An information processing method, apparatus and device

Abstract

An embodiment of this specification discloses an information processing method, apparatus, and device. The method includes: obtaining first information to be processed by a target user, where the first information includes privacy information of the target user; selecting corresponding second information for the first information, and inputting the first information and the second information into a pre-trained steganographic region model to obtain a region corresponding to the second information for performing steganographic processing on the first information, and the region for performing steganographic processing on the first information is a partial region in the region presenting the second information; inputting the first information and the region for performing steganographic processing on the first information into a pre-trained steganographic model to obtain steganographic information with the first information steganographically written in the region for performing steganographic processing on the first information; and performing business processing on a target business based on the steganographic information.

Description

Technical Field

[0001] This document relates to the field of computer technology, and particularly to an information processing method, apparatus, and device. Background Art

[0002] In recent years, information processing technology has developed rapidly. Applications for information interaction through online platforms or terminal devices have entered people's work and life, such as the face access control system in a certain community, fingerprint unlocking on mobile phones, etc. However, while information interaction through online platforms or terminal devices provides convenience for users, since the information interaction system needs to collect, transmit, process, store, etc. information, especially users' privacy information, users' privacy information is in a high-risk state. Once users' privacy information is leaked, their property and information security will be greatly threatened.

[0003] The privacy protection ability has become an important ability of the information interaction system. Generally, privacy protection processing can be carried out through information encryption. Specifically, simple linear operations are used to encrypt users' privacy information or perform operations such as row-column confusion. However, the above methods have simple atomic operations, single processes, and are easily cracked by methods such as brute force. In addition, privacy protection processing can also be carried out through deep learning, but the finally obtained information after privacy protection often has no physical meaning. In this way, it is not convenient to perform operations such as information display during the information interaction stage, resulting in certain limitations of the above methods. Based on this, an information processing solution with higher security, stronger privacy protection ability, and better displayability is needed. Summary of the Invention

[0004] The purpose of the embodiments of this specification is to provide an information processing solution with higher security, stronger privacy protection ability, and better displayability.

[0005] To achieve the above technical solution, the embodiments of this specification are implemented as follows:

[0006] An information processing method provided by an embodiment of this specification, the method includes: obtaining first information to be processed by a target user, where the first information includes privacy information of the target user. Selecting corresponding second information for the first information, and inputting the first information and the second information into a pre-trained steganographic area model to obtain an area corresponding to the second information for steganographically processing the first information, and the area for steganographically processing the first information is a partial area in the area presenting the second information, and the steganographic area model is used to determine a partial area for steganographically processing another information in the presentation area of one information. Inputting the first information and the area for steganographically processing the first information into a pre-trained steganographic model to obtain steganographic information with the first information steganographically written in the area for steganographically processing the first information, and the steganographic model is used to steganographically write one information into a partial area of the area for presenting another information. Performing business processing on a target business based on the steganographic information.

[0007] An information processing apparatus provided by an embodiment of this specification, the apparatus includes: an information acquisition module, which acquires first information to be processed by a target user, where the first information includes privacy information of the target user. A region determination module, which selects corresponding second information for the first information, and inputs the first information and the second information into a pre-trained steganographic region model to obtain a region corresponding to the second information for steganographically processing the first information, and the region for steganographically processing the first information is a partial region in the region presenting the second information, and the steganographic region model is used to determine a partial region for steganographically processing another information in the presentation region of one information. A steganographic module, which inputs the first information and the region for steganographically processing the first information into a pre-trained steganographic model to obtain steganographic information with the first information steganographically written in the region for steganographically processing the first information, and the steganographic model is used to steganographically write one information into a partial region of the region for presenting another information. A processing module, which performs business processing on a target business based on the steganographic information.

[0008] An information processing device provided by an embodiment of this specification, the information processing device includes: a processor; and a memory arranged to store computer-executable instructions, the executable instructions, when executed, cause the processor to: obtain first information to be processed by a target user, where the first information includes privacy information of the target user. Select corresponding second information for the first information, and input the first information and the second information into a pre-trained steganographic area model to obtain an area corresponding to the second information for performing steganographic processing on the first information, the area for performing steganographic processing on the first information is a partial area in the area presenting the second information, and the steganographic area model is used to determine a partial area for performing steganographic processing on another information in the presentation area of one information. Input the first information and the area for performing steganographic processing on the first information into a pre-trained steganographic model to obtain steganographic information with the first information steganographically written in the area for performing steganographic processing on the first information, and the steganographic model is used to steganographically write one information into a partial area of the area for presenting another information. Perform business processing on a target business based on the steganographic information.

[0009] An embodiment of this specification also provides a storage medium, the storage medium is used to store computer-executable instructions, and the executable instructions, when executed by a processor, implement the following process: obtain first information to be processed by a target user, where the first information includes privacy information of the target user. Select corresponding second information for the first information, and input the first information and the second information into a pre-trained steganographic area model to obtain an area corresponding to the second information for performing steganographic processing on the first information, the area for performing steganographic processing on the first information is a partial area in the area presenting the second information, and the steganographic area model is used to determine a partial area for performing steganographic processing on another information in the presentation area of one information. Input the first information and the area for performing steganographic processing on the first information into a pre-trained steganographic model to obtain steganographic information with the first information steganographically written in the area for performing steganographic processing on the first information, and the steganographic model is used to steganographically write one information into a partial area of the area for presenting another information. Perform business processing on a target business based on the steganographic information. Description of the Drawings

[0010] In order to more clearly illustrate the technical solutions in the embodiments of this specification or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments recorded in this specification. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0011] Figure 1 This is an embodiment of an information processing method in this specification;

[0012] Figure 2 This is another embodiment of an information processing method in this specification;

[0013] Figure 3 This is yet another embodiment of an information processing method in this specification;

[0014] Figure 4 This is a schematic structural diagram of an information processing system in this specification;

[0015] Figure 5 This is yet another embodiment of an information processing method in this specification;

[0016] Figure 6 This is an embodiment of an information processing device in this specification;

[0017] Figure 7 This is an embodiment of an information processing device in this specification. Detailed implementation manners

[0018] Embodiments of this specification provide an information processing method, device, and device.

[0019] In order to enable those skilled in the art to better understand the technical solutions in this specification, the following will clearly and completely describe the technical solutions in the embodiments of this specification in conjunction with the accompanying drawings in the embodiments of this specification. Obviously, the described embodiments are only a part of the embodiments of this specification, rather than all the embodiments. Based on the embodiments in this specification, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of this specification.

[0020] Embodiment 1

[0021] As Figure 1 shown, embodiments of this specification provide an information processing method. The execution subject of this method can be a terminal device or a server. Among them, the terminal device can be a certain terminal device such as a mobile phone or a tablet computer, or a computer device such as a notebook computer or a desktop computer, or can also be an IoT device (specifically such as a smart watch, a vehicle-mounted device, etc.). The server can be an independent server, or can also be a server cluster composed of multiple servers. The server can be a background server such as a financial service or an online shopping service, or can also be a background server of a certain application program, etc. This method can specifically include the following steps:

[0022] In step S102, the first information to be processed by the target user is obtained, and the privacy information of the target user is included in the first information.

[0023] Among them, the target user can be any user, such as the owner of the above terminal device. The target user can initiate an information processing request through the terminal device. The privacy information of the user can include various types. For example, the user's name, the number of the certificate proving the user's identity, the residential address, the mobile phone number, the user's biological information (specifically, such as the user's fingerprint information, facial information, etc.). Specifically, it can be set according to the actual situation, and the embodiments of this specification do not limit this.

[0024] In practice, in recent years, information processing technology has developed rapidly. Applications for information interaction through online platforms or terminal devices have entered people's work and life. For example, the face access control system in a certain community, the fingerprint unlocking of mobile phones, etc. However, while providing convenience for users through information interaction via online platforms or terminal devices, since the information interaction system needs to collect, transmit, process, store, etc. information, especially the privacy information of users, the privacy information of users is in a high-risk state. Once the privacy information of users is leaked, their property and information security will be greatly threatened.

[0025] The privacy protection ability has become an important ability of the information interaction system. Generally, privacy protection processing can be performed through information encryption. Specifically, simple linear operations are used to encrypt the privacy information of users or perform operations such as row-column confusion. However, the above methods have simple atomic operations, a single process, and are easily cracked by methods such as brute force. In addition, privacy protection processing can also be performed through deep learning. Specifically, a deep learning model (such as a neural network model, etc.) is trained and used to perform privacy protection processing on the privacy information of users to obtain the privacy-protected information. However, the finally obtained privacy-protected information often has no physical meaning. In this way, it is not convenient to perform operations such as information display during the information interaction stage, resulting in certain limitations of the above methods. Based on this, an information processing solution with higher security, stronger privacy protection ability, and better displayability is required. The embodiments of this specification provide an implementable technical solution, which can specifically include the following content:

[0026] When a user (i.e., the target user) needs to execute a specified service (such as a payment service, a login service, etc.), the execution mechanism of the above-specified service can be triggered. During the execution of the above-specified service, relevant information of the target user is often obtained. For example, the number of the certificate proving the user's identity, the residential address, the mobile phone number, the user's biometric information, etc. of the target user can be obtained. The relevant information obtained above can be analyzed to determine whether it contains the target user's privacy information. If it includes the target user's privacy information, the above information can be obtained and the obtained information can be used as the first information.

[0027] For example, before executing the above-specified service, it is often necessary to identify the identity of the user. At this time, the terminal device can activate the corresponding information collection component (such as a camera component) and can collect the user's biometric information through the information collection component (specifically, for example, the facial image of the user can be collected), so that the terminal device can obtain the first information including the user's biometric information.

[0028] In step S104, the corresponding second information is selected for the first information, and the first information and the second information are input into a pre-trained steganographic region model to obtain the region corresponding to the second information for steganographically processing the first information. The region for steganographically processing the first information is a partial region in the region presenting the second information. The steganographic region model is used to determine the partial region for steganographically processing another information in the presenting region of one information.

[0029] Among them, the steganographic region model can be a model for determining the partial content of another information into which a certain information needs to be steganographically written after steganographically processing the information. The steganographic region model can be constructed by a variety of different algorithms. For example, the steganographic region model can be constructed by a neural network algorithm, or the steganographic region model can be constructed by a random forest algorithm, etc. It can be specifically set according to the actual situation, and the embodiments of this specification do not limit this. The second information can be any information or information associated with the first information. For example, the second information can be the facial photo taken by a certain user, and the first information can be the image of the cartoon character of this user, etc. It can be specifically set according to the actual situation, and the embodiments of this specification do not limit this. The region presenting the second information can be the region for displaying or presenting the content of the second information. For example, if the second information is the facial information of a certain user, the region presenting the second information can be the region where the image containing this facial information is located.

[0030] In implementation, an initial architecture of a steganographic region model can be constructed through a preset algorithm. Then, first training sample data composed of privacy information of different users can be obtained, and second training sample data can be selected (wherein, the first training sample data and the second training sample data can be related or unrelated), and the above first training sample data and second training sample data can be used as the training sample data of the model. Then, a corresponding loss function can be set, and the obtained first training sample data, second training sample data, and the loss function can be used. At the same time, some regions can be randomly selected in the region where the second training sample data is presented, and then the steganographic region model can be trained to obtain a trained steganographic region model. Through the trained steganographic region model, a region for performing steganographic processing on the first training sample data (i.e., some regions in the region where the second training sample data is presented) can be selected.

[0031] When the first information containing the privacy information of the target user is obtained, the first information can be analyzed. Based on the obtained analysis result, corresponding second information can be selected for the first information. Then, the above first information and second information can be input into the trained steganographic region model. Through the steganographic region model, some regions for performing steganographic processing on the first information can be selected from the region where the second information is presented, that is, subsequently, the first information can be steganographically written into the selected some regions to perform privacy protection processing on the first information, so that the user's privacy information (such as fingerprint pattern information, clear facial contour information, etc.) can be written into the selected some regions in a hidden manner.

[0032] In step S106, the first information and the region for performing steganographic processing on the first information are input into a pre-trained steganographic model to obtain steganographic information in which the first information is steganographically written in the region for performing steganographic processing on the first information. The steganographic model is used to steganographically write one information into some regions of the region for presenting another information.

[0033] Among them, the steganographic model can be a model for steganographically writing one information into some regions of the region for presenting another information through information steganography. The steganographic model can be constructed through a variety of different algorithms. For example, the steganographic model can be constructed through a neural network algorithm, or the steganographic model can be constructed through a HUGO (Highly Undetectable stego) algorithm, etc. It can be specifically set according to the actual situation, and the embodiments of this specification do not limit this. The region for performing steganographic processing on the first information can be some regions in the region where the second information is presented. For example, if the second information is an image, the region for performing steganographic processing on the first information can be the region corresponding to some images in the above entire image.

[0034] In implementation, an initial architecture of a steganography model can be constructed through a preset algorithm. Then, first training sample data composed of privacy information of different users can be obtained, and second training sample data can be selected (wherein the first training sample data and the second training sample data can be related or unrelated), and some regions can be randomly selected in the region presenting the second training sample data. And through a preset corresponding loss function, the steganography model is trained to obtain a trained steganography model. Through the trained steganography model, the first training sample data can be written into the selected partial regions (i.e., partial regions in the region presenting the second training sample data) in a way of information hiding.

[0035] When the first information and the region for steganographically processing the first information are obtained, the above first information and the region for steganographically processing the first information can be input into the above trained steganography model. Through the steganography model, the first information is written into partial regions in the region presenting the second information in a way of information hiding, so as to realize privacy protection processing for the first information, so that the user's privacy information (such as information on fingerprint patterns, clear contour information of the face, etc.) is written into the selected partial regions in a hidden manner.

[0036] In step S108, business processing is performed on the target business based on the above steganographic information.

[0037] In implementation, steganographic information can be used for subsequent business processing. Specifically, for example, if the target business is a payment business, when performing the payment business, the identity of the user needs to be identified. At this time, the steganographic information (i.e., the user's biometric information after privacy protection, that is, the information obtained after the first information is steganographically written into the region for steganographically processing the first information) can be used to calculate the similarity with the reference user biometric information pre-stored locally (or on the server) (which may not include sensitive information, that is, still the user's biometric information after privacy protection). If the obtained similarity value is greater than the preset similarity threshold, the result of biometric identification of the target user is passed. At this time, the target business (i.e., the payment business) can continue to be processed. If the obtained similarity value is less than the preset similarity threshold, the result of biometric identification of the target user is failed, and the processing of the target business is terminated. In practical applications, the above processing process is only an optional method and can also include various different processing methods, which can be specifically set according to the actual situation.

[0038] An embodiment of this specification provides an information processing method. By obtaining first information to be processed by a target user, where the first information includes the privacy information of the target user, then, corresponding second information is selected for the first information, and the first information and the second information are input into a pre-trained steganographic region model to obtain a region corresponding to the second information for performing steganographic processing on the first information. The region for performing steganographic processing on the first information is a partial region in the region presenting the second information. The first information and the region for performing steganographic processing on the first information are input into a pre-trained steganographic model to obtain steganographic information with the first information steganographically embedded in the region for performing steganographic processing on the first information. Based on the steganographic information, business processing is performed on the target business. In this way, through steganography technology, the user's privacy information is steganographically embedded in the second information that is convenient for display (such as the user's cartoonified avatar, etc.). This not only desensitizes the user's privacy information but also facilitates information display. Additionally, in terms of steganography technology, different from traditional steganographic processing on the entire information, this solution proposes a local-global steganographic method, that is, a local region capable of performing steganographic processing is retrieved through a corresponding model for steganographic processing. On the one hand, the quality of steganography is improved, and on the other hand, the attacker cannot determine in which region the steganographic processing is performed, so that the privacy information is difficult to be leaked, enhancing the security of steganography.

[0039] Embodiment 2

[0040] As Figure 2 As shown, an embodiment of this specification provides an information processing method. The execution subject of this method can be a terminal device or a server. Among them, the terminal device can be a certain terminal device such as a mobile phone, a tablet computer, etc., or a computer device such as a laptop computer or a desktop computer, or it can also be an IoT device (specifically such as a smart watch, a vehicle-mounted device, etc.). The server can be an independent server or a server cluster composed of multiple servers. The server can be a background server for financial services or online shopping services, etc., or a background server for a certain application program, etc. The method can specifically include the following steps:

[0041] In step S202, first training sample data and a partial region in the region presenting second training sample data are obtained, and the first training sample data includes the privacy information of the user.

[0042] Among them, there can be multiple pieces of first training sample data. The multiple pieces of first training sample data can be composed of the privacy information of the same user, or can be composed of the privacy information of multiple different users, which can be specifically set according to the actual situation. There can be multiple pieces of second training sample data. The number of second training sample data can be the same as the number of first training sample data, or the number of second training sample data can be less than the number of first training sample data. The area presenting the second training sample data can be, for example, the area of a partial image in the image where the second training sample data is located.

[0043] In implementation, with the consent of the user, the privacy information of the user can be obtained from multiple different users, and the obtained privacy information of the user can be used as the first training sample data. Or, the privacy information of the user can be obtained from a specified database, and the obtained privacy information of the user can be used as the first training sample data, etc., which can be specifically set according to the actual situation. In addition, the second training sample data can be obtained according to the actual situation. The second training sample data can be related to the first training sample data or can be unrelated to the first training sample data. For example, with the consent of the user, the user can provide the second training sample data related to the first training sample data, or the second training sample data can be randomly selected from a specified database, etc., which can be specifically set according to the actual situation, and the embodiments of this specification do not make limitations in this regard.

[0044] The area presenting the second training sample data can be determined. For example, if the second training sample data is image data, the area presenting the second training sample data can be the area where the image presenting the second training sample data is located. Another example is that if the second training sample data is text data, the area presenting the second training sample data can be the area of the document content presenting or displaying the second training sample data, etc., which can be specifically set according to the actual situation. Then, the area presenting the second training sample data can be randomly divided according to the actual situation to obtain multiple partial areas, and one partial area can be randomly selected from the multiple partial areas as the partial area in the area presenting the obtained second training sample data.

[0045] In step S204, the steganography model, the decoding model, and the adversarial model are jointly trained with the first training sample data and a partial area in the area presenting the second training sample data to obtain the trained steganography model, the trained decoding model, and the trained adversarial model. The decoding model is used to perform restoration processing on the first training sample data after steganography processing. The adversarial model is used to determine whether the first training sample data is steganographically embedded in two partial areas in the area presenting the second training sample data, and one of the two partial areas is the area where the first training sample data is not steganographically embedded.

[0046] Among them, the first loss function can be determined in a variety of different ways. For example, corresponding loss functions can be set based on the steganography model, the decoding model, and the adversarial model respectively. Or a loss function corresponding to the input data and the final output data can also be set, etc. Alternatively, a suitable loss function can be set for the above joint training according to the actual situation, which can be specifically set according to the actual situation, and the embodiments of this specification do not limit this.

[0047] In implementation, the first training sample data and a partial area in the area presenting the second training sample data can be respectively input into the steganography model to obtain output data (that is, the first training sample data after steganography processing, which is also the data obtained after writing the first training sample data into a partial area in the area presenting the second training sample data through information steganography). The decoding model can be used to perform restoration processing on the above output data, and the adversarial model can be used to determine whether the first training sample data is steganographically written into two partial areas in the area presenting the second training sample data. Then, the corresponding loss value can be calculated through the first loss function, and it can be determined whether the above steganography model, decoding model, and adversarial model converge based on the calculated loss value. If they converge, the trained steganography model, the trained decoding model, and the trained adversarial model are obtained. If they do not converge, the steganography model, the decoding model, and the adversarial model are continuously trained based on the training sample data (that is, the first training sample data and a partial area in the area presenting the second training sample data) until the steganography model, the decoding model, and the adversarial model converge, and the trained steganography model, the trained decoding model, and the trained adversarial model are obtained.

[0048] The specific processing method of the above step S204 can be various. The following provides an optional processing method, which can specifically include the processing of steps A2 to A8.

[0049] In step A2, the first training sample data and a partial area in the area presenting the second training sample data are input into the steganography model to obtain the first training sample data after steganography processing.

[0050] In step A4, the first training sample data after steganography processing is input into the decoding model to perform restoration processing on the first training sample data after steganography processing through the decoding model to obtain the reconstructed first training sample data.

[0051] In step A6, two partial areas in the area presenting the second training sample data are input into the adversarial model to determine the probability that the first training sample data is steganographically written into each partial area in the two partial areas in the area presenting the second training sample data through the adversarial model, and the corresponding output result is obtained.

[0052] In step A8, based on the first training sample data, the first training sample data after steganography processing, the reconstructed first training sample data, a partial area in the area presenting the second training sample data, the above output result, and a preset first loss function, determine whether the steganography model, the decoding model, and the adversarial model converge. If not, obtain the first training sample data and a partial area in the area presenting the second training sample data, and continue to perform model training on the steganography model, the decoding model, and the adversarial model until the steganography model, the decoding model, and the adversarial model converge, obtaining the trained steganography model, the trained decoding model, and the trained adversarial model.

[0053] Among them, the first loss function is determined by the minimum value of the difference in peak signal-to-noise ratio (PSNR) of a partial region in the region presenting the second training sample data before and after steganography processing, the maximum value of the similarity between the first training sample data and the reconstructed first training sample data, and a preset classification sub-loss function. Specifically, it is like Lt = L1(A, At) + L2(B, Br) + L3(p, y), where A represents the first training sample data, At represents the first training sample data after steganography processing, B represents the first training sample data after steganography processing, Br represents the reconstructed first training sample data, Lt represents the first loss function corresponding to the first training sample data, p and y respectively represent the probabilities of embedding the first training sample data in each partial region of the two partial regions. L1(A, At) ensures the privacy protection effect and the displayable effect, so that the PSNR of the partial region in the region presenting the second training sample data where the first training sample data is embedded remains at a relatively high level, that is, the PSNR of the partial region in the region presenting the second training sample data before and after steganography processing is basically the same (i.e., the difference between the two is less than the preset threshold). L2(B, Br) ensures that the first training sample data after steganography processing can be better restored to the original first training sample data. L3(p, y) can be an adversarial binary classification sub-loss function, which is used to distinguish whether the second training sample data contains information written in the way of information hiding (this adversarial training method can make the steganography processing more secure). The steganography model and the decoding model can be constructed based on a variety of different methods. For example, they can be constructed based on U-Net. This U-Net is constructed by a fully connected network. The U-Net presents a structure similar to the letter "U". It consists of a contracting path on the left half and an expansive path on the right half. The contracting path can be constructed by a convolutional neural network. The structure of 2 convolutional layers and 1 max pooling layer can be repeatedly used. After each pooling operation, the dimension of the data will increase.In the extended channel, first perform one deconvolution operation to halve the dimension of the data. Then, splice and crop it corresponding to the compressed channel to obtain the corresponding feature data. Recompose the new feature data based on the above feature data, and then use two convolutional layers for feature extraction and repeat the above structure. In the final output layer, use two convolutional layers to map the high-dimensional feature data into low-dimensional output data. U-Net can be specifically divided into two parts: upsampling and downsampling. The downsampling part mainly uses continuous convolutional pooling layers to extract the feature information in the data and gradually maps the feature information to a high dimension. There is rich feature information in the whole data at the highest dimension of the entire network. U-Net does not need to directly pool this data and directly upsample it to the output data with the same size as the original data. Instead, through deconvolution processing, the high-dimensional features are mapped to a low dimension again. During the mapping process, to enhance the segmentation accuracy, the data with the same dimension in the shrinking network at the same dimension will be fused. Since the dimension will become twice the original dimension during the fusion process, convolution processing needs to be performed again to ensure that the dimension after processing is the same as the dimension before the fusion operation, so that after another deconvolution processing, it can be secondarily fused with the data at the same dimension until the output data has the same dimension as the original data. The structures of the steganography model and the decoding model in this embodiment can be composed of a U-Net with a certain number of network layers. Specifically, for example, it can be composed of a U-Net with 8 or 10 network layers, etc., and can be specifically set according to the actual situation. For another example, it can be constructed through a multi-layer perceptron MLP. In the MLP, in addition to the input layer and the output layer, there can be multiple hidden layers in the middle. The simplest MLP only contains one hidden layer, that is, a three-layer structure. The layers of the MLP are fully connected. The bottom layer of the MLP is the input layer, the middle is the hidden layer, and the last is the output layer. The steganography model and the decoding model can be specifically constructed through a three-layer MLP, and can be specifically set according to the actual situation. The adversarial model can be constructed through a specified classification algorithm, specifically such as a binary classification algorithm, etc., and an appropriate classification algorithm or binary classification algorithm, etc., can be specifically selected according to the actual situation.

[0054] In implementation, the first training sample data and a partial area in the area presenting the second training sample data can be obtained, and the first training sample data and the partial area in the area presenting the second training sample data can be input into an encoding model to obtain the first training sample data after steganography processing. The first training sample data after steganography processing can be input into a decoding model to obtain the reconstructed first training sample data. Meanwhile, two partial areas in the area presenting the second training sample data can be input into an adversarial model to obtain the probability of steganographically writing the first training sample data into each of the two partial areas. Among them, the decoding model can also be constructed based on U-Net or based on MLP. The input data of the decoding model is the first training sample data after steganography processing, and the output data is the reconstructed first training sample data. Then, based on the first training sample data, the first training sample data after steganography processing, the reconstructed first training sample data, the partial area in the area presenting the second training sample data, the above output result, and a preset first loss function, it can be determined whether the steganography model, the decoding model, and the adversarial model converge. If not, the first training sample data and the partial area in the area presenting the second training sample data are obtained to continue the model training of the steganography model, the decoding model, and the adversarial model until the steganography model, the decoding model, and the adversarial model converge, and the trained steganography model, the trained decoding model, and the trained adversarial model are obtained.

[0055] The purpose of the above processing process is to train a model that can steganographically write privacy information into a specified area of another piece of information. Additionally, to improve the performance of steganography and provide a Reward function for subsequent reinforcement learning.

[0056] In step S206, based on the first training sample data and the second training sample data, through a preset area search strategy corresponding to the steganography area model, a partial area in the area presenting the second training sample data that meets the preset conditions for steganographically processing the first training sample data is determined. And by using a preset second loss function and the determined partial area that meets the preset conditions for steganographically processing the first training sample data, it is determined whether the steganography area model converges. If not, the first training sample data and the second training sample data are obtained to continue the model training of the steganography area model until the steganography area model converges, and the trained steganography area model is obtained.

[0057] Among them, the second loss function can be determined in a variety of different ways. For example, corresponding loss functions can be set based on the steganography model, the decoding model, and the adversarial model respectively, or a loss function corresponding to the input data and the final output data can also be set, etc. And the second loss function can be constructed through the above loss functions. Or, the second loss function can also be constructed based on the first loss function, which can be specifically set according to the actual situation. Or, appropriate loss functions can also be set for the above models according to the actual situation, which can be specifically set according to the actual situation. The embodiments of this specification do not make any limitations in this regard. In practical applications, the setting methods of the second loss function can include various types. The following provides an optional processing method, which can specifically include the following content: The second loss function is negatively correlated with the first loss function, that is, the smaller the loss value corresponding to the first loss function, the larger the loss value corresponding to the second loss function, and the larger the loss value corresponding to the first loss function, the smaller the loss value corresponding to the second loss function. The region search strategy can include various types. For example, moving 5 unit lengths to the upper left, moving 2 unit lengths to the right, moving 3 unit lengths downward, etc., which can be specifically set according to the actual situation. The embodiments of this specification do not make any limitations in this regard. In this embodiment, the region search strategy can also be constructed by the search direction and / or the translation step size. The search direction among them can include one or more of the following: translating upward, translating downward, translating leftward, and translating rightward. The partial region in the region presenting the second training sample data that meets the preset conditions for steganographically processing the first training sample data can be the partial region in the region presenting the second training sample data that is suitable for steganographic processing.

[0058] In implementation, the structure of the steganographic region model can adopt the network model structure of DQN for reinforcement learning, and the reinforcement learning can be optimized by means of gradient descent. The input data of the steganographic region model is the first training sample data and the second training sample data, and the output data is the partial region suitable for steganography processing in the region presenting the second training sample data. In addition, the search process of the steganographic region model (i.e., the network model of DQN) can be motivated by the Reward function, and the steganographic region model is used to judge which regions in the region presenting the second training sample data are better regions suitable for steganography processing, and the next decision adjustment can be determined based on this. The Reward function among them can be used as the second loss function corresponding to the steganographic region model, and the Reward function can be negatively correlated with the first loss function. For the region search strategy, for example, the search direction can be [translation up, translation down, translation left, translation right], and the translation step size (or translation intensity) can be [1, 2, 4, 8]. Then the region search strategy can be 16 different situations composed of the above search directions and translation step sizes. The model training process of the steganographic region model can be: through the above network model structure of DQN combined with the region search strategy and the Reward function, the steganographic region model is trained until the Reward no longer improves, and finally the trained steganographic region model can be obtained.

[0059] Among them, DQN refers to the Q-learning algorithm based on deep learning, which combines value function approximation and neural network technology, and adopts the ways of target network and experience replay for network training. In Q-learning, a table is used to store the Reward function of the action in each state, that is, the state-action value function Q(s, a). However, in actual tasks, the number of state quantities is usually huge, and in continuous tasks, the problem of dimensionality disaster will be encountered. Therefore, it is usually impractical to use the real Value Function, so the representation method of value function approximation is used. DQN can be specifically executed according to the actual situation and will not be elaborated here. The ways of gradient descent can include various ones, such as the stochastic gradient descent SGD algorithm, the gradient descent algorithm, etc., and can be specifically set according to the actual situation.

[0060] In step S208, a biometric request of the target user is obtained. The biometric request includes the first information to be processed, and the first information includes the privacy information of the target user. The first information is the user biometric information for biometric identification.

[0061] Among them, the user's biometric information can include various types, such as the user's fingerprint information, palmprint information, facial information, or iris information, etc. In practical applications, the carriers of the above-mentioned user biometric information can include various types. For example, the above-mentioned various user biometric information can be carried in the form of images, etc., which can be specifically set according to the actual situation, and the embodiments of this specification do not limit this.

[0062] In step S210, corresponding second information is selected for the first information, and the first information and the second information are input into a pre-trained steganographic region model to obtain the region corresponding to the second information for steganographically processing the first information.

[0063] In step S212, the first information and the region for steganographically processing the first information are input into a pre-trained steganographic model to obtain steganographic information with the first information steganographically written in the region for steganographically processing the first information.

[0064] In step S214, biometric recognition processing is performed on the target user based on the above steganographic information.

[0065] In step S216, the first information is deleted.

[0066] The embodiments of this specification provide an information processing method. By obtaining the first information to be processed of the target user, where the first information includes the privacy information of the target user, then, corresponding second information is selected for the first information, and the first information and the second information are input into a pre-trained steganographic region model to obtain the region corresponding to the second information for steganographically processing the first information. The region for steganographically processing the first information is a partial region in the region presenting the second information. The first information and the region for steganographically processing the first information are input into a pre-trained steganographic model to obtain steganographic information with the first information steganographically written in the region for steganographically processing the first information. Business processing is performed on the target service based on the steganographic information. In this way, through steganography technology, the user's privacy information is steganographically written into the second information that is convenient for display (such as the user's cartoonized avatar, etc.). It not only desensitizes the user's privacy information but also facilitates the display of information. In addition, in terms of steganography technology, different from the traditional steganographic processing of the entire information, this solution proposes a local-global steganographic method, that is, a local region capable of steganographic processing is retrieved through a corresponding model for steganographic processing. On the one hand, the quality of steganography is improved, and on the other hand, the attacker cannot determine which region the steganography is performed on, so that the privacy information is difficult to be leaked, and the security of steganography is improved.

[0067] Embodiment III

[0068] Such as Figure 3As shown in the figure, an embodiment of this specification provides an information processing method, which can be jointly executed by a terminal device and a server. Among them, the terminal device can be certain terminal devices such as mobile phones and tablet computers, or computer devices such as laptop computers or desktop computers. Alternatively, it can also be an IoT device (specifically, such as smart watches, in-vehicle devices, etc.). The server can be an independent server, or a server cluster composed of multiple servers. The server can be a background server for financial services or online shopping services, or a background server for a certain application program, etc. Its system architecture can be as Figure 4 shown, and the method specifically may include the following steps:

[0069] In step S302, the server obtains the first training sample data and a partial area in the area presenting the second training sample data, and the first training sample data includes the user's privacy information.

[0070] In step S304, the server jointly trains the steganography model, the decoding model, and the adversarial model through the first training sample data and a partial area in the area presenting the second training sample data, to obtain the trained steganography model, the trained decoding model, and the trained adversarial model. The decoding model is used to restore the steganographically processed first training sample data, and the adversarial model is used to determine whether the first training sample data is steganographically written in two partial areas in the area presenting the second training sample data. One of the two partial areas is the area where the first training sample data is not steganographically written.

[0071] The specific processing method of the above step S304 can be various. The following provides an optional processing method, which specifically may include the processing from step B2 to step B8.

[0072] In step B2, the server inputs the first training sample data and a partial area in the area presenting the second training sample data into the steganography model to obtain the steganographically processed first training sample data.

[0073] In step B4, the server inputs the steganographically processed first training sample data into the decoding model to restore the steganographically processed first training sample data through the decoding model, to obtain the reconstructed first training sample data.

[0074] In step B6, the server inputs the two partial areas in the area presenting the second training sample data into the adversarial model to determine the probability of steganographically writing the first training sample data in each of the two partial areas in the area presenting the second training sample data through the adversarial model, to obtain the corresponding output result.

[0075] In step B8, the server determines whether the steganography model, the decoding model, and the adversarial model converge based on the first training sample data, the first training sample data after steganography processing, the reconstructed first training sample data, a partial area in the area presenting the second training sample data, the above output result, and a preset first loss function. If not, it obtains the first training sample data and a partial area in the area presenting the second training sample data to continue training the steganography model, the decoding model, and the adversarial model until the steganography model, the decoding model, and the adversarial model converge, obtaining the trained steganography model, the trained decoding model, and the trained adversarial model.

[0076] Among them, the first loss function is determined by the minimum value of the difference in peak signal-to-noise ratio (PSNR) of a partial area in the area presenting the second training sample data before and after steganography processing, the maximum value of the similarity between the first training sample data and the reconstructed first training sample data, and a preset classification sub-loss function.

[0077] In step S306, the server determines a partial area in the area presenting the second training sample data that meets the preset conditions for steganography processing of the first training sample data based on the first training sample data and the second training sample data through a preset area search strategy corresponding to the steganography area model, and uses a preset second loss function and the determined partial area that meets the preset conditions for steganography processing of the first training sample data to determine whether the steganography area model converges. If not, it obtains the first training sample data and the second training sample data to continue training the steganography area model until the steganography area model converges, obtaining the trained steganography area model.

[0078] Among them, the second loss function is negatively correlated with the first loss function. The area search strategy can also be constructed by a search direction and / or a translation step size, and the search direction can include one or more of the following: upward translation, downward translation, leftward translation, and rightward translation.

[0079] In step S308, the server sends the trained steganography area model and the trained encoding model to the terminal device.

[0080] In step S310, the terminal device obtains a biometric request of the target user, and the biometric request includes the first information to be processed. The first information includes the privacy information of the target user, and the first information is the user biometric information for biometric identification.

[0081] In step S312, the terminal device selects corresponding second information for the first information and inputs the first information and the second information into the pre-trained steganography area model to obtain the area corresponding to the second information for steganography processing of the first information.

[0082] In step S314, the terminal device inputs the first information and the area for steganography processing of the first information into a pre-trained steganography model, and obtains steganography information in which the first information is steganographically written in the area for steganography processing of the first information.

[0083] In step S316, the terminal device sends the above steganography information to the server, and the steganography information is used to trigger the server to perform biometric processing on the target user based on the pre-stored benchmark user biometric information and the steganography information.

[0084] In step S318, the terminal device receives the biometric result of the biometric processing of the target user sent by the server.

[0085] In step S320, the terminal device deletes the first information.

[0086] For the specific processing procedures of the above steps S302 to S320, reference may be made to the above relevant content, which will not be elaborated here.

[0087] An information processing method is provided in an embodiment of this specification. By obtaining the first information to be processed of the target user, where the first information includes the privacy information of the target user, then, a corresponding second information is selected for the first information, and the first information and the second information are input into a pre-trained steganography area model to obtain the area corresponding to the second information for steganography processing of the first information. The area for steganography processing of the first information is a partial area in the area presenting the second information. The first information and the area for steganography processing of the first information are input into a pre-trained steganography model to obtain steganography information in which the first information is steganographically written in the area for steganography processing of the first information. Business processing is performed on the target service based on the steganography information. In this way, through steganography technology, the user's privacy information is steganographically written into the second information that is convenient for display (such as the user's cartoonized avatar, etc.). It not only desensitizes the user's privacy information but also facilitates the display of information. In addition, in terms of steganography technology, different from the traditional steganography processing of the entire information, this solution proposes a local-global steganography method, that is, a local area capable of steganography processing is retrieved through a corresponding model for steganography processing. On the one hand, the quality of steganography is improved, and on the other hand, the attacker cannot determine in which area the steganography processing is performed, so that the privacy information is difficult to be leaked, and the security of steganography is improved.

[0088] Embodiment 4

[0089] This embodiment will elaborate in detail on an information processing method provided by an embodiment of the present invention in combination with a specific application scenario. The corresponding application scenario is an application scenario of biometric recognition (such as face recognition, etc.). Among them, the first training sample data, the second training sample data, the first information, the second information, etc. are all images.

[0090] As Figure 5 shown, the execution subject of this method can be a terminal device and a server. Among them, the terminal device can be a certain terminal device such as a mobile phone, a tablet computer, etc., or a computer device such as a notebook computer or a desktop computer. Or, it can also be an IoT device (specifically such as a smart watch, a vehicle-mounted device, etc.). The server can be an independent server or a server cluster composed of multiple servers. The server can be a background server such as a financial service or an online shopping service, or a background server of a certain application program, etc. This method can specifically include the following steps:

[0091] In step S502, the server obtains the first training sample data and a partial area in the area presenting the second training sample data. The first training sample data includes data of a first sample image of the user's biometric information, the second training sample data is data of a second sample image, and the area presenting the second training sample data is the area of the second sample image.

[0092] In step S504, the server inputs the data of the first sample image and a partial area in the area of the second sample image into a steganography model to obtain a steganographically processed first sample image (i.e., an image obtained by writing the first sample image into a partial area in the area of the second sample image in a way of information hiding).

[0093] In step S506, the server inputs the data of the steganographically processed first sample image into a decoding model to perform a restoration process on the steganographically processed first sample image through the decoding model to obtain a reconstructed first sample image.

[0094] In step S508, the server inputs two partial areas in the area of the second sample image into an adversarial model to determine, through the adversarial model, the probability of the first sample image being steganographically written into each of the two partial areas in the area of the second sample image, and obtains a corresponding output result.

[0095] In step S510, the server determines whether the steganography model, the decoding model, and the adversarial model converge based on the first sample image, the first sample image after steganography processing, the reconstructed first sample image, partial regions in the regions of the second sample image, the above output result, and a preset first loss function. If not, it obtains partial regions in the regions of the first sample image and the second sample image and continues to perform model training on the steganography model, the decoding model, and the adversarial model until the steganography model, the decoding model, and the adversarial model converge, obtaining the trained steganography model, the trained decoding model, and the trained adversarial model.

[0096] Among them, the first loss function is determined by the minimum value of the difference in peak signal-to-noise ratio (PSNR) of partial regions in the regions of the second sample image before and after steganography processing, the maximum value of the similarity between the first sample image and the reconstructed first sample image, and a preset classification sub-loss function.

[0097] In step S512, the server determines partial regions in the regions of the second sample image that meet the preset conditions for steganography processing of the first sample image based on the first sample image and the second sample image through a preset region search strategy corresponding to the steganography region model, and uses a preset second loss function and the determined partial regions that meet the preset conditions for steganography processing of the first sample image to determine whether the steganography region model converges. If not, it obtains the first sample image and the second sample image and continues to perform model training on the steganography region model until the steganography region model converges, obtaining the trained steganography region model.

[0098] Among them, the second loss function is negatively correlated with the first loss function. The region search strategy can also be constructed by a search direction and / or a translation step size, and the search direction can include one or more of the following: upward translation, downward translation, leftward translation, and rightward translation.

[0099] In step S514, the server sends the trained steganography region model and the trained encoding model to the terminal device.

[0100] In step S516, the terminal device obtains a biometric request of the target user, and the biometric request includes first information to be processed, and the first information is a first image containing user biometric information.

[0101] In step S518, the terminal device selects corresponding second information for the first information, and the second information is a second image, and the second image is different from the first image.

[0102] In step S520, the terminal device inputs the first image and the second image into a pre-trained steganographic region model to obtain the region corresponding to the second image for performing steganography on the first image (i.e., the region where a part of the image in the second image is located).

[0103] In step S522, the terminal device inputs the first image and the region for performing steganography on the first image into a pre-trained steganography model to obtain the steganographic information in which the first image is steganographically written in the region for performing steganography on the first image.

[0104] In step S524, the terminal device sends the above steganographic information to the server, and the steganographic information is used to trigger the server to perform biometric processing on the target user based on the pre-stored reference user biometric information and the steganographic information.

[0105] In step S526, the terminal device receives the biometric result of the biometric processing on the target user sent by the server.

[0106] In step S528, the terminal device deletes the first image.

[0107] For the specific processing procedures of the above steps S502 to S528, reference can be made to the above relevant content and will not be elaborated here.

[0108] An information processing method is provided in an embodiment of this specification. By obtaining the first information to be processed of the target user, where the first information includes the privacy information of the target user, then, corresponding second information is selected for the first information, and the first information and the second information are input into a pre-trained steganographic region model to obtain the region corresponding to the second information for performing steganography on the first information. The region for performing steganography on the first information is a partial region in the region presenting the second information. The first information and the region for performing steganography on the first information are input into a pre-trained steganography model to obtain the steganographic information in which the first information is steganographically written in the region for performing steganography on the first information. Business processing is performed on the target service based on the steganographic information. In this way, through steganography technology, the user's privacy information is steganographically written into the second information that is convenient for display (such as the user's cartoonized avatar, etc.). This not only performs desensitization processing on the user's privacy information but also facilitates the display of information. In addition, in terms of steganography technology, different from the traditional steganography processing of the entire information, this solution proposes a local-global steganography method, that is, a local region capable of performing steganography processing is retrieved through a corresponding model for steganography processing. On the one hand, the quality of steganography is improved, and on the other hand, the attacker cannot determine in which region the steganography is performed, so that the privacy information is difficult to be leaked, and the security of steganography is improved.

[0109] Embodiment Five

[0110] Based on the same idea, an embodiment of this specification also provides an information processing device, as Figure 6 shown.

[0111] The information processing device includes: an information acquisition module 601, a region determination module 602, a steganography module 603, and a processing module 604, where:

[0112] The information acquisition module 601 acquires first information to be processed by a target user, and the first information includes privacy information of the target user;

[0113] The region determination module 602 selects corresponding second information for the first information, and inputs the first information and the second information into a pre-trained steganography region model to obtain a region corresponding to the second information for performing steganography processing on the first information. The region for performing steganography processing on the first information is a partial region in the region presenting the second information. The steganography region model is used to determine a partial region for performing steganography processing on one information in the presenting region of an information;

[0114] The steganography module 603 inputs the first information and the region for performing steganography processing on the first information into a pre-trained steganography model to obtain steganography information in which the first information is steganographically written in the region for performing steganography processing on the first information. The steganography model is used to steganographically write one information into a partial region of the region for presenting another information;

[0115] The processing module 604 performs service processing on a target service based on the steganography information.

[0116] In an embodiment of this specification, the first information is user biometric information for biometric identification,

[0117] The information acquisition module 601 acquires a biometric identification request of the target user, and the biometric identification request includes the first information to be processed;

[0118] The processing module 604 performs biometric identification processing on the target user based on the steganography information;

[0119] The device further includes:

[0120] An information deletion module that deletes the first information.

[0121] In an embodiment of this specification, the first information is a first image containing user biometric information, the second information is a second image, the second image is different from the first image, and the region for performing steganography processing on the first information is the region where partial images in the second image are located.

[0122] In the embodiments of this specification, the processing module 604 includes:

[0123] An information sending unit that sends the steganographic information to a server, where the steganographic information is used to trigger the server to perform biometric processing on the target user based on pre-stored reference user biometric information and the steganographic information;

[0124] A result receiving unit that receives the biometric result of the biometric processing on the target user sent by the server.

[0125] In the embodiments of this specification, the device further includes:

[0126] A sample acquisition module that acquires first training sample data and a partial area in the area presenting second training sample data, where the first training sample data includes user privacy information;

[0127] A first model training module that jointly trains the steganographic model, the decoding model, and the adversarial model through the first training sample data and the partial area in the area presenting second training sample data to obtain a trained steganographic model, a trained decoding model, and a trained adversarial model. The decoding model is used to perform restoration processing on the steganographically processed first training sample data, and the adversarial model is used to determine whether the first training sample data is steganographically written in two partial areas in the area presenting second training sample data, and one of the two partial areas is an area where the first training sample data is not steganographically written.

[0128] In the embodiments of this specification, the first model training module includes:

[0129] A first data processing unit that inputs the first training sample data and the partial area in the area presenting second training sample data into the steganographic model to obtain steganographically processed first training sample data;

[0130] A second data processing unit that inputs the steganographically processed first training sample data into the decoding model to perform restoration processing on the steganographically processed first training sample data through the decoding model to obtain reconstructed first training sample data;

[0131] A third data processing unit that inputs two partial areas in the area presenting second training sample data into the adversarial model to determine, through the adversarial model, the probability that the first training sample data is steganographically written in each of the two partial areas in the area presenting second training sample data to obtain corresponding output results;

[0132] The model training unit determines whether the steganographic model, the decoding model, and the adversarial model converge based on the first training sample data, the first training sample data after steganographic processing, the reconstructed first training sample data, a partial area in the area presenting the second training sample data, the output result, and a preset first loss function. If not, it obtains the first training sample data and a partial area in the area presenting the second training sample data to continue model training for the steganographic model, the decoding model, and the adversarial model until the steganographic model, the decoding model, and the adversarial model converge, obtaining the trained steganographic model, the trained decoding model, and the trained adversarial model.

[0133] In the embodiments of the present specification, the first loss function is determined by the minimum value of the difference in peak signal-to-noise ratio of a partial area in the area presenting the second training sample data before and after steganographic processing, the maximum value of the similarity between the first training sample data and the reconstructed first training sample data, and a preset classification sub-loss function.

[0134] In the embodiments of the present specification, the device further includes:

[0135] The second model training module determines whether the steganographic area model converges based on the first training sample data and the second training sample data through a preset area search strategy corresponding to the steganographic area model to determine a partial area in the area presenting the second training sample data that meets the preset conditions for steganographically processing the first training sample data, and uses a preset second loss function and the determined partial area that meets the preset conditions for steganographically processing the first training sample data. If not, it obtains the first training sample data and the second training sample data to continue model training for the steganographic area model until the steganographic area model converges, obtaining the trained steganographic area model.

[0136] In the embodiments of the present specification, the second loss function is negatively correlated with the first loss function.

[0137] In the embodiments of the present specification, the area search strategy is constructed by a search direction and / or a translation step size, and the search direction includes one or more of the following: upward translation, downward translation, leftward translation, and rightward translation.

[0138] An embodiment of this specification provides an information processing apparatus. By obtaining first information to be processed by a target user, where the first information includes privacy information of the target user, then selecting corresponding second information for the first information, and inputting the first information and the second information into a pre-trained steganographic region model, a region corresponding to the second information for performing steganography on the first information is obtained. The region for performing steganography on the first information is a partial region in the region presenting the second information. Inputting the first information and the region for performing steganography on the first information into a pre-trained steganography model, steganographic information with the first information steganographically written in the region for performing steganography on the first information is obtained, and business processing is performed on the target business based on the steganographic information. In this way, through steganography technology, the user's privacy information is steganographically written into the second information that is convenient for display (such as the user's cartoonified avatar, etc.). This not only desensitizes the user's privacy information but also facilitates information display. Additionally, in terms of steganography technology, different from the traditional method of performing steganography on the entire information, this solution proposes a local-global steganography method, that is, a local region capable of performing steganography is retrieved through a corresponding model for steganography. On the one hand, the quality of steganography is improved, and on the other hand, the attacker cannot determine in which region steganography is performed, making it difficult for privacy information to be leaked and enhancing the security of steganography.

[0139] Embodiment Six

[0140] The above is the information processing apparatus provided by the embodiments of this specification. Based on the same idea, the embodiments of this specification also provide an information processing device, as Figure 7 shown.

[0141] The information processing device may be a terminal device or a server provided in the above embodiments, etc.

[0142] The information processing device may vary greatly due to configuration or performance differences. It may include one or more processors 701 and a memory 702. One or more application programs or data may be stored in the memory 702. Among them, the memory 702 may be short-term storage or persistent storage. The application programs stored in the memory 702 may include one or more modules (not shown in the figure), and each module may include a series of computer-executable instructions in the information processing device. Further, the processor 701 may be configured to communicate with the memory 702 and execute a series of computer-executable instructions in the memory 702 on the information processing device. The information processing device may also include one or more power supplies 703, one or more wired or wireless network interfaces 704, one or more input / output interfaces 705, and one or more keyboards 706.

[0143] Specifically, in this embodiment, the information processing device includes a memory and one or more programs. One or more programs are stored in the memory, and one or more programs may include one or more modules. Each module may include a series of computer-executable instructions in the information processing device and is configured to be executed by one or more processors. The one or more programs include computer-executable instructions for performing the following:

[0144] Obtain first information to be processed by a target user, where the first information includes privacy information of the target user;

[0145] Select corresponding second information for the first information, and input the first information and the second information into a pre-trained steganographic region model to obtain the region corresponding to the second information for performing steganographic processing on the first information. The region for performing steganographic processing on the first information is a partial region in the region presenting the second information. The steganographic region model is used to determine a partial region for performing steganographic processing on another information in the presentation region of one information;

[0146] Input the first information and the region for performing steganographic processing on the first information into a pre-trained steganographic model to obtain steganographic information with the first information steganographically written in the region for performing steganographic processing on the first information. The steganographic model is used to steganographically write one information into a partial region of the region for presenting another information;

[0147] Perform business processing on a target business based on the steganographic information.

[0148] In the embodiment of this specification, the first information is user biometric information for biometric identification,

[0149] The obtaining of the first information to be processed by the target user includes:

[0150] Obtain a biometric identification request of the target user, where the biometric identification request includes the first information to be processed;

[0151] The performing of business processing on the target business based on the steganographic information includes:

[0152] Perform biometric identification processing on the target user based on the steganographic information;

[0153] The method further includes:

[0154] Delete the first information.

[0155] In the embodiments of this specification, the first information is a first image containing user biometric information, the second information is a second image, the second image is different from the first image, and the area for performing steganography processing on the first information is the area where a partial image in the second image is located.

[0156] In the embodiments of this specification, the biometric recognition processing of the target user based on the steganographic information includes:

[0157] Sending the steganographic information to a server, where the steganographic information is used to trigger the server to perform biometric recognition processing on the target user based on pre-stored reference user biometric information and the steganographic information;

[0158] Receiving the biometric recognition result of the biometric recognition processing of the target user sent by the server.

[0159] In the embodiments of this specification, it further includes:

[0160] Obtaining first training sample data and a partial area in the area presenting second training sample data, where the first training sample data includes user privacy information;

[0161] Through the first training sample data and the partial area in the area presenting second training sample data, jointly training the steganography model, the decoding model, and the adversarial model to obtain a trained steganography model, a trained decoding model, and a trained adversarial model. The decoding model is used to perform restoration processing on the steganography-processed first training sample data, and the adversarial model is used to determine whether the first training sample data is steganographically written in two partial areas in the area presenting second training sample data, and one of the two partial areas is the area where the first training sample data is not steganographically written.

[0162] In the embodiments of this specification, the jointly training the steganography model, the decoding model, and the adversarial model through the first training sample data and the partial area sample in the area presenting second training sample data to obtain a trained steganography model, a trained decoding model, and a trained adversarial model includes:

[0163] Inputting the first training sample data and the partial area in the area presenting second training sample data into the steganography model to obtain steganography-processed first training sample data;

[0164] Inputting the steganography-processed first training sample data into the decoding model to perform restoration processing on the steganography-processed first training sample data through the decoding model to obtain reconstructed first training sample data;

[0165] Input two partial regions in the region presenting the second training sample data into the adversarial model to determine, through the adversarial model, the probability of the first training sample data being steganographically written in each of the two partial regions in the region presenting the second training sample data, and obtain corresponding output results;

[0166] Based on the first training sample data, the first training sample data after steganography processing, the reconstructed first training sample data, the partial regions in the region presenting the second training sample data, the output results, and a preset first loss function, determine whether the steganography model, the decoding model, and the adversarial model converge. If not, obtain the first training sample data and the partial regions in the region presenting the second training sample data and continue to perform model training on the steganography model, the decoding model, and the adversarial model until the steganography model, the decoding model, and the adversarial model converge, and obtain the trained steganography model, the trained decoding model, and the trained adversarial model.

[0167] In the embodiments of this specification, the first loss function is determined by the minimum value of the difference in peak signal-to-noise ratio of the partial regions in the region presenting the second training sample data before and after steganography processing, the maximum value of the similarity between the first training sample data and the reconstructed first training sample data, and a preset classification sub-loss function.

[0168] The embodiments of this specification further include:

[0169] Based on the first training sample data and the second training sample data, through a preset region search strategy corresponding to the steganography region model, determine partial regions in the region presenting the second training sample data that meet the preset conditions for steganographically writing the first training sample data, and use a preset second loss function and the determined partial regions that meet the preset conditions for steganographically writing the first training sample data to determine whether the steganography region model converges. If not, obtain the first training sample data and the second training sample data and continue to perform model training on the steganography region model until the steganography region model converges, and obtain the trained steganography region model.

[0170] In the embodiments of this specification, the second loss function is negatively correlated with the first loss function.

[0171] In the embodiments of this specification, the region search strategy is constructed by a search direction and / or a translation step size, and the search direction includes one or more of the following: upward translation, downward translation, left translation, and right translation.

[0172] An embodiment of this specification provides an information processing device. By obtaining first information to be processed by a target user, where the first information includes privacy information of the target user, then selecting corresponding second information for the first information, and inputting the first information and the second information into a pre-trained steganographic area model, a region corresponding to the second information for steganographically processing the first information is obtained. The region for steganographically processing the first information is a partial region in the region presenting the second information. Inputting the first information and the region for steganographically processing the first information into a pre-trained steganographic model, a steganographic information with the first information steganographically written in the region for steganographically processing the first information is obtained, and business processing is performed on the target business based on the steganographic information. In this way, through steganography technology, the user's privacy information is steganographically written into the second information that is convenient for display (such as the user's cartoonized avatar, etc.). This not only desensitizes the user's privacy information but also facilitates information display. In addition, in terms of steganography technology, different from the traditional steganographic processing of the entire information, this solution proposes a local-global steganographic method, that is, a local region capable of steganographic processing is retrieved through a corresponding model for steganographic processing. On the one hand, the quality of steganography is improved, and on the other hand, the attacker cannot determine in which region the steganographic processing is performed, so that the privacy information is difficult to be leaked, and the security of steganography is enhanced.

[0173] Embodiment Seven

[0174] Further, based on the above Figures 1 to 5 shown method, one or more embodiments of this specification also provide a storage medium for storing computer-executable instruction information. In a specific embodiment, the storage medium can be a USB flash drive, an optical disc, a hard disk, etc. When the computer-executable instruction information stored in the storage medium is executed by a processor, the following process can be realized:

[0175] Obtain first information to be processed by a target user, where the first information includes the privacy information of the target user;

[0176] Select corresponding second information for the first information, and input the first information and the second information into a pre-trained steganographic area model, to obtain a region corresponding to the second information for steganographically processing the first information. The region for steganographically processing the first information is a partial region in the region presenting the second information. The steganographic area model is used to determine a partial region for steganographically processing another information in the presentation region of an information;

[0177] Input the first information and the area for steganography processing of the first information into a pre-trained steganography model to obtain steganographic information with the first information steganographically written in the area for steganography processing of the first information. The steganography model is used to steganographically write one piece of information into a partial area of the area for presenting another piece of information;

[0178] Perform business processing on the target business based on the steganographic information.

[0179] In the embodiments of this specification, the first information is user biometric information for biometric identification,

[0180] The obtaining of the first information to be processed by the target user includes:

[0181] Obtain the biometric identification request of the target user, where the biometric identification request includes the first information to be processed;

[0182] The performing of business processing on the target business based on the steganographic information includes:

[0183] Perform biometric identification processing on the target user based on the steganographic information;

[0184] The method further includes:

[0185] Delete the first information.

[0186] In the embodiments of this specification, the first information is a first image containing user biometric information, the second information is a second image, the second image is different from the first image, and the area for steganography processing of the first information is the area where a partial image in the second image is located.

[0187] In the embodiments of this specification, the performing of biometric identification processing on the target user based on the steganographic information includes:

[0188] Send the steganographic information to the server, where the steganographic information is used to trigger the server to perform biometric identification processing on the target user based on the benchmark user biometric information stored in advance and the steganographic information;

[0189] Receive the biometric identification result of the biometric identification processing on the target user sent by the server.

[0190] In the embodiments of this specification, it further includes:

[0191] Obtain first training sample data and a partial area in the area for presenting second training sample data, where the first training sample data includes the privacy information of the user;

[0192] Jointly train the steganography model, the decoding model, and the adversarial model by using the first training sample data and partial regions in the region presenting the second training sample data, to obtain a trained steganography model, a trained decoding model, and a trained adversarial model. The decoding model is used to restore the first training sample data after steganographic processing, and the adversarial model is used to determine whether the first training sample data is steganographically embedded in two partial regions in the region presenting the second training sample data, where one of the two partial regions is a region without the first training sample data steganographically embedded therein.

[0193] In the embodiments of this specification, the joint training of the steganography model, the decoding model, and the adversarial model by using the first training sample data and partial region samples in the region presenting the second training sample data to obtain a trained steganography model, a trained decoding model, and a trained adversarial model includes:

[0194] Input the first training sample data and partial regions in the region presenting the second training sample data into the steganography model to obtain the first training sample data after steganographic processing;

[0195] Input the first training sample data after steganographic processing into the decoding model to restore the first training sample data after steganographic processing through the decoding model, to obtain the reconstructed first training sample data;

[0196] Input two partial regions in the region presenting the second training sample data into the adversarial model to determine, through the adversarial model, the probability that the first training sample data is steganographically embedded in each of the two partial regions in the region presenting the second training sample data, to obtain corresponding output results;

[0197] Based on the first training sample data, the first training sample data after steganographic processing, the reconstructed first training sample data, partial regions in the region presenting the second training sample data, the output results, and a preset first loss function, determine whether the steganography model, the decoding model, and the adversarial model converge. If not, obtain the first training sample data and partial regions in the region presenting the second training sample data and continue to perform model training on the steganography model, the decoding model, and the adversarial model until the steganography model, the decoding model, and the adversarial model converge, to obtain a trained steganography model, a trained decoding model, and a trained adversarial model.

[0198] In the embodiments of the present specification, the first loss function is determined by the minimum value of the difference between the peak signal-to-noise ratio of a partial region in the region presenting the second training sample data before and after steganography processing, the maximum value of the similarity between the first training sample data and the reconstructed first training sample data, and a preset classification sub-loss function.

[0199] In the embodiments of the present specification, it further includes:

[0200] Based on the first training sample data and the second training sample data, through a preset region search strategy corresponding to the steganographic region model, to determine a partial region in the region presenting the second training sample data that satisfies the preset conditions for steganography processing of the first training sample data, and use a preset second loss function and the determined partial region that satisfies the preset conditions for steganography processing of the first training sample data to determine whether the steganographic region model converges. If not, obtain the first training sample data and the second training sample data and continue to perform model training on the steganographic region model until the steganographic region model converges, obtaining a trained steganographic region model.

[0201] In the embodiments of the present specification, the second loss function is negatively correlated with the first loss function.

[0202] In the embodiments of the present specification, the region search strategy is constructed by a search direction and / or a translation step size, and the search direction includes one or more of the following: upward translation, downward translation, leftward translation, and rightward translation.

[0203] An embodiment of this specification provides a storage medium. By obtaining first information to be processed by a target user, where the first information includes privacy information of the target user, then selecting corresponding second information for the first information, and inputting the first information and the second information into a pre-trained steganographic region model, a region corresponding to the second information for performing steganography on the first information is obtained. The region for performing steganography on the first information is a partial region in the region presenting the second information. Inputting the first information and the region for performing steganography on the first information into a pre-trained steganography model, steganographic information with the first information steganographically written in the region for performing steganography on the first information is obtained. Based on the steganographic information, business processing is performed on the target business. In this way, through steganography technology, the user's privacy information is steganographically written into the second information that is convenient for display (such as the user's cartoonified avatar, etc.), which not only desensitizes the user's privacy information but also facilitates information display. Additionally, in terms of steganography technology, different from the traditional steganography processing on the entire information, this solution proposes a local-global steganography method, that is, a local region capable of performing steganography processing is retrieved through a corresponding model for steganography processing. On the one hand, the quality of steganography is improved, and on the other hand, the attacker cannot determine in which region the steganography is performed, so that the privacy information is difficult to be leaked, and the security of steganography is enhanced.

[0204] The above describes specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be performed in a different order than in the embodiments and still achieve the desired result. Additionally, the processes depicted in the figures do not necessarily require the specific order or sequential order shown to achieve the desired result. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0205] In the 1990s, improvements to a technology could be clearly distinguished as either hardware improvements (e.g., improvements to circuit structures such as diodes, transistors, switches, etc.) or software improvements (improvements to method flows). However, with the development of technology, many method flow improvements today can be regarded as direct improvements to hardware circuit structures. Almost all designers obtain the corresponding hardware circuit structure by programming the improved method flow into the hardware circuit. Therefore, it cannot be said that an improvement to a method flow cannot be implemented using a hardware entity module. For example, a Programmable Logic Device (PLD) (such as a Field Programmable Gate Array (FPGA)) is such an integrated circuit whose logical function is determined by the user programming the device. Designers can program themselves to "integrate" a digital system onto a single PLD, without having to ask a chip manufacturer to design and fabricate a dedicated integrated circuit chip. Moreover, nowadays, instead of manually fabricating integrated circuit chips, this programming is mostly implemented using "logic compiler" software, which is similar to the software compiler used in program development and writing. The original code before compilation also has to be written in a specific programming language, which is called a Hardware Description Language (HDL). And there is not just one type of HDL, but many, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description Language), etc. Currently, the most commonly used are VHDL (Very-High-Speed Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should also be aware that by simply performing a little logical programming on the method flow using the above-mentioned several hardware description languages and programming it into an integrated circuit, it is easy to obtain the hardware circuit that implements the logical method flow.

[0206] The controller can be implemented in any suitable manner. For example, the controller can take the form of, for example, a microprocessor or a processor and a computer-readable medium storing computer-readable program code (such as software or firmware) executable by the (micro)processor, logic gates, switches, an application specific integrated circuit (ASIC), a programmable logic controller, and an embedded microcontroller. Examples of the controller include, but are not limited to, the following microcontrollers: ARC625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicone Labs C8051F320. The memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art also know that in addition to implementing the controller in the form of pure computer-readable program code, it is entirely possible to logically program the method steps to enable the controller to be implemented in the form of logic gates, switches, application specific integrated circuits, programmable logic controllers, and embedded microcontrollers, etc. to achieve the same function. Therefore, such a controller can be considered a hardware component, and the devices included therein for implementing various functions can also be regarded as the structures within the hardware component. Or even, the devices for implementing various functions can be regarded as either software modules for implementing the method or structures within the hardware component.

[0207] The systems, devices, modules, or units illustrated in the above embodiments can be specifically implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer can be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or any combination of these devices.

[0208] For the convenience of description, when describing the above devices, they are described separately as various units according to their functions. Of course, when implementing one or more embodiments of this specification, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0209] Those skilled in the art should understand that the embodiments of this specification can be provided as a method, a system, or a computer program product. Therefore, one or more embodiments of this specification can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, one or more embodiments of this specification can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memory, CD-ROM, optical memory, etc.) containing computer-usable program code.

[0210] Embodiments of this specification are described with reference to the flowcharts and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this specification. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processors of general-purpose computers, special-purpose computers, embedded processors, or other programmable serial-parallel devices for fraud cases to generate a machine, such that the instructions executed by the processors of the computer or other programmable serial-parallel devices for fraud cases generate means for implementing the functions specified in one Figure 1 flow or multiple flows and / or blocks Figure 1 or means for implementing the functions specified in one block or multiple blocks.

[0211] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable serial-parallel device for fraud cases to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including instruction means that implement the functions specified in one Figure 1 flow or multiple flows and / or blocks Figure 1 or means for implementing the functions specified in one block or multiple blocks.

[0212] These computer program instructions can also be loaded onto a computer or other programmable serial-parallel device for fraud cases, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one Figure 1 flow or multiple flows and / or blocks Figure 1 or means for implementing the functions specified in one block or multiple blocks.

[0213] In a typical configuration, a computing device includes one or more processors (CPUs), an input / output interface, a network interface, and a memory.

[0214] The memory may include non-permanent memory in the form of computer-readable media, random access memory (RAM), and / or non-volatile memory, such as read-only memory (ROM) or flash memory (flash RAM). The memory is an example of computer-readable media.

[0215] A computer-readable medium includes both permanent and non-permanent, removable and non-removable media and can implement information storage by any method or technology. The information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transitory medium that can be used to store information accessible by a computing device. As defined herein, a computer-readable medium does not include transitory computer-readable media such as modulated data signals and carrier waves.

[0216] It should also be noted that the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, such that a process, method, article or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising a..." does not exclude the presence of additional identical elements in the process, method, article or device comprising said element.

[0217] Those skilled in the art should understand that the embodiments of this specification can be provided as a method, a system, or a computer program product. Therefore, one or more embodiments of this specification can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, one or more embodiments of this specification can 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.) that contain computer-usable program code.

[0218] One or more embodiments of this specification can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. One or more embodiments of this specification can also be practiced in a distributed computing environment where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0219] The various embodiments in this specification are described in a progressive manner. For the same or similar parts among the various embodiments, reference can be made to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and for the relevant parts, reference can be made to the partial description of the method embodiments.

[0220] The above description is only for the embodiments of this specification and is not intended to limit this application. For those skilled in the art, various modifications and changes can be made to this specification. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of this specification shall be included within the scope of the claims of this specification.

Claims

1. An information processing method, the method comprising: Obtaining first information to be processed by a target user, where the first information includes privacy information of the target user; Selecting corresponding second information for the first information, and inputting the first information and the second information into a pre-trained steganographic area model to obtain an area corresponding to the second information for performing steganographic processing on the first information. The area for performing steganographic processing on the first information is a partial area in the area presenting the second information. The steganographic area model is used to determine a partial area for performing steganographic processing on one information in the presentation area of an information; Inputting the first information and the area for performing steganographic processing on the first information into a pre-trained steganographic model to obtain steganographic information with the first information steganographically embedded in the area for performing steganographic processing on the first information. The steganographic model is used to steganographically embed one information into a partial area of the area for presenting another information; Performing business processing on a target business based on the steganographic information.

2. The method according to claim 1, wherein the first information is user biometric information for biometric identification, The obtaining first information to be processed by a target user includes: Obtaining a biometric identification request of the target user, where the biometric identification request includes the first information to be processed; The performing business processing on a target business based on the steganographic information includes: Performing biometric identification processing on the target user based on the steganographic information; The method further includes: Deleting the first information.

3. The method according to claim 1 or 2, wherein the first information is a first image containing user biometric information, the second information is a second image, the second image is different from the first image, and the area for performing steganographic processing on the first information is the area where a partial image in the second image is located.

4. The method according to claim 2, wherein the performing biometric identification processing on the target user based on the steganographic information includes: Sending the steganographic information to a server, where the steganographic information is used to trigger the server to perform biometric identification processing on the target user based on pre-stored reference user biometric information and the steganographic information; Receiving a biometric identification result of performing biometric identification processing on the target user sent by the server.

5. The method according to claim 1, the method further includes: Obtaining first training sample data and a partial area in the area presenting second training sample data, where the first training sample data includes privacy information of a user; Jointly train the steganographic model, the decoding model, and the adversarial model by using the first training sample data and partial regions in the region presenting the second training sample data, to obtain a trained steganographic model, a trained decoding model, and a trained adversarial model. The decoding model is used to restore the first training sample data after steganographic processing. The adversarial model is used to determine whether the first training sample data is steganographically embedded in two partial regions in the region presenting the second training sample data, where one of the two partial regions is a region without steganographically embedding the first training sample data.

6. The method according to claim 5, wherein the jointly training the steganographic model, the decoding model, and the adversarial model by using the first training sample data and partial region samples in the region presenting the second training sample data to obtain a trained steganographic model, a trained decoding model, and a trained adversarial model comprises: Input the first training sample data and partial regions in the region presenting the second training sample data into the steganographic model to obtain the first training sample data after steganographic processing; Input the first training sample data after steganographic processing into the decoding model to restore the first training sample data after steganographic processing through the decoding model, to obtain the reconstructed first training sample data; Input two partial regions in the region presenting the second training sample data into the adversarial model to determine, through the adversarial model, the probability of steganographically embedding the first training sample data in each of the two partial regions in the region presenting the second training sample data, to obtain corresponding output results; Based on the first training sample data, the first training sample data after steganographic processing, the reconstructed first training sample data, partial regions in the region presenting the second training sample data, the output results, and a preset first loss function, determine whether the steganographic model, the decoding model, and the adversarial model converge. If not, obtain the first training sample data and partial regions in the region presenting the second training sample data and continue to perform model training on the steganographic model, the decoding model, and the adversarial model until the steganographic model, the decoding model, and the adversarial model converge, to obtain a trained steganographic model, a trained decoding model, and a trained adversarial model.

7. The method according to claim 6, wherein the first loss function is determined by the minimum value of the difference in peak signal-to-noise ratio of partial regions in the region presenting the second training sample data before and after steganographic processing, the maximum value of the similarity between the first training sample data and the reconstructed first training sample data, and a preset classification sub-loss function.

8. The method according to claim 6, wherein the method further comprises: Based on the first training sample data and the second training sample data, through a preset region search strategy corresponding to the steganographic region model, to determine a partial region in the region presenting the second training sample data that satisfies the preset conditions for steganographically processing the first training sample data, and use a preset second loss function and the determined partial region that satisfies the preset conditions for steganographically processing the first training sample data to determine whether the steganographic region model converges. If not, obtain the first training sample data and the second training sample data and continue to perform model training on the steganographic region model until the steganographic region model converges, obtaining a trained steganographic region model.

9. The method according to claim 8, wherein the second loss function is negatively correlated with the first loss function.

10. The method according to claim 8, wherein the region search strategy is constructed by a search direction and / or a translation step size, and the search direction includes one or more of the following: upward translation, downward translation, leftward translation, and rightward translation.

11. An information processing apparatus, the apparatus comprising: An information acquisition module, which acquires first information to be processed by a target user, and the first information includes privacy information of the target user; A region determination module, which selects corresponding second information for the first information, and inputs the first information and the second information into a pre-trained steganographic region model, obtaining a region corresponding to the second information for steganographically processing the first information. The region for steganographically processing the first information is a partial region in the region presenting the second information. The steganographic region model is used to determine a partial region for steganographically processing another information in the presentation region of one information; A steganography module, which inputs the first information and the region for steganographically processing the first information into a pre-trained steganography model, obtaining steganographic information in which the first information is steganographically written in the region for steganographically processing the first information. The steganography model is used to steganographically write one information into a partial region of the region for presenting another information; A processing module, which performs business processing on a target business based on the steganographic information.

12. An information processing device, the information processing device comprising: A processor; And A memory arranged to store computer-executable instructions, and when the executable instructions are executed, the processor: Acquires first information to be processed by a target user, and the first information includes privacy information of the target user; Selects corresponding second information for the first information, and inputs the first information and the second information into a pre-trained steganographic region model, obtaining a region corresponding to the second information for steganographically processing the first information. The region for steganographically processing the first information is a partial region in the region presenting the second information. The steganographic region model is used to determine a partial region for steganographically processing another information in the presentation region of one information; Input the first information and the area for steganography processing of the first information into a pre-trained steganography model to obtain steganographic information with the first information steganographically embedded in the area for steganography processing of the first information. The steganography model is used to steganographically embed one information into a partial area of the area for presenting another information; Perform business processing on the target business based on the steganographic information.

13. A storage medium for storing computer-executable instructions that, when executed by a processor, implement the following process: Obtain the first information to be processed by a target user, where the first information includes the privacy information of the target user; Select corresponding second information for the first information, and input the first information and the second information into a pre-trained steganographic area model to obtain the area corresponding to the second information for steganography processing of the first information. The area for steganography processing of the first information is a partial area of the area for presenting the second information. The steganographic area model is used to determine a partial area for steganography processing of another information in the presentation area of one information; Input the first information and the area for steganography processing of the first information into a pre-trained steganography model to obtain steganographic information with the first information steganographically embedded in the area for steganography processing of the first information. The steganography model is used to steganographically embed one information into a partial area of the area for presenting another information; Perform business processing on the target business based on the steganographic information.

Images