Image steganalysis method and system based on octave convolution and residual learning
The image steganalysis method using octave convolution and residual learning addresses the incomplete feature extraction in existing methods by decomposing and fusing high and low-frequency features, improving the accuracy and efficiency of steganalysis through enhanced feature extraction and classification.
Patent Information
- Application Number
- CN202411023471.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-29
- Publication Date
- 2024-12-06
AI Technical Summary
Existing image steganalysis methods fail to fully consider the characteristic relationship between high and low spatial frequency groups, leading to incomplete feature extraction and reduced accuracy and efficiency, especially in complex steganographic algorithms and diverse image data.
An image steganalysis method utilizing octave convolution and residual learning, which decomposes features into high and low frequencies, performs fusion using an OC-Block structure, and nonlinearly activates the features to enhance extraction and classification, with initialization of the first convolutional layer using a high-pass DCT filter.
Improves the accuracy and efficiency of steganalysis by focusing on high-frequency features with high correlation to embedding positions, extracting more comprehensive information and reducing the influence of low-frequency content, thus enhancing the reliability of steganalysis.
Smart Images

Figure CN119091185A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of image steganalysis, and in particular to an image steganalysis method and system based on octave convolution and residual learning. Background Art
[0002] In recent years, digital images have become one of the main ways of information dissemination and communication. With the advancement of technology, image steganography technology has been widely used as a method to hide secret information in images. However, in order to prevent illegal activities, detection and analysis of image steganalysis is crucial.
[0003] For example, the Chinese invention patent publication CN115760573A discloses a method for constructing an image super-resolution reconstruction model based on a multi-frequency hierarchical residual network, including: dividing to obtain a training set and a test set; inputting a low-resolution image in the training set into a feature extraction module to obtain shallow features of the low-resolution image; decomposing the shallow features into a high spatial frequency group and a low spatial frequency group; inputting the high spatial frequency group and the low spatial frequency group into a feature mapping module, outputting and fusing the high-frequency feature group and the low-frequency feature group using a convolution operation to obtain a deep feature; performing residual learning on the deep feature and the shallow feature, performing image reconstruction, and obtaining a high-resolution image; comparing the high-resolution image with the high-resolution image in the training set, obtaining a Loss value, completing the training, and obtaining a single image super-resolution reconstruction model based on deep learning. However, this method does not fully consider the characteristic relationship between the high spatial frequency group and the low spatial frequency group, which will affect the final analysis results.
[0004] Most of the existing image steganalysis methods rely on manually designed feature extractors. Although they have achieved certain results, they are often difficult to cope with complex steganographic algorithms and diverse image data due to their limited feature extraction capabilities. Even if neural networks such as octave convolution are integrated, the respective characteristics of high-frequency and low-frequency features are still not fully considered, and the features are not integrated, which will lead to the loss of some key information and affect the accuracy and efficiency of steganalysis. Summary of the invention
[0005] To address the deficiencies of the prior art, the present invention proposes an image steganalysis method, system, electronic device and computer-readable storage medium based on octave convolution and residual learning.
[0006] To achieve the above objectives, the technical solutions adopted by the present invention include:
[0007] An image steganalysis method based on octave convolution and residual learning, characterized by comprising:
[0008] S1. Image data acquisition: acquiring target image data, wherein the target image data is compressed data containing original image data and steganographic image data;
[0009] S2, data preprocessing: decompressing the target image data and performing preprocessing to obtain preprocessed image data;
[0010] S3, feature extraction and fusion: decomposing the features of the preprocessed image data into high-frequency features and low-frequency features, performing feature fusion based on the OC-Block structure that fuses octave convolution and residual learning, and obtaining high- and low-frequency fusion features;
[0011] S4, image data output: nonlinearly activate the high- and low-frequency fusion features to obtain output features, adjust the number of channels of the output features, and add them element by element with the low-frequency features to obtain output features with residual connections, and obtain high-dimensional feature map data by stacking multiple OC-Block structures and iterating;
[0012] S5. Image classification judgment: convert the high-dimensional feature map data into probability distribution, and perform steganalysis judgment based on the probability distribution.
[0013] Furthermore, the data preprocessing in step S2 specifically includes: decompressing the target image data into the spatial domain, initializing the weights of the first convolutional layer of the convolutional neural network, performing a convolution operation on the decompressed target image data to obtain preprocessed image data, wherein the first convolutional layer weights are initialized using a high-pass DCT filter.
[0014] Furthermore, after decomposing the features of the preprocessed image data into high-frequency features and low-frequency features, step S3 also includes: performing channel halving and convolution operations on the low-frequency features to obtain intermediate low-frequency features, and performing average pooling operations on the high-frequency features to obtain pooled high-frequency features.
[0015] Furthermore, the feature fusion in step S3 specifically includes: matching the intermediate low-frequency features with the pooled high-frequency features through an upsampling operation, and performing feature fusion to obtain high- and low-frequency fusion features.
[0016] Optionally, the adjusting the number of channels of the output feature in step S4 specifically includes: adjusting the number of channels of the output feature by using convolution so that the number of channels of the adjusted output feature is the same as that of the low-frequency feature.
[0017] Optionally, converting the high-dimensional feature map data into a probability distribution in step S5 specifically includes: passing the high-dimensional feature map data through a global average pooling layer to obtain a set of low-dimensional feature map data, linearly combining the low-dimensional feature map data through a fully connected layer to generate combined features, and then using a Softmax activation function to convert the combined features into a probability distribution.
[0018] In addition, the present invention also relates to an image steganalysis system based on octave convolution and residual learning, characterized in that it includes:
[0019] An image data acquisition module, used to acquire target image data, wherein the target image data is compressed data containing original image data and steganographic image data;
[0020] A data preprocessing module, used for decompressing the target image data and performing preprocessing to obtain preprocessed image data;
[0021] A feature extraction and fusion module, used for decomposing the features of the preprocessed image data into high-frequency features and low-frequency features, performing feature fusion based on an OC-Block structure that fuses octave convolution and residual learning, and obtaining high- and low-frequency fusion features;
[0022] An image data output module is used to perform nonlinear activation on the high- and low-frequency fusion features to obtain output features, adjust the number of channels of the output features, and add them element by element with the low-frequency features to obtain output features with residual connections, and obtain high-dimensional feature map data by stacking multiple OC-Block structures and iterating;
[0023] The image classification judgment module is used to convert the high-dimensional feature map data into a probability distribution and perform steganalysis judgment based on the probability distribution.
[0024] Furthermore, the data preprocessing module is also used for:
[0025] The target image data is decompressed into the spatial domain, and the weights of the first convolutional layer of the convolutional neural network are initialized. A convolution operation is performed on the decompressed target image data to obtain preprocessed image data, wherein the weights of the first convolutional layer are initialized using a high-pass DCT filter.
[0026] Furthermore, the feature extraction and fusion module further includes:
[0027] A low-frequency feature operation unit, used for performing channel halving and convolution operations on the low-frequency features to obtain intermediate low-frequency features;
[0028] A high-frequency feature operation unit, used for performing an average pooling operation on the high-frequency features to obtain pooled high-frequency features;
[0029] The fusion unit is used to match the intermediate low-frequency features with the pooled high-frequency features through an upsampling operation, and perform feature fusion to obtain high- and low-frequency fusion features.
[0030] Furthermore, the image data output module is also used for:
[0031] The number of channels of the output feature is adjusted by using convolution so that the number of channels of the adjusted output feature is the same as that of the low-frequency feature.
[0032] Optionally, the image classification and judgment module is further used to:
[0033] The high-dimensional feature map data is passed through a global average pooling layer to obtain a set of low-dimensional feature map data, the low-dimensional feature map data is linearly combined through a fully connected layer to generate combined features, and then the Softmax activation function is used to convert the combined features into probability distribution.
[0034] In addition, the present invention also relates to an electronic device, characterized in that it comprises a processor and a memory;
[0035] The memory is used to store operation instructions;
[0036] The processor is used to execute the above method by calling the operation instruction.
[0037] In addition, the present invention also relates to a computer-readable storage medium, characterized in that a computer program is stored on the storage medium, and the computer program implements the above method when executed by a processor.
[0038] By adopting the scheme provided by the present application, compressed target image data containing original image data and steganographic image data is obtained, and then it is decompressed and preprocessed to obtain preprocessed image data; the features of the preprocessed image data are decomposed into high-frequency features and low-frequency features, the low-frequency features are subjected to channel halving and convolution operations to obtain intermediate low-frequency features, the high-frequency features are subjected to average pooling operations to obtain pooled high-frequency features, the intermediate low-frequency features are matched with the pooled high-frequency features through upsampling operations, feature fusion is performed based on an OC-Block structure that fuses octave convolution and residual learning, the fused features are nonlinearly activated and the number of channels is adjusted, and they are element-by-element added to the low-frequency features to form output features with residual connections; by stacking multiple OC-Block structures and iterating, high-dimensional feature map data is obtained and converted into probability distribution for steganalysis judgment. Therefore, the above technical solution is adopted to compress low-frequency features and focus on extracting features in high-frequency areas with high correlation with the embedding position, which is more targeted at the problem of content adaptive steganography. Feature fusion can extract more comprehensive information and improve the accuracy and efficiency of steganalysis. At the same time, the high-pass DCT filter can make full use of the prior information of the image, making the convolution operation more in line with the characteristics of the image data, thereby improving the effect of subsequent feature extraction and steganalysis. BRIEF DESCRIPTION OF THE DRAWINGS
[0039] Figure 1a It is a schematic diagram of different frequencies of the original image in the prior art;
[0040] Figure 1b A schematic diagram of embedding secret information in an image in the prior art;
[0041] Figure 2 A flowchart of an image steganalysis method based on octave convolution and residual learning provided in an embodiment of the present application;
[0042] Figure 3 A schematic diagram of image data feature decomposition provided in an embodiment of the present application;
[0043] Figure 4 A schematic diagram of an OC-Block structure provided in an embodiment of the present application;
[0044] Figure 5 A schematic diagram of the structure of an image steganalysis system based on octave convolution and residual learning provided in an embodiment of the present application;
[0045] Figure 6 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0046] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and cannot be interpreted as limiting the present invention.
[0047] Those skilled in the art can understand that, unless otherwise stated, the singular forms "one", "an", "said" and "the" used herein may also include plural forms, and the "first", "second", etc., are only used to distinguish and define objects for the purpose of introducing the scheme clearly, and do not limit the objects themselves. Of course, the objects defined by "first" and "second" may be the same terminal, device, user, etc., or the same type of terminal, device and user. It should be further understood that the term "comprising" used in the specification of the present application refers to the presence of the features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof. The term "and / or" used herein includes all or any unit and all combinations of one or more associated listed items.
[0048] The technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems are described in detail below with specific embodiments. The following embodiments may be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments.
[0049] Steganography is to embed secret information into digital carriers in an imperceptible way, and hide the message in the visible field of vision without affecting the visual quality and use effect of the carrier file, so as to achieve covert communication. Steganography is a "double-edged sword". On the one hand, it can facilitate copyright protection, legal protection, military security and other fields to ensure the security of communication; at the same time, it also provides convenience for organizations and individuals with malicious intentions or improper purposes to engage in illegal activities. The main JPEG domain adaptive steganography algorithms include MMD, JC-UED, J-UNIWARD, etc., and the corresponding steganalysis technology is the reverse technology of steganography. By analyzing the statistical characteristics of the carrier, it detects whether there is secret information hidden in the media, estimates the length or key of the secret information, and even extracts the secret information or destroys the information. Successful steganalysis is the premise of tracing the source of information and cutting off the hostile hidden channel. At present, subtle steganalysis features are mainly extracted through networks based on deep learning, such as PNet and VNet networks, CNN-SCA-GFR networks, etc.
[0050] like Figure 1aAs shown in the figure, from left to right are natural images, low-frequency signals, and high-frequency signals. The low-frequency signals generally describe the overall structure that changes smoothly, while the high-frequency signals generally contain rich details, such as Figure 1b As shown, from left to right, the J-UNIWARD content adaptive steganography algorithm is used to embed secret information at different embedding rates. After the natural image is embedded with secret information, the embedding position of the secret information is concentrated in the area with complex image content and texture, corresponding to the high-frequency area of the JPEG image, while the smoother area of the image changes little, so more attention should be paid to the high-frequency features of the image.
[0051] like Figure 2 The flowchart of the image steganalysis method based on octave convolution and residual learning provided by the embodiment of the present application is shown, which specifically includes:
[0052] S1. Image data acquisition: acquiring target image data, wherein the target image data is compressed data containing original image data and steganographic image data for a JPEG image;
[0053] S2, data preprocessing: decompressing the target image data and performing preprocessing to obtain preprocessed image data;
[0054] The data preprocessing specifically includes: decompressing the target image data into the spatial domain without rounding the pixel values to integers to avoid any information loss, initializing the weights of the first convolutional layer of the convolutional neural network, and performing a convolution operation on the decompressed target image data to obtain preprocessed image data, wherein the first convolutional layer weights use 16 4×4 high-pass DCT filters B (k,l) Initialize.
[0055] The high-pass DCT filter is expressed as: Where (k, l) is the index of the high-pass DCT filter, and (m, n) is the index of the position in the 4×4 window, the convolution operation can be expressed as:
[0056]
[0057] By initializing the weights of the first convolutional layer with a high-pass DCT filter and using a padding operation to change the size of the feature map to a multiple of 8, the image content can be effectively suppressed, the weak steganographic signal can be enhanced, and the signal-to-noise ratio can be improved.
[0058] S3, feature extraction and fusion: decomposing the features of the preprocessed image data into high-frequency features and low-frequency features, performing channel halving and convolution operations on the low-frequency features to obtain intermediate low-frequency features, performing average pooling operations on the high-frequency features to obtain pooled high-frequency features, matching the intermediate low-frequency features with the pooled high-frequency features through upsampling operations, and performing feature fusion based on the OC-Block structure that fuses octave convolution and residual learning to obtain high- and low-frequency fusion features;
[0059] like Figure 3 As shown in the figure, by decomposing the features into high-frequency features and low-frequency features for processing, steganalysis can pay more attention to the high-frequency areas where the content steganalysis is concentrated, effectively extract steganalysis features, and compress low-frequency features at the same time, reduce the influence of low-frequency area content on steganalysis signals, reduce spatial redundancy, save storage, and feature fusion can extract more comprehensive information, improve the accuracy and efficiency of steganalysis. At the same time, a BN layer can be added after the convolution operation to prevent the gradient from disappearing, exploding, and overfitting in the deep neural network, and allow a relatively large learning rate to speed up the convergence.
[0060] like Figure 4 As shown in the figure, the OC-Block structure adopts an 18-layer octave convolution structure, in which layers 1-4, 5-8, 9-12, 13-15, and 16-18 use convolution kernels of sizes 16×3×3, 32×3×3, 64×3×3, 128×3×3, and 256×3×3, respectively, which can effectively extract and process high-frequency area features with high correlation with the embedding position, and can retain the weak signal generated by message embedding in the deep layer. The convolution process of the octave convolution structure is:
[0061] X H =f(X H ; W H→H )+npsample(f(X L ; W L→H ), 2)
[0062] Y L =f(X L :W L→L )+f(pool(X H , 2); W H→L )
[0063] By representing X and Y as the input and output feature maps of the octave convolution respectively, X can be decomposed into high and low frequency components {X H , X L}, Y can be decomposed into {Y H , Y L},Y A→B represents the convolution update process from feature map group A to B, then YH =Y H→H +Y L→H and Y L =Y L →L +Y H→L , Y H→H , Y L→L Indicates the frequency information update, Y H→L , Y L→H It represents the information exchange between frequencies, W is the convolution kernel, and pool(x, k) represents average pooling.
[0064] S4, image data output: The high- and low-frequency fusion features are nonlinearly activated by using the rectified linear unit ReLU activation function to obtain output features, and the number of channels of the output features is adjusted by using convolution, so that the adjusted output features have the same number of channels as the low-frequency features, and then the output features are element-by-element added with the low-frequency features to obtain output features with residual connections, and high-dimensional feature map data is obtained by stacking multiple OC-Block structures and iterating;
[0065] Through nonlinear activation and channel number adjustment, the expressiveness of features can be further enhanced, and the stability and reliability of features can be improved. Stacking multiple OC-Block structures can extract deeper feature information.
[0066] S5, image classification judgment: the high-dimensional feature map data is passed through a global average pooling layer to obtain a set of low-dimensional feature map data, the low-dimensional feature map data is linearly combined through a fully connected layer to generate combined features, and then the combined features are converted into probability distributions using a Softmax activation function, and steganalysis is performed based on the probability distributions;
[0067] Through global average pooling and linear combination, the dimension of features can be effectively reduced and key information can be extracted, while the Softmax activation function can convert features into probability distribution, providing an intuitive and explainable basis for steganalysis judgment. The entire classification and judgment process ensures the accuracy and reliability of steganalysis.
[0068] based on Figure 2 The image steganalysis method based on octave convolution and residual learning is shown in FIG. , and on the other hand, it also relates to an image steganalysis system based on octave convolution and residual learning, such as Figure 5 As shown, including:
[0069] An image data acquisition module 501 is used to acquire target image data, wherein the target image data is compressed data containing original image data and steganographic image data;
[0070] A data preprocessing module 502 is used to decompress the target image data and perform preprocessing to obtain preprocessed image data;
[0071] A feature extraction and fusion module 503 is used to decompose the features of the preprocessed image data into high-frequency features and low-frequency features, and perform feature fusion based on an OC-Block structure that fuses octave convolution and residual learning to obtain high- and low-frequency fusion features;
[0072] An image data output module 504 is used to perform nonlinear activation on the high- and low-frequency fusion features to obtain output features, adjust the number of channels of the output features, and add them element by element with the low-frequency features to obtain output features with residual connections, and obtain high-dimensional feature map data by stacking multiple OC-Block structures and iterating;
[0073] The image classification judgment module 505 is used to convert the high-dimensional feature map data into a probability distribution and perform steganalysis judgment based on the probability distribution.
[0074] Furthermore, the data preprocessing module 502 is also used for:
[0075] The target image data is decompressed into the spatial domain, and the weights of the first convolutional layer of the convolutional neural network are initialized. A convolution operation is performed on the decompressed target image data to obtain preprocessed image data, wherein the weights of the first convolutional layer are initialized using a high-pass DCT filter.
[0076] Furthermore, the feature extraction and fusion module 503 further includes:
[0077] A low-frequency feature operation unit 5031 is used to perform channel halving and convolution operations on the low-frequency features to obtain intermediate low-frequency features;
[0078] A high-frequency feature operation unit 5032 is used to perform an average pooling operation on the high-frequency features to obtain pooled high-frequency features;
[0079] The fusion unit 5033 is used to match the intermediate low-frequency features with the pooled high-frequency features through an upsampling operation, and perform feature fusion to obtain high- and low-frequency fusion features.
[0080] Furthermore, the image data output module 504 is further used for:
[0081] The number of channels of the output feature is adjusted by using convolution so that the number of channels of the adjusted output feature is the same as that of the low-frequency feature.
[0082] Optionally, the image classification and judgment module 505 is further used to:
[0083] The high-dimensional feature map data is passed through a global average pooling layer to obtain a set of low-dimensional feature map data, the low-dimensional feature map data is linearly combined through a fully connected layer to generate combined features, and then the Softmax activation function is used to convert the combined features into probability distribution.
[0084] By using this system, the above method can be executed and the corresponding technical effects can be achieved.
[0085] An embodiment of the present invention further provides an electronic device for executing the above method, as an implementation device of the method, comprising a processor and a memory;
[0086] A memory, used for storing operation instructions;
[0087] The processor is used to execute the image steganalysis method provided in any embodiment of the present application by calling an operation instruction.
[0088] As an example, Figure 6 A schematic diagram of the structure of an electronic device applicable to the embodiment of the present application is shown. Figure 6 As shown, the electronic device 600 includes: a processor 601 and a memory 603. The processor 601 and the memory 603 are connected, such as through a bus 602. Optionally, the electronic device 600 may also include a transceiver 604. It should be noted that in actual applications, the transceiver 604 is not limited to one. It is understandable that the structure illustrated in the embodiment of the present invention does not constitute a specific limitation on the specific structure of the electronic device 600. In other embodiments of the present application, the electronic device 600 may include more or fewer components than shown in the figure, or combine certain components, or split certain components, or arrange the components differently. The illustrated components may be implemented in hardware, software, or a combination of software and hardware. Optionally, the electronic device may also include a display screen 605 for displaying images, or receiving user operation instructions when necessary.
[0089] The processor 601 is applied in the embodiment of the present application to implement the method shown in the above method embodiment. The transceiver 604 may include a receiver and a transmitter. The transceiver 604 is applied in the embodiment of the present application to implement the function of the electronic device of the embodiment of the present application communicating with other devices when executed.
[0090] Processor 601 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It may implement or execute various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 601 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, and the like.
[0091] The processor 601 may also include one or more processing units, for example, the processor 601 may include an application processor (application processor, AP), a modem processor, a graphics processor (graphics processing unit, GPU), an image signal processor (image signal processor, ISP), a controller, a memory, a video codec, a digital signal processor (digital signal processor, DSP), a baseband processor, and / or a neural network processor (Neural-network Processing Unit, NPU), etc. Different processing units may be independent devices or integrated into one or more processors. Among them, the controller may be the nerve center and command center of the electronic device 600. The controller may generate an operation control signal according to the instruction opcode and the timing signal to complete the control of fetching and executing instructions. A memory may also be set in the processor 601 for storing instructions and data. In some embodiments, the memory in the processor 601 is a high-speed cache memory. The memory may store instructions or data that the processor 601 has just used or circulated. If the processor 601 needs to use the instruction or data again, it can be directly called from the memory. Repeated access is avoided, the waiting time of the processor 601 is reduced, and the efficiency of the system is improved.
[0092] The processor 601 can run the image steganalysis method provided in the embodiment of the present application to reduce the complexity of user operations, improve the intelligence of the terminal device, and enhance the user experience. The processor 601 can include different devices. For example, when the CPU and GPU are integrated, the CPU and GPU can cooperate to execute the image steganalysis method provided in the embodiment of the present application, with part of the algorithm executed by the CPU and the other part of the algorithm executed by the GPU to obtain faster processing efficiency.
[0093] The bus 602 may include a path to transmit information between the above components. The bus 602 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The bus 602 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 6 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.
[0094] The memory 603 may be a ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, a RAM (Random Access Memory) or other types of dynamic storage devices that can store information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory), or may include a high-speed random access memory, or may include a non-volatile memory, such as at least one disk storage device, a flash memory device, a universal flash storage (UFS), or other optical disk storage, optical disk storage (including compressed optical disk, laser disk, optical disk, digital versatile disk, Blu-ray disk, etc.), a disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited to this.
[0095] Optionally, the memory 603 is used to store application code for executing the solution of the present application, and the execution is controlled by the processor 601. The processor 601 is used to execute the application code stored in the memory 603 to implement the image steganalysis method provided in any embodiment of the present application.
[0096] The memory 603 can be used to store computer executable program codes, which include instructions. The processor 601 executes various functional applications and data processing of the electronic device 600 by running the instructions stored in the memory 603. The memory 603 may include a program storage area and a data storage area. Among them, the program storage area may store the code of the operating system, the application program, etc. The data storage area may store data created during the use of the electronic device 600 (such as images, videos, etc. collected by the camera application), etc.
[0097] The memory 603 may also store one or more computer programs corresponding to the image steganalysis method provided in the embodiment of the present application. The one or more computer programs are stored in the above-mentioned memory 603 and are configured to be executed by the one or more processors 601. The one or more computer programs include instructions, and the above-mentioned instructions can be used to execute each step in the above-mentioned corresponding embodiment.
[0098] Of course, the code of the image steganalysis method provided in the embodiment of the present application can also be stored in an external memory.
[0099] The display screen 605 includes a display panel. The display panel can be a liquid crystal display (LCD), an organic light-emitting diode (OLED), an active-matrix organic light-emitting diode or an active-matrix organic light-emitting diode (AMOLED), a flexible light-emitting diode (FLED), Miniled, MicroLed, Micro-oLed, a quantum dot light-emitting diode (QLED), etc. In some embodiments, the electronic device 600 may include 1 or N display screens 605, where N is a positive integer greater than 1. The display screen 605 can be used to display information input by a user or information provided to a user and various graphical user interfaces (GUIs). For example, the display screen 605 can display photos, videos, web pages, or files.
[0100] The electronic device provided in the embodiment of the present application is applicable to any embodiment of the above method. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding method provided above, which will not be repeated here.
[0101] An embodiment of the present invention also provides a computer-readable storage medium capable of implementing all the steps in the method in the above embodiment, wherein a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, all the steps in the method in the above embodiment are implemented.
[0102] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0103] The present invention is described with reference to the flowchart and / or block diagram of the method, device (system) and computer program product according to the embodiment of the present invention. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the process and / or box in the flowchart and / or block diagram can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device produce a system for implementing the functions specified in one process or multiple processes in the flowchart and / or one box or multiple boxes in the block diagram.
[0104] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction system that implements the functions specified in one or more processes in the flowchart and / or one or more boxes in the block diagram.
[0105] These computer program instructions may also be loaded onto a computer or other programmable data processing device so that a series of operating steps are performed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one or more processes of the flowchart and / or one or more blocks of the block diagram. Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they are aware of the basic creative concepts. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.
[0106] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.
Claims
1. An image steganalysis method based on octave convolution and residual learning, characterized in that: include: S1. Image data acquisition: acquiring target image data, wherein the target image data is compressed data containing original image data and steganographic image data; S2, data preprocessing: decompressing the target image data and performing preprocessing to obtain preprocessed image data; S3, feature extraction and fusion: decomposing the features of the preprocessed image data into high-frequency features and low-frequency features, performing feature fusion based on the OC-Block structure that fuses octave convolution and residual learning, and obtaining high- and low-frequency fusion features; S4, image data output: nonlinearly activate the high- and low-frequency fusion features to obtain output features, adjust the number of channels of the output features, and add them element by element with the low-frequency features to obtain output features with residual connections, and obtain high-dimensional feature map data by stacking multiple OC-Block structures and iterating; S5. Image classification judgment: convert the high-dimensional feature map data into probability distribution, and perform steganalysis judgment based on the probability distribution.
2. The method according to claim 1, characterized in that The data preprocessing in step S2 specifically includes: decompressing the target image data into the spatial domain, initializing the weights of the first convolutional layer of the convolutional neural network, performing a convolution operation on the decompressed target image data to obtain preprocessed image data, wherein the weights of the first convolutional layer are initialized using a high-pass DCT filter.
3. The method according to claim 2, characterized in that After decomposing the features of the preprocessed image data into high-frequency features and low-frequency features, step S3 further includes: performing channel halving and convolution operations on the low-frequency features to obtain intermediate low-frequency features, and performing average pooling operations on the high-frequency features to obtain pooled high-frequency features.
4. The method according to claim 3, characterized in that The feature fusion in step S3 specifically includes: matching the intermediate low-frequency features with the pooled high-frequency features through an upsampling operation, and performing feature fusion to obtain high- and low-frequency fusion features.
5. The method according to claim 1 or 4, characterized in that: The step S4 of adjusting the number of channels of the output feature specifically includes: using convolution to adjust the number of channels of the output feature so that the number of channels of the adjusted output feature is the same as that of the low-frequency feature.
6. The method according to claim 1, characterized in that The step S5 of converting the high-dimensional feature map data into a probability distribution specifically includes: obtaining a set of low-dimensional feature map data through a global average pooling layer, linearly combining the low-dimensional feature map data through a fully connected layer to generate combined features, and then converting the combined features into a probability distribution using a Softmax activation function.
7. An image steganalysis system based on octave convolution and residual learning, characterized in that: include: An image data acquisition module, used to acquire target image data, wherein the target image data is compressed data containing original image data and steganographic image data; A data preprocessing module, used for decompressing the target image data and performing preprocessing to obtain preprocessed image data; A feature extraction and fusion module, used for decomposing the features of the preprocessed image data into high-frequency features and low-frequency features, performing feature fusion based on an OC-Block structure that fuses octave convolution and residual learning, and obtaining high- and low-frequency fusion features; An image data output module is used to perform nonlinear activation on the high- and low-frequency fusion features to obtain output features, adjust the number of channels of the output features, and add them element by element with the low-frequency features to obtain output features with residual connections, and obtain high-dimensional feature map data by stacking multiple OC-Block structures and iterating; The image classification judgment module is used to convert the high-dimensional feature map data into a probability distribution and perform steganalysis judgment based on the probability distribution.
8. The system according to claim 7, characterized in that The feature extraction and fusion module also includes: A low-frequency feature operation unit, used for performing channel halving and convolution operations on the low-frequency features to obtain intermediate low-frequency features; A high-frequency feature operation unit, used for performing an average pooling operation on the high-frequency features to obtain pooled high-frequency features; The fusion unit is used to match the intermediate low-frequency features with the pooled high-frequency features through an upsampling operation, and perform feature fusion to obtain high- and low-frequency fusion features.
9. An electronic device, characterized in that: including a processor and a memory; The memory is used to store operation instructions; The processor is used to execute the method according to any one of claims 1 to 6 by calling the operation instruction.
10. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.
