Method, device and equipment for blind extraction of privacy information of thumbnail-kept encrypted image

Abstract

The invention relates to a method, a device and equipment for blind extraction of privacy information of a thumbnail-kept encrypted image, which can provide an emergency solution for a user passing identity verification and legal authorization under the abnormal condition that a key cannot be obtained, and improve the security redundancy. The method comprises the following steps: inputting a ciphertext image sample generated by a thumbnail preserving encryption method and a corresponding plaintext image sample as training samples into a privacy information extraction network constructed based on a generative adversarial network, and alternately training a discriminator and a generator in the privacy information extraction network; in the training process, a generator generates prediction privacy information according to a ciphertext image sample, and after parameters of a discriminator are updated by using the ciphertext image sample and the prediction privacy information, the parameters of the generator are updated by using the ciphertext image sample and the prediction privacy information. A user inputs a to-be-processed ciphertext image into a generator in the trained privacy information extraction network, and decryption and privacy information extraction are performed on the ciphertext image in a key-free manner.

Description

Technical Field

[0001] The present application relates to the field of image security technology, and in particular to a method, device and equipment for blindly extracting privacy information of encrypted images with thumbnails preserved. Background Art

[0002] With the popularity of the Internet and various electronic devices, people have captured a large number of images. Due to the limited storage space of electronic devices and the convenience of cloud storage, more and more users choose to save a large number of images on cloud servers, but the security of cloud images has become one of the key issues that people are concerned about. In order to improve the security of cloud images and prevent the leakage of private information in images, researchers have proposed a series of image encryption algorithms, such as traditional image encryption (the ciphertext image is a noise-like image), content-based ciphertext image retrieval, and region of interest image encryption. However, these image encryption methods cannot balance the privacy and availability of ciphertext images well. In order to solve this problem, researchers proposed a thumbnail preservation encryption scheme.

[0003] The thumbnail-preserving encryption scheme erases the image details while keeping the ciphertext image thumbnail the same or similar to the original image thumbnail. The image owner with prior knowledge can identify the content of the ciphertext image through the thumbnail of the ciphertext image. Therefore, this type of scheme has a good balance between the privacy and availability of the ciphertext image. To ensure security, this type of scheme usually needs to use the relevant information of the image (such as generation time, name, etc.) to generate keys through a dynamic update mechanism. However, when this information is abnormally missing or incorrectly modified during image transmission, storage, and processing, the decryption party cannot obtain the correct key and decrypt the ciphertext image. Summary of the invention

[0004] Based on this, it is necessary to address the above technical issues and provide a key-free thumbnail-preserving blind extraction method, device and equipment for privacy information of encrypted images for users who have passed identity authentication and legal authorization under abnormal circumstances.

[0005] A method for blindly extracting private information of an encrypted image by retaining a thumbnail, the method comprising: Acquire a training sample set, wherein the training sample set includes a plurality of ciphertext image samples and corresponding plaintext image samples related to a certain type of target, wherein the ciphertext image samples are generated according to the plaintext image samples by using a thumbnail preservation encryption method; The ciphertext image samples and the corresponding plaintext image samples in the training sample set are used as training samples, and are input into a private information extraction network constructed based on a generative adversarial network. The discriminator and the generator in the private information extraction network are alternately trained to obtain a trained private information extraction network. In the iterative process of training the private information extraction network, the generator generates predicted private information according to the ciphertext image samples, the ciphertext image samples and the predicted private information are used to update the parameters in the discriminator, and then the parameters in the generator are updated using the discriminator after the updated parameters. Obtain a ciphertext image for which private information extraction is to be performed, input the ciphertext image into a generator in the trained private information extraction network, and obtain the private information of the ciphertext image.

[0006] In one embodiment, the generator includes a downsampling convolution block, a residual block, and an upsampling deconvolution block, wherein each of the blocks is also provided with a corresponding batch normalization operation and an activation function.

[0007] In one embodiment, the discriminator extracts multi-scale features after downsampling the input data, and performs true or false discrimination based on features of different sizes.

[0008] In one embodiment, when updating the discriminator: The discriminator calculates the truth-discrimination loss according to the ciphertext image sample and the plaintext image sample; The discriminator calculates the false judgment loss according to the ciphertext image sample and the predicted privacy information; The parameters of the discriminator are updated according to the true judgment loss and the false judgment loss to obtain the discriminator with updated parameters.

[0009] In one embodiment, the discriminator is a Markov discriminator.

[0010] In one embodiment, the private information extraction network also includes a target recognition network, and when the generator is updated: Using the target recognition network to perform target recognition according to the predicted private information and the plaintext image sample, and calculating the target recognition loss according to the recognition result; Calculate the structural similarity loss and the L1 norm loss based on the predicted private information and the plaintext image sample; Calculating the truth-telling loss using the discriminator with updated parameters according to the ciphertext image sample and the predicted privacy information; The parameters of the generator are updated according to the target recognition loss, structural similarity loss, L1 norm loss and true discrimination loss.

[0011] In one embodiment, the target in the ciphertext image sample is a human face.

[0012] The present application also provides a device for blindly extracting private information of an encrypted image with a thumbnail, the device comprising: A training sample set acquisition module, used to acquire a training sample set, wherein the training sample set includes a plurality of ciphertext image samples and corresponding plaintext image samples related to a certain type of target, wherein the ciphertext image samples are generated from the plaintext image samples using a thumbnail preservation encryption method; a private information extraction network training module, for taking the ciphertext image samples and the corresponding plaintext image samples in the training sample set as training samples, inputting them into the private information extraction network constructed based on the generative adversarial network, alternately training the discriminator and the generator in the private information extraction network to obtain a trained private information extraction network, and in the iterative process of training the private information extraction network, the generator generates predicted private information according to the ciphertext image samples, and the parameters in the discriminator are updated by using the ciphertext image samples and the predicted private information, and then the parameters in the generator are updated by using the discriminator after the updated parameters; The private information blind extraction module is used to obtain a ciphertext image for private information extraction, input the ciphertext image into the generator in the trained private information extraction network, and obtain the private information of the ciphertext image.

[0013] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented: Acquire a training sample set, wherein the training sample set includes a plurality of ciphertext image samples and corresponding plaintext image samples related to a certain type of target, wherein the ciphertext image samples are generated according to the plaintext image samples by using a thumbnail preservation encryption method; The ciphertext image samples and the corresponding plaintext image samples in the training sample set are used as training samples, and are input into a private information extraction network constructed based on a generative adversarial network. The discriminator and the generator in the private information extraction network are alternately trained to obtain a trained private information extraction network. In the iterative process of training the private information extraction network, the generator generates predicted private information according to the ciphertext image samples, the ciphertext image samples and the predicted private information are used to update the parameters in the discriminator, and then the parameters in the generator are updated using the discriminator after the updated parameters. Obtain a ciphertext image for which private information extraction is to be performed, input the ciphertext image into a generator in the trained private information extraction network, and obtain the private information of the ciphertext image.

[0014] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the following steps: Acquire a training sample set, wherein the training sample set includes a plurality of ciphertext image samples and corresponding plaintext image samples related to a certain type of target, wherein the ciphertext image samples are generated according to the plaintext image samples by using a thumbnail preservation encryption method; The ciphertext image samples and the corresponding plaintext image samples in the training sample set are used as training samples, and are input into a private information extraction network constructed based on a generative adversarial network. The discriminator and the generator in the private information extraction network are alternately trained to obtain a trained private information extraction network. In the iterative process of training the private information extraction network, the generator generates predicted private information according to the ciphertext image samples, the ciphertext image samples and the predicted private information are used to update the parameters in the discriminator, and then the parameters in the generator are updated using the discriminator after the updated parameters. Obtain a ciphertext image for which private information extraction is to be performed, input the ciphertext image into a generator in the trained private information extraction network, and obtain the private information of the ciphertext image.

[0015] The above-mentioned method, device and equipment for blindly extracting private information of thumbnail-preserving encrypted images use the ciphertext image samples and the corresponding plaintext image samples in the training sample set generated by the thumbnail-preserving encryption method as training samples, and input them into the private information extraction network constructed based on the generative adversarial network, and alternately train the discriminator and generator therein. In the iterative training process, the generator generates predicted private information based on the ciphertext image samples, and the parameters in the discriminator are updated using the ciphertext image samples and the predicted private information, and then the parameters in the generator are updated using the ciphertext image samples and the predicted private information, and the ciphertext image to be subjected to private information extraction is input into the generator in the trained private information extraction network, so that users who have passed identity authentication and legal authorization can decrypt the ciphertext image and extract private information in abnormal circumstances such as being unable to obtain the key, thereby improving the security redundancy of the thumbnail-preserving encryption scheme. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A schematic flow chart of a method for blindly extracting private information of an encrypted image with thumbnails in one embodiment; Figure 2 is a schematic diagram of the structure of a discriminator in one embodiment; Figure 3 A schematic diagram of training a private information extraction network in one embodiment; Figure 4A structural block diagram of a device for blindly extracting private information of an encrypted image with thumbnails in one embodiment; Figure 5 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0018] In view of the problem that during the transmission, storage and processing of thumbnail encrypted images, abnormal situations such as missing information and incorrect modification may occur, and the key cannot be generated through the key update mechanism to achieve correct decryption of the image and extraction of private information, in this application, Figure 1 As shown, a key-free thumbnail-preserving encrypted image privacy information blind extraction method is provided for authenticated and legally authorized users in emergency situations, including the following steps: Step S100, obtaining a training sample set, which includes a plurality of ciphertext image samples and corresponding plaintext image samples related to a certain type of target, and the ciphertext image samples are generated according to the plaintext image samples using a thumbnail preservation encryption method.

[0019] Step S110, the ciphertext image samples and the corresponding plaintext image samples in the training sample set are used as training samples, and are input into the private information extraction network constructed based on the generative adversarial network. The discriminator and the generator in the private information extraction network are alternately trained to obtain a trained private information extraction network. In the iterative process of training the private information extraction network, the generator generates predicted private information based on the ciphertext image samples, the ciphertext image samples and the predicted private information are used to update the parameters in the discriminator, and then the parameters in the generator are updated using the discriminator after the updated parameters.

[0020] Step S120, obtaining a ciphertext image for private information extraction, inputting the ciphertext image into a generator in a trained private information extraction network, and obtaining the private information of the ciphertext image.

[0021] In this application, a keyless thumbnail-preserving encrypted image privacy information blind extraction method is proposed. This method is based on the deep learning GAN model, i.e., the generative adversarial network, and is optimized by combining the residual network and the multi-scale feature discriminator. The ciphertext image is input into the generator network model to process the ciphertext image and extract the privacy information therein. This method can provide an emergency solution for users who have passed identity authentication and legal authorization when the key cannot be generated normally, thereby improving the security redundancy of the thumbnail-preserving encryption scheme.

[0022] The application scenario of this method can be that the user encrypts the image thumbnail to protect his privacy and uploads the encrypted image to a cloud storage platform. However, during the transmission, storage and processing of the encrypted image, due to some reasons, abnormal situations such as missing image information or incorrect modification occur, resulting in the inability to obtain the correct decryption key. After passing the identity authentication and legal authorization, the user can use this method to decrypt the encrypted image and extract the privacy information, thereby improving the security redundancy of the thumbnail encryption scheme.

[0023] Furthermore, the method can also be used to assist law enforcement, and with the authorization of security agencies or judicial agencies, assist in cracking encrypted information related to public security.

[0024] In step S100, since this method mainly processes ciphertext images encrypted by the thumbnail-preserving encryption method, when constructing a training sample set, first, based on multiple plaintext image samples of a certain type of target, ciphertext image samples are encrypted by the thumbnail-preserving encryption method, and a training sample set is constructed based on the ciphertext image samples and the corresponding plaintext image samples, wherein the plaintext image samples are used as true value labels.

[0025] In this embodiment, the target type in the image may be a human face, an animal, a specific object such as a vehicle, a building, daily necessities, etc. In this article, a human face target is taken as an example for description.

[0026] Specifically, several images are selected from the face image dataset FFHQ (Flickr-Faces-HQ) and scaled to 3×256×256 pixels as the training set trainY. The training set trainY belongs to the plaintext image sample Y domain. The images of the training set trainY are thumbnailed and encrypted to generate the training set trainX. The training set trainX belongs to the ciphertext image sample X domain. Then, the training sample set is constructed according to the training set trainY and the training set trainX.

[0027] In one embodiment, there are 20,000 images in trainX and trainY, and the image size is 3×256×256. The TPE-ADE encryption method is used for the plaintext image samples, and each JPEG plaintext image sample is encrypted according to the thumbnail block size of 16×16 to form the data set trainX required for model training.

[0028] In step S110, the ciphertext image samples and the corresponding plaintext image samples in the training sample set are input as training samples into the private information extraction network for training.

[0029] In this embodiment, the private information extraction network is composed of a generator G and a discriminator D. The generator G includes a downsampling convolution block, a residual block, and an upsampling deconvolution block, wherein each block is also provided with a corresponding batch normalization operation and an activation function. After the discriminator D downsamples the input data, it extracts multi-scale features and distinguishes true from false based on the features of different sizes.

[0030] Specifically, the discriminator D downsamples the input image and outputs the sampling convolution results at different levels to calculate the loss, that is, multi-scale feature discrimination is performed. For example, image X is obtained by convolution layer 1 to obtain X1, and X1 is obtained by convolution layer 2 to obtain X2. Then both X1 and X2 are output to calculate the loss of the discriminator D. At the same time, in this method, the ciphertext image samples also need to be input into the discriminator to extract their features, so as to improve the ability of the discriminator D to fight against the generator G, thereby improving the ability of the generator G to extract private information.

[0031] like Figure 2 As shown in the figure, it is the specific network structure of the discriminator D, whose input is the ciphertext image sample (i.e. Figure 2 ), and the corresponding plaintext image samples or predicted private information (i.e. Figure 2 The generated image or plaintext image in the image is first concatenated with the two input data to obtain the 6-channel image data (i.e. Figure 2), and then downsample the 6-channel image data, where the downsampled input is 6 channels, the convolution kernel is 4×4, the stride is 2, and the output is 64 channels. The downsampled data passes through 4 multi-scale feature extraction units in sequence. The inputs of these 4 different scale feature extraction units are 64 channels, 128 channels, 256 channels, and 512 channels, respectively. The convolution kernel size is 4×4, the stride of the first two scale feature extraction units is 2, and the stride of the last two scale feature extraction units is 1. The outputs are 128 channels, 256 channels, 512 channels, and 1 channel, respectively. The output convolution results are 128×63×63, 256×30×30, 512×27×27, and 1×26×26, respectively. The multi-scale features extracted by these four different scale feature extraction units are then used for discrimination.

[0032] In this embodiment, in the private information blind extraction network, the generator G first generates the predicted private information fakeY according to the ciphertext image sample realX, using the following formula: ; Next, the parameters in the discriminator D are preliminarily updated using the adversarial loss. The adversarial loss is a loss function that constrains the discriminator D, including the false judgment loss and the true judgment loss. Among them, the discriminator D is based on the ciphertext image sample realX and the plaintext image sample , calculate the true judgment loss, and calculate the false judgment loss based on the ciphertext image sample realX and the predicted private information fakeY.

[0033] Furthermore, the discriminator D uses the adversarial loss function For plaintext image samples And the predicted private information fakeY is identified, and the process is expressed as: ; In the above formula, Indicates that Input the discriminator D to get the discrimination result, true represents 1, false represents 0, the result is When the discriminator D thinks belongs to the plaintext image sample Y domain. When the result is false, it means that the discriminator D believes that realY does not belong to the plaintext image sample Y domain. Indicates that the images are concatenated in the channel dimension. Represents the calculation of cross entropy loss.

[0034] In this embodiment, a Markov discriminator is used in the discriminator D, which outputs an n×n matrix. When calculating the adversarial loss, the true loss is calculated with the all-1 matrix, and the false loss is calculated with the all-0 matrix.

[0035] Furthermore, when using the adversarial loss function After the parameters of the discriminator D are preliminarily updated, the parameters in the generator G are updated using the discriminator D with updated parameters.

[0036] In this embodiment, the private information extraction network also includes a target recognition network. When the generator is updated: the target recognition network is used to perform target recognition based on the predicted private information and the plaintext image samples, and the target recognition loss is calculated based on the recognition results. The structural similarity loss and the L1 norm loss are calculated based on the predicted private information and the plaintext image samples. The discriminator D with updated parameters is used to calculate the true detection loss based on the ciphertext image samples and the predicted private information. Finally, the parameters of the generator are updated based on the target recognition loss, the structural similarity loss, the L1 norm loss and the true detection loss.

[0037] Specifically, the discriminator D with updated parameters is used to calculate the true loss according to the ciphertext image sample and the predicted privacy information. , that is, the discriminator D with updated parameters is used to judge whether the image generated by the generator G belongs to the Y domain, thereby constraining whether the private information generated by the generator G belongs to the Y domain. It is expressed as: ; Furthermore, we also use the structural similarity loss And L1 norm loss The constructed perceptual loss function constrains the generator G, that is, compares the plaintext image sample realY with the generated predicted privacy information fakeY, so as to constrain the image generated by the generator G to have richer texture and contain more privacy information. The formula is: ; ; In the above formula, Indicates the calculation of L1 norm distance, Indicates the calculation of structural similarity, where a value of 1 indicates the same and a value of 0 indicates completely different.

[0038] Furthermore, the perceptual loss function is expressed as: ; In the above formula, and It is a constant coefficient, ranging from 0 to 100.

[0039] Furthermore, the target recognition loss L is also used IDThe loss function constrains the generator G, that is, the plaintext image sample realY and the generated predicted private information fakeY are input into the target recognition network, thereby constraining the private information in the image generated by the generator G to be the same as the private information in the real image.

[0040] In this embodiment, the target is a human face as an example, that is, the target recognition network is a face recognition model, which is expressed as: ; In the above formula, recognizer(realY) and recognizer(fakeY) respectively represent inputting realY and fakeY into the face recognition model recognizer to obtain their respective identity feature vectors.

[0041] In this embodiment, the face recognition model used is SEResNet50. Refers to calculating cosine similarity.

[0042] It should be noted here that when the target in the ciphertext image is not a human face, a network that can recognize the target information can be used.

[0043] In this embodiment, the loss of the discriminator D is expressed as: ;

[0044] Furthermore, the loss of the generator is expressed as: ; In the above formula, and It is a constant coefficient, ranging from 0 to 100.

[0045] In step S120, the trained generator G performs blind extraction of private information on the encrypted ciphertext image whose thumbnail remains encrypted. The ciphertext image from which private information needs to be extracted is input into the trained generator G. The image is firstly feature extracted through a convolutional neural network, and the feature matrix is ​​transferred through a residual block. Then, the feature matrix is ​​decoded using a deconvolution layer, and finally a 3×256×256 image in the generated plaintext domain is obtained. That is, the ciphertext image is processed and the private information therein is extracted under the key-free condition.

[0046] In this embodiment, since the thumbnail-preserving encryption method includes multiple other methods in addition to the TPE-ADE encryption method proposed above, at this time, the corresponding generator G can be trained for each specific method, that is, the training data set is concentrated, and the corresponding thumbnail-preserving encryption method is used to generate the corresponding ciphertext image sample based on the plaintext image sample.

[0047] like Figure 3 As shown in Figure 1, it is a schematic diagram of training a private information extraction network.

[0048] In the above-mentioned method for blindly extracting private information of thumbnail-preserving encrypted images, an emergency solution can be provided for users who have passed identity authentication and legal authorization in abnormal situations such as failure to obtain keys, thereby improving security redundancy. The ciphertext image samples and the corresponding plaintext image samples in the training sample set generated by the thumbnail-preserving encryption method are used as training samples and input into the private information extraction network constructed based on the generative adversarial network, and the discriminator and generator in the private information extraction network are alternately trained. During the iterative training process, the generator generates predicted private information based on the ciphertext image samples, and the parameters in the discriminator are updated using the ciphertext image samples and the predicted private information, and then the parameters in the generator are updated using the ciphertext image samples and the predicted private information. The user can input the ciphertext image to be processed into the generator in the trained private information extraction network, and decrypt the ciphertext image and extract private information in a key-free manner. The private information extracted by this method includes facial contours, eye features, hairstyles, etc. Compared with ciphertext images, more detailed private information can be intuitively reflected, so that the predicted private information is very close to the original plaintext image samples in visual perception, thereby enhancing the availability of predicted private information.

[0049] It should be understood that although Figure 1 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 1 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0050] In one embodiment, Figure 4 As shown, a device for blindly extracting private information of an encrypted image with thumbnails is provided, comprising: a training sample set acquisition module 200, a private information extraction network training module 210 and a private information blind extraction module 220, wherein: The training sample set acquisition module 200 is used to acquire a training sample set, wherein the training sample set includes a plurality of ciphertext image samples and corresponding plaintext image samples related to a certain type of target, wherein the ciphertext image samples are generated from the plaintext image samples using a thumbnail preservation encryption method; The private information extraction network training module 210 is used to input the ciphertext image samples and the corresponding plaintext image samples in the training sample set as training samples into the private information extraction network constructed based on the generative adversarial network, alternately train the discriminator and the generator in the private information extraction network to obtain a trained private information extraction network. In the iterative process of training the private information extraction network, the generator generates predicted private information according to the ciphertext image samples, uses the ciphertext image samples and the predicted private information to update the parameters in the discriminator, and then uses the discriminator with updated parameters to update the parameters in the generator. The private information blind extraction module 220 is used to obtain a ciphertext image for private information extraction, input the ciphertext image into the generator in the trained private information extraction network, and obtain the private information of the ciphertext image.

[0051] For the specific definition of the blind extraction device for privacy information of encrypted images preserved in thumbnails, please refer to the definition of the blind extraction method for privacy information of encrypted images preserved in thumbnails above, which will not be repeated here. Each module in the above-mentioned blind extraction device for privacy information of encrypted images preserved in thumbnails can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0052] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as follows: Figure 5 As shown. The computer device includes a processor, a memory, a network interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for blindly extracting privacy information of an encrypted image with a thumbnail is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a key, trackball or touchpad set on the computer device housing, or an external keyboard, touchpad or mouse, etc.

[0053] Those skilled in the art will understand that Figure 5The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0054] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented: Acquire a training sample set, wherein the training sample set includes a plurality of ciphertext image samples and corresponding plaintext image samples related to a certain type of target, wherein the ciphertext image samples are generated according to the plaintext image samples by using a thumbnail preservation encryption method; The ciphertext image samples and the corresponding plaintext image samples in the training sample set are used as training samples, and are input into a private information extraction network constructed based on a generative adversarial network. The discriminator and the generator in the private information extraction network are alternately trained to obtain a trained private information extraction network. In the iterative process of training the private information extraction network, the generator generates predicted private information according to the ciphertext image samples, the ciphertext image samples and the predicted private information are used to update the parameters in the discriminator, and then the parameters in the generator are updated using the discriminator after the updated parameters. Obtain a ciphertext image for which private information extraction is to be performed, input the ciphertext image into a generator in the trained private information extraction network, and obtain the private information of the ciphertext image.

[0055] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented: Acquire a training sample set, wherein the training sample set includes a plurality of ciphertext image samples and corresponding plaintext image samples related to a certain type of target, wherein the ciphertext image samples are generated according to the plaintext image samples by using a thumbnail preservation encryption method; The ciphertext image samples and the corresponding plaintext image samples in the training sample set are used as training samples, and are input into a private information extraction network constructed based on a generative adversarial network. The discriminator and the generator in the private information extraction network are alternately trained to obtain a trained private information extraction network. In the iterative process of training the private information extraction network, the generator generates predicted private information according to the ciphertext image samples, the ciphertext image samples and the predicted private information are used to update the parameters in the discriminator, and then the parameters in the generator are updated using the discriminator after the updated parameters. Obtain a ciphertext image for which private information extraction is to be performed, input the ciphertext image into a generator in the trained private information extraction network, and obtain the private information of the ciphertext image.

[0056] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0057] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0058] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention. It should be pointed out that, for a person of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A method for blindly extracting private information of encrypted images by keeping thumbnails, characterized in that: The method comprises: Acquire a training sample set, wherein the training sample set includes a plurality of ciphertext image samples and corresponding plaintext image samples related to a certain type of target, wherein the ciphertext image samples are generated according to the plaintext image samples by using a thumbnail preservation encryption method; The ciphertext image samples and the corresponding plaintext image samples in the training sample set are used as training samples, and are input into a private information extraction network constructed based on a generative adversarial network. The discriminator and the generator in the private information extraction network are alternately trained to obtain a trained private information extraction network. In the iterative process of training the private information extraction network, the generator generates predicted private information according to the ciphertext image samples, the ciphertext image samples and the predicted private information are used to update the parameters in the discriminator, and then the parameters in the generator are updated using the discriminator after the updated parameters. Obtain a ciphertext image for which private information extraction is to be performed, input the ciphertext image into a generator in the trained private information extraction network, and obtain the private information of the ciphertext image.

2. The method for blindly extracting private information of an encrypted image by retaining a thumbnail according to claim 1, characterized in that: The generator includes a downsampling convolution block, a residual block, and an upsampling deconvolution block, wherein each of the blocks is also provided with a corresponding batch normalization operation and an activation function.

3. The method for blindly extracting private information of an encrypted image by retaining a thumbnail according to claim 2, characterized in that: After downsampling the input data, the discriminator extracts multi-scale features and performs true and false discrimination based on features of different sizes.

4. The method for blindly extracting private information of an encrypted image by retaining a thumbnail according to claim 3, characterized in that: When updating the discriminator: The discriminator calculates the truth-discrimination loss according to the ciphertext image sample and the plaintext image sample; The discriminator calculates the false judgment loss according to the ciphertext image sample and the predicted privacy information; The parameters of the discriminator are updated according to the true judgment loss and the false judgment loss to obtain the discriminator with updated parameters.

5. The method for blindly extracting private information of an encrypted image by retaining a thumbnail according to claim 4, characterized in that: The discriminator adopts a Markov discriminator.

6. The method for blindly extracting private information of an encrypted image by keeping a thumbnail according to claim 5, characterized in that: The privacy information extraction network also includes a target recognition network. When the generator is updated: Using the target recognition network to perform target recognition according to the predicted private information and the plaintext image sample, and calculating the target recognition loss according to the recognition result; Calculate the structural similarity loss and the L1 norm loss based on the predicted private information and the plaintext image sample; Calculating the truth-telling loss using the discriminator with updated parameters according to the ciphertext image sample and the predicted privacy information; The parameters of the generator are updated according to the target recognition loss, structural similarity loss, L1 norm loss and true discrimination loss.

7. The method for blindly extracting private information of an encrypted image with thumbnails preserved according to any one of claims 1 to 6, characterized in that: The target in the ciphertext image sample is a human face.

8. A device for blindly extracting private information of an encrypted image by retaining a thumbnail, characterized in that: The device comprises: A training sample set acquisition module, used to acquire a training sample set, wherein the training sample set includes a plurality of ciphertext image samples and corresponding plaintext image samples related to a certain type of target, wherein the ciphertext image samples are generated from the plaintext image samples using a thumbnail preservation encryption method; a private information extraction network training module, for taking the ciphertext image samples and the corresponding plaintext image samples in the training sample set as training samples, inputting them into the private information extraction network constructed based on the generative adversarial network, alternately training the discriminator and the generator in the private information extraction network to obtain a trained private information extraction network, and in the iterative process of training the private information extraction network, the generator generates predicted private information according to the ciphertext image samples, and the parameters in the discriminator are updated by using the ciphertext image samples and the predicted private information, and then the parameters in the generator are updated by using the discriminator after the updated parameters; The private information blind extraction module is used to obtain a ciphertext image for private information extraction, input the ciphertext image into the generator in the trained private information extraction network, and obtain the private information of the ciphertext image.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

Images