness, while information embedding at low frequency has poor robustness, therefore, for robustness considerations, this application can choose... The corresponding quantization step size Information is embedded in larger locations.

[0070] However, for safety reasons, robust vector sequences When embedding information, the modification increment is either "+1" or "-1", resulting in an average modification increment of the DCT coefficients that can reach [amount missing]. (Calculated based on modulation to the midpoint of the interval), excessive modification will lead to a decrease in image quality and a reduction in anti-detection performance. Therefore, the quantization step size should be adjusted accordingly. It's better to embed it in a smaller location.

[0071] Taking into account both robustness and security, this application sets two hyperparameters. and , The threshold can be understood as The corresponding quantization step size Greater than Only elements that meet the criteria will be selected as embedding coefficients. The number of elements selected as embedding fields in an 8x8 DCT block, i.e., in this embodiment, the last selected embedding field has a quantization step size greater than 1. The element with the smallest quantization step size Each element serves as an embedding coefficient.

[0072] It is important to note that hyperparameters and It can be obtained by professionals in this field from existing data or simulation experiments, depending on the specific circumstances and actual needs, and no specific restrictions are imposed here.

[0073] Optionally, in one embodiment of this application, adjusting the instability coefficients in the initial DCT coefficients to obtain the final DCT coefficients includes: performing inverse discrete cosine transform, spatial rounding, and truncation on the initial DCT coefficients to obtain corresponding first simulated spatial pixel values; modifying the first simulated spatial pixel values ​​based on the first simulated spatial pixel values ​​to obtain second simulated spatial pixel values; obtaining a pre-transmission intermediate image that satisfies a second preset scaling condition based on the second simulated spatial pixel values; and performing simulated channel processing on the pre-transmission intermediate image to adjust the instability coefficients in the initial DCT coefficients to obtain the final DCT coefficients.

[0074] In some embodiments, since there are unstable coefficients in the initial DCT coefficients, i.e., poor robustness, this application needs to adjust the unstable coefficients in the initial DCT coefficients to obtain the final DCT coefficients with higher robustness.

[0075] Specifically, this application can adjust the initial DCT coefficients using a coefficient adjustment module based on analog channel processing. First, the initial DCT coefficients are subjected to inverse discrete cosine transform, spatial rounding, and truncation to obtain the corresponding first analog spatial pixel values. Then, these first analog spatial pixel values ​​are modified to obtain second analog spatial pixel values. Next, a pre-transmission intermediate image satisfying a second preset scaling condition is obtained based on the second analog spatial pixel values. This pre-transmission intermediate image is then subjected to analog channel processing to adjust the unstable coefficients in the initial DCT coefficients, resulting in the final DCT coefficients.

[0076] The coefficient adjustment module based on analog channel processing in this embodiment mainly includes, but is not limited to, three steps: intermediate image generation, analog channel processing, and coefficient adjustment. The second preset scaling condition can be understood here as modifying the pixel values ​​of the spatial domain image of the original carrier image using the second analog spatial domain pixel values. The resulting analog spatial domain image, after scaling, is equal to the spatial pixel values ​​corresponding to the initial carrier density DCT coefficients after dithering modulation.

[0077] For example, the initial carrier density DCT coefficients after dithering modulation can be first... Inverse DCT transformation, spatial rounding, and truncation are performed to obtain the corresponding first analog spatial pixel value. Then, the pixel value modification module based on inverse interpolation and the first simulated spatial pixel value are used. Pixel values ​​of the spatial domain image of the original carrier image Modify the values ​​to obtain the second simulated spatial domain pixel values, so that the modified second simulated spatial domain values ​​correspond to the simulated spatial domain image. The scaled spatial pixel value is exactly equal to the initial carrier density DCT coefficient after dithering modulation. Corresponding spatial pixel value Then, the modified simulated spatial image is recompressed using the quality factor corresponding to the original carrier image to obtain the intermediate image to be sent. The obtained intermediate image is then scaled and recompressed using the same processing parameters as the target channel to obtain the image to be received.

[0078] After processing the generated intermediate image through an analog channel, the QDCT (Quaternion Discrete Cosine Transform) coefficients of the pre-received image within the embedding domain are compared. With the secret sequence If the elements in the table are equal, no modification is made if they are equal. For positions where elements are unequal, if the corresponding element... In non-robust areas In the middle, then will Adjusted to the robust zone Neutral The nearest location; if In the robust zone In the middle, then Move one position towards the midpoint of the interval. Because the coefficients in the robust region were adjusted with a step size of 1 during coefficient adjustment, the adjustment range is small, and the adjusted coefficients may still not meet the robustness requirements. Furthermore, the effect of spatial rounding changes after each coefficient modification; therefore, multiple rounds of coefficient adjustment are needed to ensure the robustness of the final DCT coefficients. Let the number of coefficient adjustment rounds be denoted as... The process can be represented as follows:

[0079] (5)

[0080] in, For the sign function.

[0081] Step S303: Transform the final DCT coefficients of the carrier image to the spatial domain to obtain the corresponding second spatial domain pixel values, and modify the first spatial domain pixel values ​​according to the second spatial domain pixel values ​​until the spatial image corresponding to the modified first spatial domain pixel values ​​meets the first preset scaling condition. Compress the spatial image corresponding to the modified first spatial domain pixel values ​​according to the quality factor of the carrier image to generate the final carrier image that meets the preset size.

[0082] As one possible approach, after obtaining the final DCT coefficients, this application can transform the final DCT coefficients to the spatial domain to obtain the corresponding second spatial pixel value, and modify the first spatial pixel value according to the second spatial pixel value until the spatial image corresponding to the modified first spatial pixel value meets the first preset scaling condition. The large-size spatial image corresponding to the modified first spatial pixel value is then compressed according to the quality factor of the carrier image to generate the final DCT image that meets the preset size.

[0083] Here, the first preset scaling condition can be understood as the spatial image corresponding to the modified first spatial pixel value being scaled to exactly equal the scaled spatial image of the encrypted image. The preset size can be understood as the size that meets the requirements for sending the encrypted image on the target network platform.

[0084] The encrypted image generated by this method has the same quality factor and size as the original carrier image and is resistant to scaling and JPEG recompression attacks in lossy social media platform channels. After the generated encrypted image is transmitted through the communication channel of a social media platform that includes interpolation scaling and JPEG recompression, the receiver will receive a scaled and recompressed encrypted image. Secret message extraction. In this case, the received image can first be entropy decoded to obtain the QDCT coefficients of the received image. Then, according to the same embedding field as when embedding the message, in The following steps involve selecting encryption coefficients, which together constitute the encryption carrier for extracting the secret message. Finally, the STC decoding algorithm is used to extract the secret message. .

[0085] Optionally, in one embodiment of this application, modifying the first spatial domain pixel value according to the second spatial domain pixel value includes: calculating the modification cost of each pixel in the spatial domain image using a target distortion function; calculating the interpolation block corresponding to each element in the second spatial domain pixel value in the spatial domain image to obtain the weight value of each element in the interpolation block; sorting the weight values ​​according to the modification cost to generate a modification sequence; and modifying the first spatial domain pixel value according to the second spatial domain pixel value based on the modification sequence.

[0086] Based on the descriptions of other embodiments, to ensure the robustness of the final carrier-density DCT coefficients, simulation processing was employed, utilizing a coefficient adjustment module based on simulated channel processing and a pixel value modification module based on inverse interpolation. The main function of the pixel value modification module based on inverse interpolation is to modify the first spatial domain pixel value based on the second spatial domain pixel value, that is, to modify the pixel value of the spatial domain image corresponding to the original carrier image. Modifications were made to obtain the modified spatial domain image. After scaling, it is exactly equal to the spatial pixel value corresponding to the final DCT coefficients after dithering modulation. .

[0087] Specifically, this application can first use a distortion function to calculate The cost of modifying each pixel; for Each element in Calculate its corresponding interpolation blocks on and interpolation blocks The weight value corresponding to each element in Then, based on the calculated modification cost, Sort the elements in the middle according to the order of modification, modifying elements with lower modification costs first and elements with higher modification costs last; then modify them sequentially according to the modification order. Modify the elements in the image based on the original scaled image pixel values. And the scaled image pixel values The difference between the original spatial image and the original spatial image The modification process can be represented by the following formula:

[0088] (6)

[0089] (7)

[0090] in, This refers to the position of the pixel to be modified within the interpolation block. It can be determined based on... and The difference determines the direction of modification. Simultaneously, the `sign` function can be used to limit the modification step size to 1. After modifying each element in the interpolation block, the modified interpolation block is checked to determine if it satisfies the condition that the current interpolation block, after computation, is exactly equal to... This involves checking whether the pixel values ​​in the interpolation block satisfy the formula shown below:

[0091] (8)

[0092] If the condition is met, the current interpolation block modification ends; otherwise, the modification continues. After all the interpolation blocks corresponding to all elements in the image have been modified, the modified spatial domain image will be satisfied. After scaling, it is exactly equal to .

[0093] The robust steganography method for images with high quality factor and large size proposed in this application uses a pre-scaled image of the original carrier image as a medium. It embeds secret messages using a dithering modulation algorithm combined with an adaptive steganography framework, adjusts unstable DCT coefficients through simulated channel processing, and finally generates a large-size, high-quality steganography image through inverse interpolation. This ensures that the generated steganography image has the same quality factor and size as the original carrier image, enabling covert communication and resisting scaling and JPEG recompression attacks in lossy social media channels, correctly extracting the secret information from the steganography image. This solves the problems of related image steganography algorithms that do not consider lossy processing in the channel, often resulting in steganography images that cannot correctly extract secret messages, or those that only consider single lossy processing such as scaling or recompression. These algorithms allow monitoring parties to easily observe the quality factor and size of user-uploaded images, thus distinguishing between communication and general public behavior, limiting their application and potentially leading to a series of behavioral security issues.

[0094] Next, referring to the accompanying drawings, a robust steganography apparatus for images with high quality factors and large sizes, proposed according to embodiments of this application, is described.

[0095] Figure 5 This is a schematic diagram of the structure of a robust steganography device for images with high quality factor and large size according to an embodiment of this application.

[0096] like Figure 5 As shown, the robust steganography device 10 for images with high quality factor and large size includes: a first processing module 100, a second processing module 200 and a steganography module 300.

[0097] The first processing module 100 is used to decode the carrier image to obtain the first spatial pixel value of the carrier image, and prescale the spatial image corresponding to the carrier image based on the channel characteristics of the target network platform to obtain the prescaled spatial image, and obtain the discrete cosine transform (DCT) coefficients of the prescaled spatial image.

[0098] The second processing module 200 is used to embed the target secret information into the DCT coefficients to obtain the initial secret DCT coefficients, and adjust the unstable coefficients in the initial secret DCT coefficients to obtain the final secret DCT coefficients.

[0099] The steganography module 300 is used to transform the final carrier DCT coefficients to the spatial domain to obtain the corresponding second spatial pixel values, and modify the first spatial pixel values ​​according to the second spatial pixel values ​​until the spatial image corresponding to the modified first spatial pixel values ​​meets the first preset scaling condition. The spatial image corresponding to the modified first spatial pixel values ​​is compressed according to the quality factor of the carrier image to generate the final carrier image that meets the preset size.

[0100] Optionally, in one embodiment of this application, the second processing module 200 includes: a first processing unit, an embedding unit, and a first modification unit.

[0101] The first processing unit is used to perform steganalytic quantization on the DCT coefficients to generate a steganalytic vector.

[0102] An embedding unit is used to embed target secret information into a steganographic carrier to obtain a steganographic sequence.

[0103] The first modification unit is used to modify the DCT coefficients according to the carrier sequence and the steganographic sequence of the steganographic carrier to obtain the initial steganographic DCT coefficients.

[0104] Optionally, in one embodiment of this application, it further includes: an acquisition module and a determination module.

[0105] The acquisition module acquires the quantization step size information of the coefficient block composed of DCT coefficients before embedding the target secret information into the DCT coefficients.

[0106] The determination module is used to determine the target location where the target secret information is embedded into the DCT coefficients based on the quantization step size information.

[0107] Optionally, in one embodiment of this application, the second processing module 200 includes: a second processing unit, a second modification unit, a generation unit, and an adjustment unit.

[0108] The second processing unit is used to perform inverse discrete cosine transform, spatial rounding and truncation on the initial dense DCT coefficients to obtain the corresponding first simulated spatial pixel values.

[0109] The second modification unit is used to modify the first simulated spatial pixel value based on the first simulated spatial pixel value to obtain the second simulated spatial pixel value.

[0110] The generation unit is used to obtain a pre-transmission intermediate image that meets the second preset scaling condition based on the second simulated spatial domain pixel value;

[0111] The adjustment unit is used to perform simulated channel processing on the pre-sent intermediate image to adjust the unstable coefficients in the initial DCT coefficients and obtain the final DCT coefficients.

[0112] Optionally, in one embodiment of this application, the steganography module 300 includes: a first calculation unit, a second calculation unit, a sorting unit, and a third modification unit.

[0113] The first computing unit is used to calculate the modification cost of each pixel in the spatial domain image using the target distortion function.

[0114] The second calculation unit is used to calculate the interpolation block corresponding to each element in the second spatial domain pixel value on the spatial domain image, and to obtain the weight value of each element in the interpolation block.

[0115] A sorting unit is used to sort the weight values ​​according to the modification cost to generate a modification sequence;

[0116] The third modification unit is used to modify the first spatial domain pixel value based on the modification sequence and the second spatial domain pixel value.

[0117] It should be noted that the foregoing explanation of the robust steganography method embodiment for images with high quality factor and large size also applies to the robust steganography apparatus for images with high quality factor and large size in this embodiment, and will not be repeated here.

[0118] The robust steganography device for large-size images with high quality factor proposed in this application uses a pre-scaled image of the original carrier image as a medium. It embeds secret messages based on a dithering modulation algorithm combined with an adaptive steganography framework, adjusts unstable DCT coefficients through simulated channel processing, and finally generates a large-size, high-quality steganography image through inverse interpolation. This ensures that the generated steganography image has the same quality factor and size as the original carrier image, enabling covert communication and resisting scaling and JPEG recompression attacks in lossy social media channels, correctly extracting the secret information from the steganography image. This solves the problems of related image steganography algorithms that do not consider lossy processing in the channel, often resulting in steganography images that cannot correctly extract secret messages, or those that only consider single lossy processing such as scaling or recompression. These algorithms allow monitoring parties to easily observe the quality factor and size of user-uploaded images, thus distinguishing communication behavior from general public behavior, limiting their application scope and potentially leading to a series of behavioral security issues.

[0119] Figure 6 A schematic diagram of the structure of an electronic device provided in an embodiment of this application. The electronic device may include:

[0120] The memory 601, the processor 602, and the computer program stored on the memory 601 and capable of running on the processor 602.

[0121] When the processor 602 executes the program, it implements the robust steganography method for images with high quality factor and large size provided in the above embodiments.

[0122] Furthermore, electronic devices also include:

[0123] Communication interface 603 is used for communication between memory 601 and processor 602.

[0124] The memory 601 is used to store computer programs that can run on the processor 602.

[0125] The memory 601 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.

[0126] If the memory 601, processor 602, and communication interface 603 are implemented independently, then the communication interface 603, memory 601, and processor 602 can be interconnected via a bus to complete communication between them. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 6 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0127] Optionally, in a specific implementation, if the memory 601, processor 602, and communication interface 603 are integrated on a single chip, then the memory 601, processor 602, and communication interface 603 can communicate with each other through an internal interface.

[0128] The processor 602 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application.

[0129] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the robust steganography method for images with high quality factors and large sizes as described above.

[0130] This application also provides a computer program product, including a computer program that can run computer instructions. When these computer instructions are executed by a processor, they implement the robust steganography method for images with high quality factors and large sizes provided in this application.

[0131] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0132] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0133] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or N executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0134] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0135] It should be understood that the various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. If implemented in hardware, as in another embodiment, it can be implemented using any one or more of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0136] Those skilled in the art will understand that all or part of the steps of the methods described in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it includes one or a combination of the steps of the method embodiments.

[0137] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0138] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.

Claims

1. A robust steganography method for images with high quality factor and large size, characterized in that, Includes the following steps: The carrier image is decoded to obtain the first spatial pixel value of the carrier image. Based on the channel characteristics of the target network platform, the spatial image corresponding to the carrier image is pre-scaled to obtain the pre-scaled spatial image. The discrete cosine transform (DCT) coefficients of the pre-scaled spatial image are then obtained. The target secret information is embedded into the DCT coefficients to obtain the initial secret-carrying DCT coefficients. The instability coefficients in the initial secret-carrying DCT coefficients are adjusted to obtain the final secret-carrying DCT coefficients. The final DCT coefficients of the carrier image are transformed to the spatial domain to obtain the corresponding second spatial pixel value. The first spatial pixel value is modified according to the second spatial pixel value until the spatial image corresponding to the modified first spatial pixel value meets the first preset scaling condition. The spatial image corresponding to the modified first spatial pixel value is compressed according to the quality factor of the carrier image to generate the final carrier image that meets the preset size. The step of modifying the first spatial domain pixel value according to the second spatial domain pixel value includes: calculating the modification cost of each pixel in the spatial domain image using the target distortion function; Calculate the interpolation block corresponding to each element in the second spatial domain pixel value on the spatial domain image, and obtain the weight value of each element in the interpolation block; sort the weight values ​​according to the modification cost to generate a modification sequence; modify the first spatial domain pixel value according to the second spatial domain pixel value based on the modification sequence.

2. The method according to claim 1, characterized in that, The step of embedding the target secret information into the DCT coefficients to obtain the initial secret-carrying DCT coefficients includes: The DCT coefficients are subjected to steganalytic quantization to generate a steganalytic vector; The target secret information is embedded into the steganographic carrier to obtain the steganographic sequence; The DCT coefficients are modified according to the carrier sequence of the steganographic carrier and the steganographic sequence to obtain the initial steganographic DCT coefficients.

3. The method according to claim 1, characterized in that, Before embedding the target secret information into the DCT coefficients, the method further includes: Obtain the quantization step size information of the coefficient block composed of the DCT coefficients; The target location where the target secret information is embedded into the DCT coefficients is determined based on the quantization step size information.

4. The method according to claim 1, characterized in that, The adjustment of the instability coefficient in the initial density DCT coefficient to obtain the final density DCT coefficient includes: The initial DCT coefficients are subjected to inverse discrete cosine transform, spatial rounding, and truncation to obtain the corresponding first simulated spatial pixel values. Based on the first simulated spatial pixel value, the first spatial pixel value is modified to obtain the second simulated spatial pixel value; Based on the second simulated spatial domain pixel value, a pre-sent intermediate image that meets the second preset scaling condition is obtained; The pre-sent intermediate image is processed using a simulated channel to adjust the instability coefficient in the initial DCT coefficients, thus obtaining the final DCT coefficients.

5. A robust steganography device for images with high quality factor and large size, characterized in that, include: The first processing module is used to decode the carrier image to obtain the first spatial pixel value of the carrier image, and pre-scale the spatial image corresponding to the carrier image based on the channel characteristics of the target network platform to obtain the pre-scaled spatial image, and obtain the discrete cosine transform (DCT) coefficients of the pre-scaled spatial image. The second processing module is used to embed the target secret information into the DCT coefficients to obtain the initial secret-carrying DCT coefficients, and adjust the instability coefficients in the initial secret-carrying DCT coefficients to obtain the final secret-carrying DCT coefficients. The steganography module is used to transform the final carrier DCT coefficients to the spatial domain to obtain the corresponding second spatial pixel value, and modify the first spatial pixel value according to the second spatial pixel value until the spatial image corresponding to the modified first spatial pixel value meets the first preset scaling condition. The spatial image corresponding to the modified first spatial pixel value is compressed according to the quality factor of the carrier image to generate a final carrier image that meets the preset size. The steganography module includes: a first calculation unit, used to calculate the modification cost of each pixel in the spatial image using a target distortion function; a second calculation unit, used to calculate the interpolation block corresponding to each element in the second spatial pixel value on the spatial image, and obtain the weight value of each element in the interpolation block; a sorting unit, used to sort the weight values ​​according to the modification cost to generate a modification sequence; and a third modification unit, used to modify the first spatial pixel value according to the second spatial pixel value based on the modification sequence.

6. The apparatus according to claim 5, characterized in that, The second processing module includes: The processing unit is used to perform steganalytic quantization on the DCT coefficients to generate a steganalytic vector. An embedding unit is used to embed the target secret information into the steganographic carrier to obtain a steganographic sequence. The modification unit is used to modify the DCT coefficients according to the carrier sequence of the steganographic carrier and the steganographic sequence to obtain the initial steganographic DCT coefficients.

7. An electronic device, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, the processor executing the program to implement the robust steganography method for images with high quality factor and large size as described in any one of claims 1-4.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, The program is executed by the processor to implement the robust steganography method for images with high quality factor and large size as described in any one of claims 1-4.

9. A computer program product, comprising a computer program, characterized in that, When the computer program is executed, it is used to implement the robust steganography method for images with high quality factors and large sizes as described in any one of claims 1-4.

Images