Image data protection method and device based on distributed nodes

Abstract

The present invention discloses an image data protection method and device based on distributed nodes. The method comprises: firstly, dividing the image data to be processed according to the number of nodes to obtain sub-data, and sending the obtained sub-data to each node, wherein each node receives only one sub-data; then, processing the sub-data by a multimodal model in each node to obtain processed sub-data; then, generating noise by each node, and adding the generated noise to the processed sub-data in the node itself to obtain noise data; finally, dividing the noise data in the node itself by each node to obtain multiple segmented data, the number of segmented data is consistent with the number of nodes, and sending the segmented data obtained by segmentation to all nodes, so as to realize distributed storage and calculation of image data, and each node only transmits part of the data, thereby ensuring the overall integrity and security, and reducing the risk of leakage of sensitive information when the image data is applied in the multimodal model.

Description

Technical Field

[0001] The present invention belongs to the field of computer technology, and in particular relates to an image data protection method and device based on distributed nodes. Background Art

[0002] Image data is often input into multimodal models for image classification. That is, the image data is input into the multimodal model, and the model classifies the image data into different categories and generates corresponding text descriptions based on the pictures. However, image data often contains a large amount of sensitive information, and the image data is concentrated in one node for processing, posing a high risk of leakage.

[0003] Therefore, how to reduce the risk of leakage of sensitive information when image data is applied in a multimodal model is a technical problem to be solved by those skilled in the art. Summary of the invention

[0004] The purpose of the present invention is to solve the technical problem in the prior art that sensitive information is easily leaked when image data is applied in a multimodal model.

[0005] To achieve the above technical objectives, on the one hand, the present invention provides an image data protection method based on distributed nodes, which is applied to a distributed node architecture, and each node in the architecture is provided with a multimodal model. The method includes:

[0006] The image data to be processed is divided into sub-data according to the number of nodes, and the sub-data obtained are sent to each node, wherein each node only receives one sub-data;

[0007] The sub-data are processed by the multimodal model in each node to obtain processed sub-data;

[0008] Generate noise through each node, and add the generated noise to the processed sub-data in the node itself to obtain noise data;

[0009] After each node divides the noise data in the node itself, a plurality of divided data are obtained, the number of the divided data is consistent with the number of nodes, and the divided data obtained by the division are sent to all nodes, and each node only receives one piece of divided data.

[0010] Furthermore, the method further includes verifying the noise data to be verified when any node as a verification node receives a verification instruction to verify a certain noise data, and the verification process is as follows;

[0011] Sending a first instruction to a node corresponding to the noise data to be verified, using the node as a proof node, so that the proof node sends a verification value to the verification node;

[0012] In response to the verification value, sending a random challenge value generated by a pseudo-random number generator to the proof node through the verification node, so that the proof node sends response information to the verification node after receiving the random challenge value, the random challenge value specifically includes a first random challenge value and a second random challenge value, and the response information specifically includes first response information and second response information;

[0013] The verification node determines whether the corresponding noise data is correct based on the random challenge value, the verification value and the response information.

[0014] Further, the verification node determines whether the corresponding noise data is correct based on the random challenge value, the verification value and the response information. Specifically, when the following formula holds true, the corresponding noise data is correct:

[0015] ,

[0016] In the formula, and is a public generator, For the first response information, is the second response information, To verify the value, is the first random challenge value, is the second random challenge value, For the modulo operation, is a prime number.

[0017] Furthermore, the method further includes verifying the noise data to be verified when any node as a verification node receives a verification instruction to verify at least two noise data at the same time, and the verification process is as follows;

[0018] Sending a first instruction to all nodes corresponding to the noise data to be verified, and using the corresponding nodes as proof nodes, so that all proof nodes send verification values ​​to the verification node;

[0019] After the verification node receives the verification value, the verification node sends a random challenge value generated by a pseudo-random number generator to the proof node corresponding to the received verification value, so that the corresponding proof node sends a response message to the verification node after receiving the random challenge value;

[0020] The verification node determines whether the corresponding noise data is correct based on the random challenge value, verification value and response information corresponding to the same proof node.

[0021] Furthermore, the verification value is specifically determined by the following formula:

[0022] ,

[0023] In the formula, To verify the value, and is a public generator, To prove the value a node has, A random number generated by the proof node. For the modulo operation, is a prime number.

[0024] Furthermore, the response information is specifically determined by the following formula:

[0025] ,

[0026] ,

[0027] In the formula, is not equal to prime number.

[0028] Further, the generating noise specifically includes:

[0029] Transforming the corresponding processed sub-data to generate Gaussian distribution samples;

[0030] Standardizing the Gaussian distribution samples to obtain standardized samples;

[0031] A sampling value is obtained by randomly sampling from a gamma distribution, and the noise is obtained by multiplying the sampling value with a standardized sample.

[0032] On the other hand, the present invention also provides an image data protection device based on distributed nodes, which is applied to a distributed node architecture, and each node in the architecture is provided with a multimodal model, and the device comprises:

[0033] A segmentation module is used to segment the image data to be processed according to the number of nodes to obtain sub-data, and send the obtained sub-data to each node, wherein each node only receives one copy of the sub-data;

[0034] A generation module, located in each node, for generating noise and adding the noise to processed sub-data obtained after the multimodal model in each node processes the sub-data, thereby obtaining noise data;

[0035] The sending module is located in each node and is used to segment the noise data in the node itself to obtain multiple segmented data, the number of which is consistent with the number of nodes, and send the segmented data to all nodes, and each node only receives one piece of segmented data.

[0036] The present invention provides an image data protection method and device based on distributed nodes. Compared with the prior art, the method first divides the image data to be processed according to the number of nodes to obtain sub-data, and sends the obtained sub-data to each node, wherein each node receives only one sub-data; then the sub-data is processed by the multimodal model in each node to obtain processed sub-data; then noise is generated by each node, and the generated noise is added to the processed sub-data in the node itself to obtain noise data; finally, the noise data in the node itself is divided by each node to obtain multiple divided data, the number of the divided data is consistent with the number of nodes, and the divided data obtained by the division is sent to all nodes, each node receives only one divided data, and can realize distributed storage and calculation of image data, each node only transmits part of the data, even if the transmitted data is attacked, it will not cause the loss or leakage of the entire data set, thereby ensuring the overall integrity and security, and reducing the risk of leakage of sensitive information when the image data is applied in the multimodal model. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the embodiments of this specification or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0038] Figure 1 The figure is a flow chart of the distributed node-based image data protection method provided in the embodiment of this specification;

[0039] Figure 2 Shown is a schematic diagram of the structure of an image data protection device based on distributed nodes provided in an embodiment of this specification. DETAILED DESCRIPTION

[0040] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of this application.

[0041] like Figure 1The flowchart of the distributed node-based image data protection method provided by the embodiment of this specification is shown. Although this specification provides the method operation steps or device structures shown in the following embodiments or drawings, the method or device may include more or fewer operation steps or module units after partial merger based on routine or no creative labor. In the steps or structures where there is no necessary causal relationship logically, the execution order of these steps or the module structure of the device is not limited to the execution order or module structure shown in the embodiments of this specification or drawings. When the method or module structure is applied in actual devices, servers or terminal products, it can be executed sequentially or in parallel according to the method or module structure shown in the embodiments or drawings (for example, in a parallel processor or multi-threaded processing environment, or even in an implementation environment of a server cluster).

[0042] The distributed node-based image data protection method provided in the embodiments of this specification is applied to a distributed node architecture, and a multimodal model is set in each node in the architecture, such as Figure 1 As shown, the method specifically comprises the following steps:

[0043] Step S101 : dividing the image data to be processed into sub-data according to the number of nodes, and sending the sub-data to each node, wherein each node receives only one copy of the sub-data.

[0044] Specifically, after determining the image data to be processed, the image data to be processed is divided or partitioned according to the number of nodes in the distributed node architecture. Each node can only process the corresponding sub-data. Even if a node is attacked, it will not cause the entire image data to be processed to be lost or leaked, thus realizing the distributed storage and processing of data and ensuring the overall integrity and security. After receiving the corresponding sub-data, each node also performs L2 standardization on the corresponding sub-data to facilitate the processing of the sub-data.

[0045] Step S102: Process the sub-data using the multimodal model in each node to obtain processed sub-data.

[0046] Specifically, each node processes the sub-data it receives respectively, thereby processing the sub-data; in this embodiment, the processing is the operation that the node needs to perform after receiving the data. The specific operation is set according to the node's own needs and has nothing to do with the solution of the present application. For example, the operation can be a certain calculation on the data, or it can be the use of the data to train a certain neural network, etc.

[0047] Step S103: Generate noise through each node, and add the generated noise to the processed sub-data in the node itself to obtain noise data.

[0048] The generated noise specifically includes:

[0049] Transforming the corresponding processed sub-data to generate Gaussian distribution samples;

[0050] Standardizing the Gaussian distribution samples to obtain standardized samples;

[0051] A sampling value is obtained by randomly sampling from a gamma distribution, and the noise is obtained by multiplying the sampling value with a standardized sample.

[0052] That is to say, by adding noise to different processed sub-data, privacy protection is combined to achieve overall distribution. Not only the data privacy of a single node is protected, but also the privacy protection of the entire computing process is ensured. Since the noise is generated and added during the computing process, even if the computing results are made public, the actual data cannot be recovered from it, thereby protecting the privacy of the data owner.

[0053] Step S104: After each node divides the noise data in the node itself, a plurality of divided data are obtained, the number of the divided data is consistent with the number of nodes, and the divided data obtained by the division are sent to all nodes, and each node only receives one copy of the divided data.

[0054] It should be noted that the segmentation process here is as follows. If the number of images in the noise data is 3 images (image A, image B, image C), and the number of nodes in the distributed node structure is three, then each image must be segmented. For example, for image A, image A is segmented to form image A1, image A2, and image A3. Similarly, image B1, image B2, image B3, image C1, image C2, and image C3 can also be obtained. Then image A1, image B1, and image C1 are composed of sub-data, and image A1, image B2, and image C3 can also be composed of sub-data. It can be understood that the number of segmentations is the same as the number of nodes, and each image in the noise data will also be segmented according to the number of nodes. That is to say, the segmentation of the noise data here is not simply to segment the images in the noise data into the same number as the number of nodes, but to segment each image in the noise data according to the number of nodes, and then extract one image fragment from each image at a time to form the segmented data, and send the segmented data to each node for subsequent processing by each node.

[0055] In a specific application scenario, the noise data is shown as ,Will Split into share, is the number of nodes, so we get , , , , , For the The first node Split data into pieces, The original data, that is, the noise data, can be reconstructed by Lagrange interpolation method. The reconstruction process is shown as follows:

[0056] ,

[0057] ,

[0058] In the formula, is a polynomial, is a large prime number, is the preset number and is less than , for The coefficient of for The coefficient of , and so on, is a constant term, and its value is After being converted into a standard polynomial form, the value in front of the corresponding power term is the value of the corresponding coefficient.

[0059] The segmentation data also needs to satisfy a constraint formula, which is as follows:

[0060] ,

[0061] In some embodiments, the method further includes verifying the noise data to be verified when any node as a verification node receives a verification instruction to verify a certain noise data, and the verification process is as follows;

[0062] Sending a first instruction to a node corresponding to the noise data to be verified, using the node as a proof node, so that the proof node sends a verification value to the verification node;

[0063] In response to the verification value, sending a random challenge value generated by a pseudo-random number generator to the proof node through the verification node, so that the proof node sends response information to the verification node after receiving the random challenge value, the random challenge value specifically includes a first random challenge value and a second random challenge value, and the response information specifically includes first response information and second response information;

[0064] The verification node determines whether the corresponding noise data is correct based on the random challenge value, the verification value and the response information.

[0065] The verification value is specifically determined by the following formula:

[0066] ,

[0067] In the formula, To verify the value, and is a public generator, To prove the value a node has, A random number generated by the proof node. For the modulo operation, For prime numbers, the generators can be obtained in a random way and then verified. When enough elements can be generated by group operations, they are retained. If not, they are obtained by random means.

[0068] The response information is specifically determined by the following formula:

[0069] ,

[0070] ,

[0071] In the formula, For the first response information, is the second response information, is the first random challenge value, is the second random challenge value, is not equal to prime number.

[0072] The verification node determines whether the corresponding noise data is correct based on the random challenge value, the verification value and the response information. Specifically, when the following formula is established, the corresponding noise data is correct:

[0073] ,

[0074] Specifically, under special circumstances, a node may receive a verification instruction issued by an operator, or the system may issue a verification instruction to a node after determining that there is a risk. The verification instruction includes noise data and the corresponding proof node to be verified. After receiving the instruction, the verification node sends a first instruction to the proof node. The first instruction can be any specified instruction. When the proof node receives the first instruction, it generates a random number, and then generates a verification value based on the random number, and sends the verification value to the verification node. After receiving the verification value, the verification node uses a pseudo-random number generator to use cryptography to securely generate a random challenge value, which specifically includes a first random challenge value and a second random challenge value. After receiving the random challenge value, the proof node calculates response information based on the random challenge value and sends it to the verification node. The response information specifically includes first response information and second response information. Then the verification node determines whether the noise data in the proof node is correct based on the random challenge value, the verification value and the response information.

[0075] In some embodiments, the method further includes verifying the noise data to be verified when any node as a verification node receives a verification instruction to verify at least two of the noise data at the same time, and the verification process is as follows;

[0076] Sending a first instruction to all nodes corresponding to the noise data to be verified, and using the corresponding nodes as proof nodes, so that all proof nodes send verification values ​​to the verification node;

[0077] After the verification node receives the verification value, the verification node sends a random challenge value generated by a pseudo-random number generator to the proof node corresponding to the received verification value, so that the corresponding proof node sends a response message to the verification node after receiving the random challenge value;

[0078] The verification node determines whether the corresponding noise data is correct based on the random challenge value, verification value and response information corresponding to the same proof node.

[0079] Specifically, that is to say, the verification node can verify the noise data in multiple proof nodes at the same time, and send the corresponding random challenge value to each proof node after receiving the verification value sent by each proof node, so that each proof node generates a corresponding response information after receiving the corresponding random challenge value and sends it to the verification node. After receiving the corresponding response information of each proof node, the verification node determines whether the noise data of the corresponding proof node is correct through the above formula.

[0080] The present application provides an image data protection method based on distributed nodes. The method first divides the image data to be processed according to the number of nodes to obtain sub-data, and sends the obtained sub-data to each node, wherein each node receives only one sub-data; then, the sub-data is processed by the multimodal model in each node to obtain processed sub-data; then, noise is generated by each node, and the generated noise is added to the processed sub-data in the node itself to obtain noise data; finally, the noise data in the node itself is divided by each node to obtain multiple divided data and divided information, and each node also sends all the divided data and divided information in itself to all nodes except the node itself, which can realize distributed storage and calculation of image data. Each node only transmits part of the data. Even if the transmitted data is attacked, it will not cause the loss or leakage of the entire data set, thereby ensuring the overall integrity and security, and reducing the risk of leakage of sensitive information when the image data is applied in the multimodal model.

[0081] In addition, the scheme of the present application combines the privacy protection of the overall distribution by adding noise to the different processed sub-data. Not only the data privacy of a single node is protected, but also the privacy protection of the entire calculation process is ensured. Since the noise is generated and added during the calculation process, even if the calculation result is made public, the actual data cannot be recovered from it, thereby protecting the privacy of the data owner, and the noise generation includes sampling from a Gaussian distribution, standardization, and sampling a value from a gamma distribution to adjust the noise size. The multi-level noise addition mechanism provides a flexible level of privacy protection, and the size of the noise can be adjusted as needed.

[0082] At the same time, the verification mechanism allows nodes to prove the correctness of noise data without leaking any input data, protecting the privacy of the data because the verification process does not involve the transmission of any actual data.

[0083] In addition, data segmentation computing allows computing directly on segmented data fragments, which can reduce the delay caused by data transmission and improve computing efficiency. Data segmentation computing supports a wider range of computing models, including addition, subtraction, multiplication, division, and derivation, while existing technologies are usually limited to specific computing tasks. Data segmentation computing is more flexible and can adapt to more application scenarios, such as machine learning and big data analysis.

[0084] Based on the above-mentioned distributed node-based image data protection method, one or more embodiments of this specification also provide a platform and terminal for distributed node-based image data protection. The platform or terminal may include a device, software, module, plug-in, server, client, etc. that uses the method described in the embodiments of this specification and is combined with necessary implementation hardware. Based on the same innovative concept, the system in one or more embodiments provided in the embodiments of this specification is as described in the following embodiments. Since the implementation scheme and method for solving the problem of the system are similar, the implementation of the specific system in the embodiments of this specification can refer to the implementation of the aforementioned method, and the repetitions will not be repeated. The terms "unit" or "module" used below can be a combination of software and / or hardware that implements predetermined functions. Although the system described in the following embodiments is preferably implemented in software, hardware and a combination of software and hardware are also possible and conceived.

[0085] Specifically, Figure 2 It is a schematic diagram of the module structure of an embodiment of the distributed node-based image data protection device provided in this specification, which is applied to a distributed node architecture, and each node in the architecture is provided with a multimodal model, such as Figure 2 As shown, the distributed node-based image data protection device provided in this specification includes:

[0086] The segmentation module 201 is used to segment the image data to be processed according to the number of nodes to obtain sub-data, and send the obtained sub-data to each node, wherein each node only receives one sub-data;

[0087] A generating module 202, located in each node, is used to generate noise and add the noise to the processed sub-data obtained after the multimodal model in each node processes the sub-data, thereby obtaining noise data;

[0088] The sending module 203 is located in each node and is used to segment the noise data in the node itself to obtain multiple segmented data, the number of which is consistent with the number of nodes, and send the segmented data obtained by segmentation to all nodes, and each node only receives one piece of segmented data.

[0089] It should be noted that the above-mentioned system may also include other implementation modes according to the description of the corresponding method embodiment. The specific implementation modes may refer to the description of the above-mentioned corresponding method embodiment, which will not be described one by one here.

[0090] The present application also provides an electronic device, including:

[0091] processor;

[0092] a memory for storing instructions executable by the processor;

[0093] The processor is configured to execute the method provided in the above embodiment.

[0094] The electronic device provided by the embodiment of the present application stores the executable instructions of the processor in a memory. When the processor executes the executable instructions, it can first divide the image data to be processed according to the number of nodes to obtain sub-data, and send the obtained sub-data to each node, wherein each node only receives one copy of the sub-data; then the sub-data is processed by the multimodal model in each node to obtain processed sub-data; then noise is generated by each node, and the generated noise is added to the processed sub-data in the node itself to obtain noise data; finally, the noise data in the node itself is divided by each node to obtain multiple segmented data, the number of the segmented data is consistent with the number of nodes, and the segmented data obtained by the segmentation is sent to all nodes, and each node only receives one copy of the segmented data, which can realize distributed storage and calculation of image data. Each node only transmits part of the data. Even if the transmitted data is attacked, it will not cause the loss or leakage of the entire data set, thereby ensuring the overall integrity and security, and reducing the risk of leakage of sensitive information when the image data is applied in the multimodal model.

[0095] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0096] The methods or devices described in the above embodiments provided in this specification can implement business logic through computer programs and record them on storage media, and the storage media can be read and executed by computers to achieve the effects of the solutions described in the embodiments of this specification, such as:

[0097] The image data to be processed is divided into sub-data according to the number of nodes, and the sub-data obtained are sent to each node, wherein each node only receives one sub-data;

[0098] The sub-data are processed by the multimodal model in each node to obtain processed sub-data;

[0099] Generate noise through each node, and add the generated noise to the processed sub-data in the node itself to obtain noise data;

[0100] After each node divides the noise data in the node itself, a plurality of divided data are obtained, the number of the divided data is consistent with the number of nodes, and the divided data obtained by the division are sent to all nodes, and each node only receives one piece of divided data.

[0101] The storage medium may include a physical device for storing information, which is usually a medium that digitizes information and then stores it in an electrical, magnetic or optical manner. The storage medium may include: a device that stores information in an electrical energy manner, such as various memories, such as RAM, ROM, etc.; a device that stores information in a magnetic energy manner, such as a hard disk, a floppy disk, a magnetic tape, a magnetic core memory, a magnetic bubble memory, a USB flash drive; a device that stores information in an optical manner, such as a CD or a DVD. Of course, there are other readable storage media, such as quantum memory, graphene memory, etc.

[0102] The embodiments of this specification are not limited to complying with industry communication standards, standard computer resource data update and data storage rules, or the situations described in one or more embodiments of this specification. Certain industry standards or slightly modified implementation plans based on the implementation described in the custom method or embodiment can also achieve the same, equivalent or similar, or predictable implementation effects after deformation of the above-mentioned embodiments. The embodiments obtained by using these modified or deformed data acquisition, storage, judgment, processing methods, etc. can still fall within the scope of the optional implementation plans of the embodiments of this specification.

[0103] The controller can be implemented in any appropriate manner, for example, the controller can take the form of a computer-readable medium such as a microprocessor or processor and a computer-readable program code (such as software or firmware) that can be executed by the (micro)processor, a logic gate, a switch, an application specific integrated circuit (ASIC), a programmable logic controller, and an embedded microcontroller. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, ATMEL AT91SAM, MICROCHIP PIC18F26K20, and SILICONE LABS C8051F320. The memory controller can also be implemented as part of the control logic of the memory. It is also known to those skilled in the art that in addition to implementing the controller in a purely computer-readable program code manner, the controller can be implemented in the form of a logic gate, a switch, an application specific integrated circuit, a programmable logic controller, and an embedded microcontroller by logically programming the method steps. Therefore, such a controller can be considered as a hardware component, and the devices included therein for implementing various functions can also be considered as structures within the hardware component. Or even, the devices for implementing various functions can be considered as both software modules for implementing the method and structures within the hardware component.

[0104] The device embodiments described above are merely illustrative. For example, the division of the units is merely a logical function division. There may be other divisions in actual implementation, such as multiple units or plug-ins may be combined or integrated into another system, or some features may be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed may be through some interface, indirect coupling or communication connection of the device or unit, which may be electrical, mechanical or other forms.

[0105] These computer program instructions may also be loaded onto a computer or other programmable resource data updating device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions executed on the computer or other programmable device for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0106] Each embodiment in this specification is described in a progressive manner, and the same and similar parts between the embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can refer to the partial description of the method embodiment. In the description of this specification, the description of the reference terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of this specification. In this specification, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples without contradiction.

[0107] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific variations and combinations that do not deviate from the essence of the present invention based on the technical revelations disclosed by the present invention, and these variations and combinations are still within the protection scope of the present invention.

Claims

1. A distributed node-based image data protection method, characterized in that: Applied to a distributed node architecture, wherein each node in the architecture is provided with a multimodal model, the method comprises: The image data to be processed is divided into sub-data according to the number of nodes, and the sub-data obtained are sent to each node, wherein each node only receives one sub-data; The sub-data are processed by the multimodal model in each node to obtain processed sub-data; Generate noise through each node, and add the generated noise to the processed sub-data in the node itself to obtain noise data; After each node divides the noise data in the node itself, a plurality of divided data are obtained, the number of the divided data is consistent with the number of nodes, and the divided data obtained by the division are sent to all nodes, and each node only receives one piece of divided data.

2. The distributed node-based image data protection method according to claim 1, characterized in that: The method further includes verifying the noise data to be verified when any node as a verification node receives a verification instruction to verify a certain noise data, and the verification process is as follows; Sending a first instruction to a node corresponding to the noise data to be verified, using the node as a proof node, so that the proof node sends a verification value to the verification node; In response to the verification value, sending a random challenge value generated by a pseudo-random number generator to the proof node through the verification node, so that the proof node sends response information to the verification node after receiving the random challenge value, the random challenge value specifically includes a first random challenge value and a second random challenge value, and the response information specifically includes first response information and second response information; The verification node determines whether the corresponding noise data is correct based on the random challenge value, the verification value and the response information, wherein the verification node determines whether the corresponding noise data is correct based on the random challenge value, the verification value and the response information, specifically when the following formula is established, the corresponding noise data is correct: ， In the formula, and is a public generator, For the first response information, is the second response information, To verify the value, is the first random challenge value, is the second random challenge value, For the modulo operation, is a prime number.

3. The distributed node-based image data protection method according to claim 2, characterized in that: The method further includes verifying the noise data to be verified when any node as a verification node receives a verification instruction to verify at least two of the noise data at the same time, and the verification process is as follows; Sending a first instruction to all nodes corresponding to the noise data to be verified, and using the corresponding nodes as proof nodes, so that all proof nodes send verification values ​​to the verification node; After the verification node receives the verification value, the verification node sends a random challenge value generated by a pseudo-random number generator to the proof node corresponding to the received verification value, so that the corresponding proof node sends a response message to the verification node after receiving the random challenge value; The verification node determines whether the corresponding noise data is correct based on the random challenge value, verification value and response information corresponding to the same proof node.

4. The distributed node-based image data protection method according to claim 2, characterized in that: The verification value is specifically determined by the following formula: ， In the formula, To prove the value a node has, A random number generated for the proving node.

5. The distributed node-based image data protection method according to claim 4, characterized in that: The response information is specifically determined by the following formula: ， ， In the formula, is not equal to prime number.

6. The distributed node-based image data protection method according to claim 1, characterized in that: The generated noise specifically includes: Transforming the corresponding processed sub-data to generate Gaussian distribution samples; Standardizing the Gaussian distribution samples to obtain standardized samples; A sampling value is obtained by randomly sampling from a gamma distribution, and the noise is obtained by multiplying the sampling value with a standardized sample.

7. An image data protection device based on distributed nodes, characterized in that: Applied to a distributed node architecture, in which each node is provided with a multimodal model, the device comprises: A segmentation module is used to segment the image data to be processed according to the number of nodes to obtain sub-data, and send the obtained sub-data to each node, wherein each node only receives one copy of the sub-data; A generation module, located in each node, for generating noise, and adding the noise to processed sub-data obtained after the multimodal model in each node processes the sub-data, thereby obtaining noise data; The sending module is located in each node and is used to segment the noise data in the node itself to obtain multiple segmented data, the number of which is consistent with the number of nodes, and send the segmented data to all nodes, and each node only receives one piece of segmented data.

Images