Image tampering detection method, apparatus, equipment, and storage medium based on adaptive hypergraph modeling
The image tampering detection method based on adaptive hypergraph modeling solves the problem of reduced image detection accuracy caused by the upgrading of forgery technology, realizes the forgery detection of original images, improves detection accuracy and prevents the spread of forged information.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- MACAO POLYTECHNIC INST
- Filing Date
- 2024-05-14
- Publication Date
- 2026-06-30
AI Technical Summary
The problem of reduced image detection accuracy due to upgrades in forgery techniques in existing technologies.
An image tampering detection method based on adaptive hypergraph modeling is adopted, which includes preprocessing the original image data, performing feature extraction, hypergraph generation and image detection through a trained adaptive hypergraph modeling image detection model, and generating hypergraph Laplacian data to achieve image tampering detection.
It improves the accuracy of image detection and effectively prevents the spread of forged image information.
Smart Images
Figure CN118429786B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of image detection technology, and in particular to an image tampering detection method, apparatus, device, and storage medium based on adaptive hypergraph modeling. Background Technology
[0002] Currently, the rapid advancements in image editing technology are driving the development of industries such as creative design, but they have also raised concerns about the manipulation of original image content. Therefore, when faced with scenarios involving verifying the authenticity of content and protecting intellectual property, image forgery detection methods, due to their ability to accurately locate tampered areas of the image, offer a solution that meets the needs of forgery detection.
[0003] While image forgery detection technology has potential in intellectual property protection, academic research, and digital evidence collection, it still needs to overcome problems such as the risk of forgery technology upgrades. Therefore, there is an urgent need for an image tampering detection method that can improve the accuracy of image detection.
[0004] The above content is only used to help understand the technical solution of the present invention and does not represent an admission that the above content is prior art. Summary of the Invention
[0005] The main objective of this invention is to provide an image tampering detection method, apparatus, device, and storage medium based on adaptive hypergraph modeling, aiming to solve the technical problem of reduced image detection accuracy caused by the upgrading of forgery technology in the prior art.
[0006] To achieve the above objectives, the present invention provides an image tampering detection method based on adaptive hypergraph modeling, the image tampering detection method based on adaptive hypergraph modeling comprising the following steps:
[0007] The original image data is preprocessed to obtain the processed image data;
[0008] The processed image data is input into the trained adaptive hypergraph modeling image detection model for image feature extraction to obtain a set of feature images.
[0009] Hypergraph generation is performed based on the feature image set to obtain hypergraph Laplacian data;
[0010] Image detection is performed based on the feature image set and the hypergraph Laplacian data to obtain the original image tampering detection result.
[0011] Optionally, the processed image data is input into a trained adaptive hypergraph modeling image detection model for image feature extraction to obtain a feature image set. The specific process includes:
[0012] Multi-scale feature extraction is performed on the processed image data to obtain a multi-scale high-level feature map;
[0013] Edge features are extracted from the processed image data to obtain a multi-scale edge feature map;
[0014] Noise features are extracted from the processed image data to obtain a noise feature map;
[0015] The multi-scale high-level feature map, the multi-scale edge feature map, and the noise feature map are integrated to obtain the feature image set.
[0016] Optionally, a hypergraph is generated based on the feature image set to obtain hypergraph Laplacian data. The specific process includes:
[0017] The feature image set is filtered by image resolution to obtain a low-resolution feature image set;
[0018] Hypergraph construction is performed based on the set of low-resolution feature images to obtain adaptive hypergraph data;
[0019] Hypergraph generation is performed based on the adaptive hypergraph data to obtain the hypergraph Laplacian data.
[0020] Optionally, a hypergraph is constructed based on the low-resolution feature image set to obtain adaptive hypergraph data, specifically including:
[0021] The low-resolution feature image set is stitched together to obtain the stitched feature image;
[0022] Based on the stitched feature image, feature learning is performed to obtain node feature data and hyperedge feature data;
[0023] Hypergraph is constructed based on the node feature data and the hyperedge feature data to obtain the adaptive hypergraph feature data.
[0024] Optionally, image detection is performed based on the feature image set and the hypergraph Laplacian data to obtain the original image tampering detection result, specifically including:
[0025] Construct a hypergraph convolutional layer based on the hypergraph Laplacian data;
[0026] The set of feature images is input into the hypergraph convolutional layer for feature extraction to obtain the set of feature images to be detected.
[0027] Image detection is performed based on the set of feature images to be detected and the hypergraph Laplacian data to obtain the original image tampering detection result.
[0028] Optionally, the specific training process of the trained adaptive hypergraph modeling image detection model includes:
[0029] Acquire sample image data and corresponding pixel-level label information;
[0030] The sample image data is preprocessed to obtain the processed sample image data;
[0031] The processed sample image data and the corresponding pixel-level label information are input into the adaptive hypergraph modeling image detection model to be trained for machine learning, and the stage training results of machine learning are detected to obtain the corresponding model training detection results.
[0032] If the model training detection results meet the preset model qualification standard, the trained adaptive hypergraph modeling image detection model is obtained.
[0033] Optionally, after inputting the processed sample image data and the corresponding pixel-level label information into the adaptive hypergraph modeling image detection model to be trained for machine learning, and detecting the stage training results of machine learning to obtain the corresponding model training detection results, the method further includes:
[0034] If the model training detection result does not meet the preset model qualification standard, the model weight parameters of the adaptive hypergraph modeling image detection model to be trained are adjusted and machine learning continues until the subsequent model training detection result meets the preset model qualification standard, and the trained adaptive hypergraph modeling image detection model is obtained.
[0035] Furthermore, to achieve the above objectives, the present invention also proposes an image tampering detection device based on adaptive hypergraph modeling, the image tampering detection device based on adaptive hypergraph modeling comprising:
[0036] Preprocessing module: preprocesses the original image data to obtain processed image data;
[0037] Feature extraction module: Inputs the processed image data into the trained adaptive hypergraph modeling image detection model to extract image features and obtain a set of feature images;
[0038] Hypergraph generation module: Generates a hypergraph based on the set of feature images to obtain hypergraph Laplacian data;
[0039] Image detection module: Performs image detection based on the feature image set and the hypergraph Laplacian data to obtain the original image tampering detection result.
[0040] Furthermore, to achieve the above objectives, the present invention also proposes an image tampering detection device based on adaptive hypergraph modeling. The image tampering detection device based on adaptive hypergraph modeling includes: a memory, a processor, and an image tampering detection program based on adaptive hypergraph modeling stored in the memory and executable on the processor. The image tampering detection program based on adaptive hypergraph modeling is configured to implement the steps of the image tampering detection method based on adaptive hypergraph modeling as described above.
[0041] Furthermore, to achieve the above objectives, the present invention also proposes a computer-readable storage medium storing a computer program, wherein the storage medium stores an image tampering detection program based on adaptive hypergraph modeling, and when the image tampering detection program based on adaptive hypergraph modeling is executed by a processor, it implements the steps of the image tampering detection method based on adaptive hypergraph modeling as described above.
[0042] This invention preprocesses the original image data to obtain processed image data; it then inputs the processed image data into a trained adaptive hypergraph modeling image detection model for image feature extraction, resulting in a feature image set; based on the feature image set, it generates a hypergraph to obtain hypergraph Laplacian data; finally, it performs image detection based on the feature image set and the hypergraph Laplacian data to obtain the original image tampering detection result. This invention, based on a trained adaptive hypergraph modeling image detection model, sequentially performs feature extraction, hypergraph generation, and image detection on the original image data, thereby ultimately obtaining the corresponding original image tampering detection result, achieving forgery detection of the original image, and effectively preventing the spread of forged image information. Attached Figure Description
[0043] Figure 1 This is a schematic diagram of the structure of an image tampering detection device based on adaptive hypergraph modeling in the hardware operating environment of the embodiment of the present invention;
[0044] Figure 2 This is a flowchart illustrating the first embodiment of the image tampering detection method based on adaptive hypergraph modeling of the present invention;
[0045] Figure 3 This is a flowchart illustrating the second embodiment of the image tampering detection method based on adaptive hypergraph modeling of the present invention;
[0046] Figure 4 This is a flowchart illustrating the third embodiment of the image tampering detection method based on adaptive hypergraph modeling of the present invention;
[0047] Figure 5 This is a flowchart illustrating the fourth embodiment of the image tampering detection method based on adaptive hypergraph modeling of the present invention;
[0048] Figure 6 This is a structural block diagram of the first embodiment of the image tampering detection device based on adaptive hypergraph modeling of the present invention.
[0049] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation
[0050] It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention.
[0051] Reference Figure 1 , Figure 1 This is a schematic diagram of the structure of an image tampering detection device based on adaptive hypergraph modeling, which is part of the hardware operating environment of the embodiment of the present invention.
[0052] like Figure 1 As shown, the image tampering detection device based on adaptive hypergraph modeling may include: a processor 1001, such as a central processing unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. The communication bus 1002 is used to enable communication between these components. The user interface 1003 may include a display screen or an input unit such as a keyboard; optionally, the user interface 1003 may also include a standard wired interface or a wireless interface. The network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a Wireless-Fidelity (Wi-Fi) interface). The memory 1005 may be high-speed random access memory (RAM) or stable non-volatile memory (NVM), such as a disk drive. The memory 1005 may also optionally be a storage device independent of the aforementioned processor 1001.
[0053] Those skilled in the art will understand that Figure 1 The structure shown does not constitute a limitation on image tampering detection devices based on adaptive hypergraph modeling, and may include more or fewer components than shown, or combine certain components, or have different component arrangements.
[0054] like Figure 1 As shown, the memory 1005, which serves as a storage medium, may include an operating system, a network communication module, a user interface module, and an image tampering detection program based on adaptive hypergraph modeling.
[0055] exist Figure 1In the image tampering detection device based on adaptive hypergraph modeling shown, the network interface 1004 is mainly used for data communication with the network server; the user interface 1003 is mainly used for data interaction with the user; the processor 1001 and memory 1005 in the image tampering detection device based on adaptive hypergraph modeling of the present invention can be set in the image tampering detection device based on adaptive hypergraph modeling. The image tampering detection device based on adaptive hypergraph modeling calls the image tampering detection program based on adaptive hypergraph modeling stored in the memory 1005 through the processor 1001 and executes the image tampering detection method based on adaptive hypergraph modeling provided in the embodiment of the present invention.
[0056] This invention provides an image tampering detection method based on adaptive hypergraph modeling, referring to... Figure 2 , Figure 2 This is a flowchart illustrating the first embodiment of an image tampering detection method based on adaptive hypergraph modeling according to the present invention.
[0057] In this embodiment, the image tampering detection method based on adaptive hypergraph modeling includes the following steps:
[0058] Step S10: Preprocess the original image data to obtain processed image data;
[0059] It should be noted that in the specific implementation, image preprocessing is required before the original image to be detected is performed. The purpose is to improve the quality and usability of the original image data so as to better perform subsequent image analysis and tamper detection tasks.
[0060] It should also be noted that, in specific implementations, the preprocessing operations performed on the original image may include image data cleaning, image size adjustment, image normalization, and image brightness adjustment. For example, in this embodiment, the original image size may be scaled to 512*512.
[0061] Step S20: Input the processed image data into the trained adaptive hypergraph modeling image detection model to extract image features and obtain a set of feature images;
[0062] It is understandable that, in specific implementations, the trained adaptive hypergraph modeling image detection model is an image detection system model that includes a multi-view, multi-scale feature extraction module, an adaptive hypergraph convolution module, and a hypergraph filtering module. Specifically, in the multi-view, multi-scale feature extraction module, ConvNeXt (Convolutional Neural Network Next) can be used as the feature extraction network. That is, the multi-view, multi-scale feature extraction module in the trained adaptive hypergraph modeling image detection model can extract image features from the processed image data to obtain the corresponding feature image set.
[0063] Step S30: Generate a hypergraph based on the feature image set to obtain hypergraph Laplacian data;
[0064] It should be noted that a hypergraph is an extension of the traditional graph concept in graph theory, in which edges can connect any number of nodes, not just two nodes. Compared to traditional graphs, its characteristic is that it can contain multiple hyperedges between nodes, which makes it better able to represent complex nonlinear structures.
[0065] It should also be noted that, in specific implementation, hypergraph generation essentially refers to the process of representing and learning data on the hypergraph structure, that is, the process of representing and learning feature image set data on the hypergraph structure, and the corresponding result is hypergraph Laplacian data.
[0066] Step S40: Perform image detection based on the feature image set and hypergraph Laplacian data to obtain the original image tampering detection result.
[0067] It should be noted that, in the specific implementation, image detection based on the feature image set and hypergraph Laplacian data is achieved through the hypergraph filtering module in the trained adaptive hypergraph modeling image detection model, and the forgery localization result finally output by the hypergraph filtering module is the original image tampering detection result.
[0068] This embodiment preprocesses the original image data to obtain processed image data; the processed image data is then input into a trained adaptive hypergraph modeling image detection model for image feature extraction, resulting in a feature image set; a hypergraph is generated based on the feature image set, yielding hypergraph Laplacian data; and image detection is performed based on the feature image set and the hypergraph Laplacian data to obtain the original image tampering detection result. This embodiment sequentially performs feature extraction, hypergraph generation, and image detection on the original image data using a trained adaptive hypergraph modeling image detection model, ultimately obtaining the corresponding original image tampering detection result, achieving forgery detection of the original image, and effectively preventing the spread of forged image information.
[0069] refer to Figure 3 , Figure 3 This is a flowchart illustrating a second embodiment of an image tampering detection method based on adaptive hypergraph modeling according to the present invention.
[0070] Based on the first embodiment described above, in this embodiment, step S20 specifically includes:
[0071] Step S21: Perform multi-scale feature extraction on the processed image data to obtain multi-scale high-level feature maps;
[0072] It should be noted that, in the specific implementation, the multi-view multi-scale feature extraction module specifically includes a multi-scale feature extraction network, an edge feature extraction network, and a noise feature extraction network. The multi-scale feature extraction network can specifically be ConvNeXt (Convolutional Neural Network Next).
[0073] Step S22: Extract edge features from the processed image data to obtain a multi-scale edge feature map;
[0074] Understandably, in the specific implementation, the multi-scale edge feature map is obtained based on the edge feature extraction network, and the edge feature extraction network can be composed of three convolutional layers with kernel sizes of {1, 3, 3} respectively.
[0075] Step S23: Extract noise features from the processed image data to obtain a noise feature map;
[0076] It should be noted that, in the specific implementation, noise feature extraction of the processed image data is achieved through the SRM filter (Spatial Rich Model) and convolutional layers in the noise feature extraction network.
[0077] Step S24: Integrate the multi-scale high-level feature map, multi-scale edge feature map, and noise feature map to obtain a feature image set.
[0078] Understandably, in practical implementation, the role of the feature image set is to extract key information from the original image, such as color, texture, shape, and spatial relationships, to provide a foundation for subsequent image processing tasks, that is, mainly for the subsequent image tampering detection process.
[0079] This embodiment extracts multi-scale features, edge features, and noise features from the processed image data to obtain a set of feature images, which provides a foundation for subsequent image detection and hypermap generation processes.
[0080] refer to Figure 4 , Figure 4 This is a flowchart illustrating the third embodiment of an image tampering detection method based on adaptive hypergraph modeling according to the present invention.
[0081] Based on the first embodiment described above, in this embodiment, step S30 specifically includes:
[0082] Step S31: Perform image resolution filtering on the feature image set to obtain a low-resolution feature image set;
[0083] It should be noted that in the specific implementation, the image resolutions of the various feature images in the feature image set are not exactly the same. Therefore, they can be filtered and classified according to their different image resolutions to obtain a low-resolution feature image set, where low resolution is defined as 1 / 16 of the original image resolution.
[0084] Step S32: Construct a hypergraph based on a set of low-resolution feature images to obtain adaptive hypergraph data;
[0085] Understandably, in the specific implementation, the process of constructing a hypergraph based on a set of low-resolution feature images includes first obtaining the corresponding node features and hyperedge features from the set of low-resolution feature images, then constructing an association matrix by multiplying the node features, the learnable parameter matrix, and the hyperedge features to obtain the association matrix representing the adaptive hypergraph, thus obtaining the adaptive hypergraph data.
[0086] Step S33: Generate a hypergraph based on the adaptive hypergraph data to obtain hypergraph Laplacian data.
[0087] It should be noted that the specific process of generating a hypergraph based on adaptive hypergraph data is essentially the process of building or expanding a hypergraph on the existing adaptive hypergraph data structure.
[0088] This embodiment constructs a hypergraph based on a low-resolution feature image set in the feature image set to obtain corresponding adaptive hypergraph data. Finally, it constructs a hypergraph based on the adaptive hypergraph data to obtain hypergraph Laplacian data, which provides a foundation for subsequent image detection and improves the accuracy of image detection.
[0089] refer to Figure 5 , Figure 5 This is a flowchart illustrating the fourth embodiment of an image tampering detection method based on adaptive hypergraph modeling according to the present invention.
[0090] Based on the third embodiment described above, in this embodiment, step S32 specifically includes:
[0091] Step S321: Perform image stitching on the low-resolution feature image set to obtain the stitched feature image;
[0092] It should be noted that, in the specific implementation, the process of image stitching of a low-resolution feature image set includes combining each image in the low-resolution feature image set into a larger image with a wider field of view through an image stitching algorithm. The final result is the stitched feature image.
[0093] Step S322: Perform feature learning based on the stitched feature image to obtain node feature data and hyperedge feature data;
[0094] It should be noted that, in the specific implementation, the feature learning process based on the stitched feature image is used to extract useful image information from the stitched feature image, namely node feature data and hyperedge feature data. Specifically, node feature data describes the attributes or information of each node in the stitched feature image; hyperedge feature data refers to the features obtained in the hypergraph by learning the features of all the hyperedges to which a node belongs. These features are formed by aggregating the node features of a node and the hyperedges it connects to.
[0095] Step S323: Construct a hypergraph based on node feature data and hyperedge feature data to obtain adaptive hypergraph feature data.
[0096] Understandably, in practice, adaptive hypergraph feature data is used to further construct the hypergraph Laplacian.
[0097] In this embodiment, the low-resolution feature image set is first stitched together to obtain the stitched feature image. Then, subsequent feature learning is performed to obtain the corresponding node feature data and hyperedge feature data, thus providing a foundation for the subsequent hypergraph construction process, and finally obtaining adaptive hypergraph feature data.
[0098] Further, image detection is performed based on the feature image set and the hypergraph Laplacian data to obtain the original image tampering detection result. Specifically, this includes: constructing a hypergraph convolutional layer based on the hypergraph Laplacian data; inputting the feature image set into the hypergraph convolutional layer for feature extraction to obtain the feature image set to be detected; and performing image detection based on the feature image set to be detected and the hypergraph Laplacian data to obtain the original image tampering detection result.
[0099] It should be noted that, in the specific implementation, the image detection process based on the feature image set and hypergraph Laplacian data is implemented through a hypergraph filtering module. This module can consist of multiple convolutional layers, hypergraph filtering layers, and upsampling layers. Furthermore, the convolutional and upsampling layers can be designed to progressively increase the feature map resolution, thereby restoring the images in the feature image set to their original input size. The final predicted forgery localization result output by the hypergraph filtering module is the original image tampering detection result.
[0100] Furthermore, the specific training process of the trained adaptive hypergraph modeling image detection model includes: acquiring sample image data and corresponding pixel-level label information; preprocessing the sample image data to obtain processed sample image data; inputting the processed sample image data and corresponding pixel-level label information into the adaptive hypergraph modeling image detection model to be trained for machine learning, and detecting the stage training results of machine learning to obtain the corresponding model training detection results; if the model training detection results meet the preset model qualification standard, the trained adaptive hypergraph modeling image detection model is obtained.
[0101] Understandably, in specific implementations, the machine learning qualification standard for the adaptive hypergraph modeling image detection model to be trained is a model parameter preset according to user needs. For example, if the preset model qualification standard is a tamper detection accuracy of 90%, then only when the model training detection result is higher than 90% can the corresponding trained adaptive hypergraph modeling image detection model be obtained.
[0102] Further, the processed sample image data and the corresponding pixel-level label information are input into the adaptive hypergraph modeling image detection model to be trained for machine learning, and the phase training results of machine learning are detected. After obtaining the corresponding model training detection results, the method further includes: if the model training detection results do not meet the preset model qualification standard, adjusting the model weight parameters of the adaptive hypergraph modeling image detection model to be trained and continuing machine learning until the subsequent model training detection results meet the preset model qualification standard, thereby obtaining the trained adaptive hypergraph modeling image detection model.
[0103] Furthermore, embodiments of the present invention also propose a computer-readable storage medium storing a computer program, wherein the storage medium stores an image tampering detection program based on adaptive hypergraph modeling, and when the image tampering detection program based on adaptive hypergraph modeling is executed by a processor, it implements the steps of the image tampering detection method based on adaptive hypergraph modeling as described above.
[0104] Since this storage medium adopts all the technical solutions of all the above embodiments, it has at least all the beneficial effects brought about by the technical solutions of the above embodiments, which will not be repeated here.
[0105] Reference Figure 6 , Figure 6 This is a structural block diagram of the first embodiment of the image tampering detection device based on adaptive hypergraph modeling of the present invention.
[0106] like Figure 6 As shown, the image tampering detection device based on adaptive hypergraph modeling proposed in this embodiment of the invention includes:
[0107] Preprocessing module 10: preprocesses the original image data to obtain processed image data;
[0108] Feature extraction module 20: Inputs the processed image data into the trained adaptive hypergraph modeling image detection model to extract image features and obtain a set of feature images;
[0109] Hypergraph generation module 30: Generates hypergraphs based on feature image sets to obtain hypergraph Laplacian data;
[0110] Image detection module 40: performs image detection based on feature image set and hypergraph Laplacian data to obtain the original image tampering detection result.
[0111] This embodiment preprocesses the original image data to obtain processed image data; the processed image data is then input into a trained adaptive hypergraph modeling image detection model for image feature extraction, resulting in a feature image set; a hypergraph is generated based on the feature image set, yielding hypergraph Laplacian data; and image detection is performed based on the feature image set and the hypergraph Laplacian data to obtain the original image tampering detection result. This embodiment sequentially performs feature extraction, hypergraph generation, and image detection on the original image data using a trained adaptive hypergraph modeling image detection model, ultimately obtaining the corresponding original image tampering detection result, achieving forgery detection of the original image, and effectively preventing the spread of forged image information.
[0112] It should be understood that the above are merely illustrative examples and do not constitute any limitation on the technical solutions of the present invention. In specific applications, those skilled in the art can make settings as needed, and the present invention does not impose any restrictions on this.
[0113] It should be noted that the workflow described above is merely illustrative and does not limit the scope of protection of this invention. In practical applications, those skilled in the art can select some or all of the workflow to achieve the purpose of this embodiment according to actual needs, and no restrictions are imposed here.
[0114] In addition, for technical details not described in detail in this embodiment, please refer to the image tampering detection method based on adaptive hypergraph modeling provided in any embodiment of the present invention, which will not be repeated here.
[0115] Furthermore, it should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.
[0116] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.
[0117] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as read-only memory (ROM) / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.
[0118] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.
Claims
1. An image tampering detection method based on adaptive hypergraph modeling, characterized in that, include: The original image data is preprocessed to obtain the processed image data; The processed image data is input into a trained adaptive hypergraph modeling image detection model for image feature extraction to obtain a feature image set. The specific process includes: Multi-scale feature extraction is performed on the processed image data to obtain a multi-scale high-level feature map; Edge features are extracted from the processed image data to obtain a multi-scale edge feature map; Noise features are extracted from the processed image data to obtain a noise feature map; The multi-scale high-level feature map, the multi-scale edge feature map, and the noise feature map are integrated to obtain the feature image set; Hypergraph generation is performed based on the aforementioned feature image set to obtain hypergraph Laplacian data. The specific process includes: The feature image set is filtered by image resolution to obtain a low-resolution feature image set; Hypergraph construction is performed based on the set of low-resolution feature images to obtain adaptive hypergraph data; Hypergraph generation is performed based on the adaptive hypergraph data to obtain the hypergraph Laplacian data; Image detection is performed based on the feature image set and the hypergraph Laplacian data to obtain the original image tampering detection result. The specific process includes: Construct a hypergraph convolutional layer based on the hypergraph Laplacian data; The set of feature images is input into the hypergraph convolutional layer for feature extraction to obtain the set of feature images to be detected. Image detection is performed based on the set of feature images to be detected and the hypergraph Laplacian data to obtain the original image tampering detection result.
2. The image manipulation detection method based on adaptive hypergraph modeling according to claim 1, characterized in that, Hypergraph construction is performed based on the aforementioned low-resolution feature image set to obtain adaptive hypergraph data, specifically including: The low-resolution feature image set is stitched together to obtain the stitched feature image; Based on the stitched feature image, feature learning is performed to obtain node feature data and hyperedge feature data; The adaptive hypergraph data is obtained by constructing a hypergraph based on the node feature data and the hyperedge feature data.
3. The image tampering detection method based on adaptive hypergraph modeling according to claim 1 or 2, characterized in that, The specific training process of the trained adaptive hypergraph modeling image detection model includes: Acquire sample image data and corresponding pixel-level label information; The sample image data is preprocessed to obtain the processed sample image data; The processed sample image data and the corresponding pixel-level label information are input into the adaptive hypergraph modeling image detection model to be trained for machine learning, and the stage training results of machine learning are detected to obtain the corresponding model training detection results. If the model training detection results meet the preset model qualification standard, the trained adaptive hypergraph modeling image detection model is obtained.
4. The image tampering detection method based on adaptive hypergraph modeling according to claim 3, characterized in that, The processed sample image data and the corresponding pixel-level label information are input into the adaptive hypergraph modeling image detection model to be trained for machine learning. After detecting the stage training results of machine learning and obtaining the corresponding model training detection results, the process further includes: If the model training detection result does not meet the preset model qualification standard, the model weight parameters of the adaptive hypergraph modeling image detection model to be trained are adjusted and machine learning continues until the subsequent model training detection result meets the preset model qualification standard, and the trained adaptive hypergraph modeling image detection model is obtained.
5. An image tampering detection device based on adaptive hypergraph modeling, characterized in that, The image tampering detection device based on adaptive hypergraph modeling includes: Preprocessing module: preprocesses the original image data to obtain processed image data; Feature extraction module: Inputs the processed image data into the trained adaptive hypergraph modeling image detection model for image feature extraction, obtaining a feature image set; the specific process includes: Multi-scale feature extraction is performed on the processed image data to obtain a multi-scale high-level feature map; Edge features are extracted from the processed image data to obtain a multi-scale edge feature map; Noise features are extracted from the processed image data to obtain a noise feature map; The multi-scale high-level feature map, the multi-scale edge feature map, and the noise feature map are integrated to obtain the feature image set; Hypergraph generation module: Generates a hypergraph based on the feature image set to obtain hypergraph Laplacian data. The specific process includes: The feature image set is filtered by image resolution to obtain a low-resolution feature image set; Hypergraph construction is performed based on the set of low-resolution feature images to obtain adaptive hypergraph data; Hypergraph generation is performed based on the adaptive hypergraph data to obtain the hypergraph Laplacian data; Image detection module: performs image detection based on the feature image set and the hypergraph Laplacian data to obtain the original image tampering detection result; the specific process includes: Construct a hypergraph convolutional layer based on the hypergraph Laplacian data; The set of feature images is input into the hypergraph convolutional layer for feature extraction to obtain the set of feature images to be detected. Image detection is performed based on the set of feature images to be detected and the hypergraph Laplacian data to obtain the original image tampering detection result.
6. An image tampering detection device based on adaptive hypergraph modeling, characterized in that, The image tampering detection device based on adaptive hypergraph modeling includes: a memory, a processor, and an image tampering detection program based on adaptive hypergraph modeling stored in the memory and executable on the processor. The image tampering detection program based on adaptive hypergraph modeling is configured to implement the image tampering detection method based on adaptive hypergraph modeling according to any one of claims 1 to 4.
7. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it can implement the steps in the image tampering detection method based on adaptive hypergraph modeling as described in any one of claims 1 to 4.