Method and device for protecting privacy of fully-encrypted database

By extracting deep feature vectors and generating hash codes on the client, the resource waste and server congestion problems during cloud data retrieval are solved, and efficient encrypted data retrieval and decryption are realized, ensuring data security.

CN118981785AActive Publication Date: 2024-11-19SOUTH CHINA UNIV OF TECH
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202410934032.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-12
Publication Date
2024-11-19
Estimated Expiration
2044-07-12

AI Technical Summary

Technical Problem

When users need to retrieve encrypted data stored in the cloud, they must first download all the data and then decrypt it before retrieval, resulting in waste of network bandwidth, computing resources and storage space, which may lead to server congestion.

Method used

By extracting the deep feature vector of query fragment data on the client, generating hash codes and features, and uploading them to the cloud server to find and send target encrypted data. The client uses key decryption to obtain the original data, avoiding the download and decryption of the entire data operate.

Benefits of technology

It improves query efficiency and response speed, reduces the resources consumed in data retrieval, ensures that the required data is retrieved without obtaining data content, avoids the necessity of data download and decryption, and ensures data security.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118981785A_ABST
    Figure CN118981785A_ABST
Patent Text Reader

Abstract

The invention discloses a fully-encrypted database privacy protection method and device, and relates to the technical field of data encryption. The method comprises the following steps: extracting depth feature vectors of query fragment data through a preset model to obtain a first depth feature data set; obtaining a query fragment hash code and a query fragment feature through the first depth feature data set; and uploading the query fragment hash code and the query fragment feature to a cloud server to determine the target encrypted data, and decrypting the target encrypted data through the first key to obtain the original data. The encrypted data set corresponding to the query fragment hash code in the preset database is searched through the generated query fragment hash code, the query efficiency and the response speed are improved, operation on all data is avoided, then the target encrypted data is determined through the query fragment feature, and it is guaranteed that under the condition that the data content is not obtained, the query efficiency is improved. The required data is retrieved, and resources consumed by data retrieval are reduced under the condition that the data security is ensured.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of data encryption, and in particular relates to a privacy protection method and device for a fully encrypted database. Background Art

[0002] Fully Homomorphic Encryption (FHE) is an encryption technology that allows calculations to be performed on encrypted data without first decrypting it. This means that various operations and analyses can be performed on the data while keeping it encrypted, thereby protecting the privacy and security of the data.

[0003] With the rapid development of network multimedia information and intelligent applications, many enterprises and individuals choose to store local data in cloud servers. However, as a third party, cloud servers are not completely trustworthy, which makes users lose physical control over data when outsourcing data to the cloud, thus raising issues of data security and privacy protection, especially for sensitive voice data. To solve this problem, data is usually encrypted and then stored in the cloud in ciphertext form. However, when users need to retrieve encrypted data stored in the cloud, they must first download all the data and then decrypt it before retrieving it. This operation not only consumes a lot of network bandwidth, computing resources and local storage space, but may also cause server congestion. Summary of the invention

[0004] The purpose of the present invention is to solve the problem that when a user needs to retrieve encrypted data stored in the cloud, all the data must be downloaded first and then decrypted before retrieval. This operation not only consumes a large amount of network bandwidth, computing resources and local storage space, but may also cause server congestion. A privacy protection method and device for a fully encrypted database is proposed.

[0005] In the first aspect of the present invention, a privacy protection method for a fully secret database is first proposed, and the method is applied to a client:

[0006] Extracting deep feature vectors of query fragment data using a preset model to obtain a first deep feature data set;

[0007] The first deep feature data set obtains a query fragment hash code and a query fragment feature corresponding to the query fragment data by a hash generation method and a query feature generation method;

[0008] The query segment hash code and the query segment feature are uploaded to the cloud server; the cloud server searches for an encrypted data set corresponding to the query segment hash code in a preset database according to the query segment hash code, and for each encrypted data, determines the target encrypted data according to the query segment feature and the complete original feature corresponding to the encrypted audio data, and sends the target encrypted data to the client; the target encrypted data is the original data encrypted by the first key; the query segment data is a segment in the original data;

[0009] The target encrypted data is obtained, and the target encrypted data is decrypted using the first key to obtain original data.

[0010] Optionally, the original data is audio data, and extracting deep feature data of the query segment data by using a preset model includes:

[0011] Opening the query segment data from a preset target library to obtain speech waveform data of the query segment data;

[0012] Removing data from all parts of the speech waveform data that do not contain actual speech to obtain a plurality of speech waveform data segments; the parts that do not contain actual speech are silence segments and background noise segments;

[0013] Calculate the MFCC coefficients of the speech frames in each speech waveform data segment to obtain multiple sets of feature data;

[0014] All group feature data are brought into the CNN model to obtain deep feature data.

[0015] Optionally, the hash generation method specifically includes:

[0016] Sort the deep feature vectors in the deep feature data set by size to obtain a feature sorting set;

[0017] Determine the depth feature vector value corresponding to the median position in the feature sorting set, and record it as the target judgment value;

[0018] A target hash code is obtained by performing a hash transformation on all the depth feature vectors in the depth feature data set with the target judgment value as the judgment condition; a depth feature vector value greater than the target judgment value in the depth feature data set is recorded as 1, and a depth feature vector value less than or equal to the target judgment value is recorded as 0.

[0019] Optionally, the query feature generation method specifically includes: performing data processing on the first deep feature data set using a second key to obtain a query fragment feature; the second key includes a first random number and a preset matrix;

[0020] For the first depth feature data set, convert all first depth feature data vectors to a first target dimension using the first random number to obtain a first target feature vector; the first target dimension is the dimension of the first depth feature data vector in the first depth feature data set plus 3;

[0021] The first target feature vector is encrypted by a preset matrix to obtain a first target fragment feature; the number of rows and columns of the preset matrix is ​​the same as the dimension size of the first depth feature data vector in the first depth feature data set.

[0022] Optionally, the method applied to the client further includes:

[0023] Extracting the depth feature vector of the original data through a preset model to obtain a second depth feature data set;

[0024] The second deep feature data set obtains a complete hash code and a complete original feature corresponding to the original data by a hash generation method and a complete feature generation method;

[0025] Encrypting the original data using the first key to obtain encrypted data;

[0026] The encrypted data is bound to the complete hash code and the complete original feature and then uploaded to the cloud server.

[0027] Optionally, the complete feature generation method specifically includes: performing data processing on the second deep feature data set using a third key to obtain an original complete feature; the third key includes a second random number, a third random number and a preset inverse matrix;

[0028] For the second depth feature data set, convert all second depth feature data vectors into a second target dimension by using the second random number and the third random number to obtain a second target feature vector; the second target dimension is the dimension of the second depth feature data vector in the second depth feature data set plus 3;

[0029] The second target feature vector is encrypted by a preset inverse matrix to obtain the original complete feature; the number of rows and columns of the preset inverse matrix is ​​the same as the dimension size of the second depth feature data vector in the inverse depth feature data set.

[0030] In a second aspect of the present invention, a privacy protection method for a fully encrypted database is proposed, and the method is applied to a cloud server:

[0031] Obtain the query fragment hash code, and search a preset database for an encrypted data set corresponding to the query fragment hash code; the encrypted data set includes encrypted data, a complete hash code, and complete original features;

[0032] The query fragment feature is obtained, and the target distance between the query fragment feature and the complete original feature corresponding to each encrypted data in the encrypted data set is calculated by Euclidean distance, the encrypted data corresponding to the minimum target distance is determined and recorded as the target encrypted data, and the target encrypted data is sent to the client.

[0033] In a third aspect of the present invention, a privacy protection device for a fully secret database is proposed, wherein the device is applied to a client and comprises:

[0034] A first deep feature generation module, configured to extract a deep feature vector of the query fragment data through a preset model to obtain a first deep feature data set;

[0035] A first deep feature processing module, configured to obtain a query fragment hash code and a query fragment feature corresponding to the query fragment data by using a hash generation method and a query feature generation method for the first deep feature data set;

[0036] A query module is used to upload the query segment hash code and the query segment feature to a cloud server; so that the cloud server searches for an encrypted data set corresponding to the query segment hash code in a preset database according to the query segment hash code, and for each encrypted data, determines a target encrypted data according to the query segment feature and the complete original feature corresponding to the encrypted audio data, and sends the target encrypted data to a client; the target encrypted data is the original data encrypted by a first key; the query segment data is a segment in the original data;

[0037] The decryption module is used to obtain the target encrypted data and decrypt the target encrypted data using the first key to obtain the original data.

[0038] In a fourth aspect of the present invention, a privacy protection device for a fully secret database is proposed, wherein the device is applied to a cloud server:

[0039] An encrypted data set search module is used to obtain the query fragment hash code and search for an encrypted data set corresponding to the query fragment hash code in a preset database; the encrypted data set includes encrypted data, a complete hash code and complete original features;

[0040] The target encrypted data determination module is used to obtain the query fragment feature, calculate the target distance between the query fragment feature and the complete original feature corresponding to each encrypted data in the encrypted data set through Euclidean distance, determine the encrypted data corresponding to the minimum target distance as the target encrypted data, and send the target encrypted data to the client.

[0041] Beneficial effects of the present invention:

[0042] The present invention proposes a privacy protection method for a fully encrypted database, wherein a first deep feature data set is obtained by extracting a deep feature vector of query fragment data through a preset model; the first deep feature data set obtains a query fragment hash code and a query fragment feature corresponding to the query fragment data through a hash generation method and a query feature generation method; the query fragment hash code and the query fragment feature are uploaded to a cloud server; the cloud server searches for an encrypted data set corresponding to the query fragment hash code in a preset database according to the query fragment hash code, determines target encrypted data according to the query fragment feature and the complete original feature corresponding to the encrypted audio data for each encrypted data, and sends the target encrypted data to a client; obtains the target encrypted data, and decrypts the target encrypted data through a first key to obtain the original data. The encrypted data set corresponding to the query fragment hash code in the preset database is searched through the generated query fragment hash code, thereby improving the query efficiency and response speed, avoiding the need to operate all data, and then determining the target encrypted data through the query fragment feature, ensuring that the required data is retrieved without obtaining the data content, avoiding the problem that when a user needs to retrieve encrypted data stored in the cloud, all data must be downloaded first, and then decrypted before retrieval, thereby reducing the resources consumed by data retrieval while ensuring data security. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The present invention will be further described below in conjunction with the accompanying drawings.

[0044] Figure 1 A flowchart of a fully encrypted database privacy protection method is provided for an embodiment of the present invention;

[0045] Figure 2 A flowchart of another method for protecting privacy in a fully encrypted database is provided for an embodiment of the present invention;

[0046] Figure 3 A schematic diagram of the structure of a fully encrypted database privacy protection device is provided for an embodiment of the present invention;

[0047] Figure 4 A structural schematic diagram of another fully confidential database privacy protection device is provided for an embodiment of the present invention. DETAILED DESCRIPTION

[0048] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. The term "and / or" herein is only a description of the association relationship of the associated objects, indicating that there can be three relationships. For example, A and B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, in the present invention, the description of "first", "second", etc. is only used for descriptive purposes, and cannot be understood as indicating or implying its relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" can explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but it must be based on the ability of ordinary technicians in the field to implement. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0049] Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative work shall fall within the scope of protection of the present invention.

[0050] The embodiment of the present invention provides a privacy protection method for a fully encrypted database. Figure 1 , Figure 1 A flowchart of a privacy protection method for a fully secret database provided by an embodiment of the present invention. The method is applied to a client and specifically includes the following steps:

[0051] S101, extracting deep feature vectors of query fragment data through a preset model to obtain a first deep feature data set.

[0052] S102: The first deep feature data set obtains a query fragment hash code and a query fragment feature corresponding to the query fragment data through a hash generation method and a query feature generation method.

[0053] S103, upload the query segment hash code and query segment features to the cloud server; so that the cloud server searches the preset database for an encrypted data set corresponding to the query segment hash code according to the query segment hash code, and for each encrypted data, determines the target encrypted data according to the query segment features and the complete original features corresponding to the encrypted audio data, and sends the target encrypted data to the client.

[0054] S104, obtaining target encrypted data, and decrypting the target encrypted data using a first key to obtain original data.

[0055] The target encrypted data is the original data encrypted by the first key; the query fragment data is the fragment in the original data;

[0056] A privacy protection method for a fully secret database provided by an embodiment of the present invention uses a generated query fragment hash code to search for an encrypted data set corresponding to the query fragment hash code in a preset database, thereby improving query efficiency and response speed, avoiding operations on all data, and then determining the target encrypted data through query fragment features, thereby ensuring that the required data is retrieved without obtaining the data content, avoiding the problem that when a user needs to retrieve encrypted data stored in the cloud, all data must be downloaded first and then decrypted before retrieval, thereby reducing the resources consumed by data retrieval while ensuring data security.

[0057] In one implementation, a preset model is used to extract deep feature vectors of the query segment data, thereby obtaining a first deep feature data set, which helps to extract high-level feature representations of the data.

[0058] In one implementation, the target encrypted data is obtained by searching the fragment hash code, which can reduce the scope that needs to be verified by querying the fragment features, while increasing the speed and ensuring that data will not be leaked.

[0059] In one embodiment, the original data is audio data, and extracting deep feature data of the query segment data by using a preset model includes:

[0060] Open the query segment data from the preset target library to obtain the speech waveform data of the query segment data;

[0061] Removing all parts of the speech waveform data that do not contain actual speech data to obtain multiple speech waveform data segments; the parts that do not contain actual speech are silence segments and background noise segments;

[0062] Calculate the MFCC coefficients of the speech frames in each speech waveform data segment to obtain multiple sets of feature data;

[0063] All group feature data are brought into the CNN model to obtain deep feature data.

[0064] In one implementation, the preset target library is a library that can load voice files and obtain original voice waveforms. The preset target library can be any one of Librosa, PyAudio, Soundfile, TensorFlow IO, Essentia, Kaldi, and Soundfile.

[0065] In one implementation, multiple speech waveform data segments are obtained by removing all parts of the speech waveform data that do not contain actual speech, and all segments that do not contain actual speech parts are selected and deleted, thereby reducing the dimension of subsequent hash generation, while also ensuring that the hash value generated by the query segment data can correspond to the hash value generated by the original data.

[0066] In one implementation, the CNN model consists of 8 layers, of which the first 5 are convolutional layers. Each convolutional layer contains 8 convolutional filters. ReLU is used as the activation function. Batch normalization is performed in convolutional layers 2, 4, and 5 to improve the fitting speed. A dropout regularization method with a value of 0.25 is also introduced to prevent overfitting. The formula for extracting deep feature data is: Among them, V is the depth feature data, m i is the weight of the i-th recurrent convolutional layer and pooling layer, h i is the MFCC coefficient input for the i-th cycle, σ is the bias parameter to prevent overfitting, and n is the number of cycles.

[0067] In one embodiment, the hash generation method specifically includes:

[0068] Sort the deep feature vectors in the deep feature data set by size to obtain a feature sorting set;

[0069] Determine the deep feature vector value corresponding to the median position in the feature sorting set, and record it as the target judgment value;

[0070] With the target judgment value as the judgment condition, all deep feature vectors in the deep feature data set are hashed to obtain the target hash code; the depth feature vector value greater than the target judgment value in the deep feature data set is recorded as 1, and the depth feature vector value less than or equal to the target judgment value is recorded as 0.

[0071] In one implementation, the deep feature vectors in the deep feature dataset are converted into irreversible target hash codes, thereby better protecting the privacy and security of the data, reducing the complexity and dimension of the data, and simplifying subsequent data retrieval.

[0072] In one embodiment, the query feature generation method specifically includes: processing the first deep feature data set by using a second key to obtain a query fragment feature; the second key includes a first random number and a preset matrix;

[0073] For the first depth feature data set, all first depth feature data vectors are converted to a first target dimension by a first random number to obtain a first target feature vector; the first target dimension is the dimension of the first depth feature data vector in the first depth feature data set plus 3;

[0074] The first target feature vector is encrypted by a preset matrix to obtain a first target fragment feature; the number of rows and columns of the preset matrix is ​​the same as the dimension size of the first depth feature data vector in the first depth feature data set.

[0075] In one implementation, the embodiments described below are only used to illustrate the application of the query feature generation method in the present invention, which includes a second key K2{S1, M}, S1 is a first random number, M is a preset matrix, a first deep feature data set T{t1, t2, ..., ti}; all first deep feature data vectors are converted to the first target dimension to obtain the first target feature vector reference formula Then, the first target feature vector is encrypted by a preset matrix to obtain the first target segment feature

[0076] In one embodiment, the method is applied to a client and further includes:

[0077] Extracting the depth feature vector of the original data through a preset model to obtain a second depth feature data set;

[0078] The second deep feature data set obtains the complete hash code and the complete original feature corresponding to the original data through the hash generation method and the complete feature generation method;

[0079] Encrypting the original data using the first key to obtain encrypted data;

[0080] The encrypted data is bound with the complete hash code and the complete original features and uploaded to the cloud server.

[0081] An implementation method, encrypting the original data by using a first key to obtain encrypted data is specifically as follows: the first key includes a preset first sequence, a preset second sequence and a preset third sequence; wherein encrypting the original data by using the first key specifically includes: performing non-overlapping needle division on the original data to obtain M speech frames, each speech frame contains N sampling points; performing discrete cosine transform on each speech frame to obtain a coefficient domain matrix; the size of the coefficient domain matrix is ​​M*N; after performing row interference on the coefficient domain matrix by using the preset first sequence, column scrambling on the interfered coefficient domain matrix by using the preset second sequence to obtain a target coefficient domain matrix; after connecting all target coefficient domain matrices in frame order, performing inverse discrete cosine transform to obtain target data; the target data is a one-dimensional time domain signal; performing bitwise XOR operation on the target data and the preset third sequence to obtain encrypted data.

[0082] In one embodiment, the complete feature generation method specifically includes: processing the second deep feature data set by using a third key to obtain an original complete feature; the third key includes a second random number, a third random number and a preset inverse matrix;

[0083] For the second depth feature data set, all second depth feature data vectors are converted to a second target dimension by using a second random number and a third random number to obtain a second target feature vector; the second target dimension is the dimension of the second depth feature data vector in the second depth feature data set plus 3;

[0084] The second target feature vector is encrypted by a preset inverse matrix to obtain the original complete feature; the number of rows and columns of the preset inverse matrix is ​​the same as the dimension size of the second depth feature data vector in the inverse depth feature data set.

[0085] In one implementation, the following embodiment is only used to illustrate the application of the query feature generation method of the present invention, which includes the third key K2 {S2, S3, M -1}, S2 is the second random number, S3 is the third random number, M -1 is the preset inverse matrix, the second deep feature data set R{r1, r2, ..., rn}; all second deep feature data vectors are converted to the second target dimension to obtain the second target feature vector reference formula Then, the first target feature vector is encrypted by a preset matrix to obtain the first target segment feature

[0086] Based on the same inventive concept, the present invention also provides another privacy protection method for a fully secret database. Figure 2 , Figure 2 A flowchart of another fully secret database privacy protection method provided by an embodiment of the present invention, the method is applied to a cloud server:

[0087] S201, obtaining a query fragment hash code, and searching a preset database for an encrypted data set corresponding to the query fragment hash code.

[0088] S202, obtaining query fragment features, calculating the target distance between the query fragment features and the complete original features corresponding to each encrypted data in the encrypted data set through Euclidean distance, determining the encrypted data corresponding to the minimum target distance as target encrypted data, and sending the target encrypted data to the client.

[0089] Among them, the encrypted data set includes encrypted data, complete hash codes and complete original features;

[0090] In one implementation, the Euclidean distance is calculated to calculate the target distance between the query fragment feature and the complete original feature corresponding to each encrypted data in the encrypted data set. This can find the original data most similar to the query fragment, avoiding the need for customers to download all the data first and then decrypt it before retrieving it when they need to retrieve encrypted data stored in the cloud, thereby improving efficiency.

[0091] Based on the same inventive concept, the present invention also provides a privacy protection device for a fully encrypted database. Figure 3 , Figure 3 A schematic diagram of the structure of a fully secret database privacy protection device provided by an embodiment of the present invention, the device is applied to a client:

[0092] A first deep feature generation module, configured to extract a deep feature vector of the query fragment data through a preset model to obtain a first deep feature data set;

[0093] A first deep feature processing module, used for obtaining a query fragment hash code and a query fragment feature corresponding to the query fragment data by using a hash generation method and a query feature generation method for the first deep feature data set;

[0094] A query module is used to upload the query segment hash code and the query segment feature to the cloud server; so that the cloud server can search the encrypted data set corresponding to the query segment hash code in the preset database according to the query segment hash code, and for each encrypted data, determine the target encrypted data according to the query segment feature and the complete original feature corresponding to the encrypted audio data, and send the target encrypted data to the client; the target encrypted data is the original data encrypted by the first key; the query segment data is the segment in the original data;

[0095] The decryption module is used to obtain target encrypted data and decrypt the target encrypted data using a first key to obtain original data.

[0096] Based on the same inventive concept, the present invention also provides another fully confidential database privacy protection device. Figure 4 , Figure 4 A schematic diagram of the structure of another fully secret database privacy protection device provided by an embodiment of the present invention, the device is applied to a cloud server:

[0097] In one embodiment, the device is applied to the client and further includes:

[0098] The encrypted data set search module is used to obtain the query fragment hash code and search for the encrypted data set corresponding to the query fragment hash code in the preset database; the encrypted data set includes the encrypted data, the complete hash code and the complete original features;

[0099] The target encrypted data determination module is used to obtain the query fragment features, calculate the target distance between the query fragment features and the complete original features corresponding to each encrypted data in the encrypted data set through Euclidean distance, determine the encrypted data corresponding to the minimum target distance as the target encrypted data, and send the target encrypted data to the client.

[0100] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A privacy protection method for a fully confidential database, characterized in that: The method is applied on the client side: Extracting deep feature vectors of query fragment data using a preset model to obtain a first deep feature data set; The first deep feature data set obtains a query fragment hash code and a query fragment feature corresponding to the query fragment data by a hash generation method and a query feature generation method; Uploading the query fragment hash code and the query fragment feature to a cloud server; The cloud server searches for an encrypted data set corresponding to the query segment hash code in a preset database according to the query segment hash code, determines target encrypted data according to the query segment features and the complete original features corresponding to the encrypted audio data for each encrypted data, and sends the target encrypted data to the client; the target encrypted data is the original data encrypted by using the first key; The query fragment data is a fragment in the original data; The target encrypted data is obtained, and the target encrypted data is decrypted using the first key to obtain original data.

2. A privacy protection method for a fully secret database according to claim 1, characterized in that: The original data is audio data, and the deep feature data of the query fragment data extracted by the preset model includes: Opening the query segment data from a preset target library to obtain speech waveform data of the query segment data; Removing data from all parts of the speech waveform data that do not contain actual speech to obtain a plurality of speech waveform data segments; the parts that do not contain actual speech are silence segments and background noise segments; Calculate the MFCC coefficients of the speech frames in each speech waveform data segment to obtain multiple sets of feature data; All group feature data are brought into the CNN model to obtain deep feature data.

3. A privacy protection method for a fully secret database according to claim 1, characterized in that: The hash generation method specifically includes: Sort the deep feature vectors in the deep feature data set by size to obtain a feature sorting set; Determine the depth feature vector value corresponding to the median position in the feature sorting set, and record it as the target judgment value; A target hash code is obtained by performing a hash transformation on all the depth feature vectors in the depth feature data set with the target judgment value as the judgment condition; a depth feature vector value greater than the target judgment value in the depth feature data set is recorded as 1, and a depth feature vector value less than or equal to the target judgment value is recorded as 0.

4. A privacy protection method for a fully secret database according to claim 1, characterized in that: The query feature generation method specifically includes: processing the first deep feature data set by using a second key to obtain a query fragment feature; the second key includes a first random number and a preset matrix; For the first depth feature data set, convert all first depth feature data vectors to a first target dimension using the first random number to obtain a first target feature vector; the first target dimension is the dimension of the first depth feature data vector in the first depth feature data set plus 3; The first target feature vector is encrypted by a preset matrix to obtain a first target fragment feature; the number of rows and columns of the preset matrix is ​​the same as the dimension size of the first depth feature data vector in the first depth feature data set.

5. A privacy protection method for a fully secret database according to claim 1, characterized in that: The method is applied to the client and further includes: Extracting the depth feature vector of the original data through a preset model to obtain a second depth feature data set; The second deep feature data set obtains a complete hash code and a complete original feature corresponding to the original data by a hash generation method and a complete feature generation method; Encrypting the original data using the first key to obtain encrypted data; The encrypted data is bound to the complete hash code and the complete original feature and then uploaded to the cloud server.

6. A privacy protection method for a fully secret database according to claim 5, characterized in that: The complete feature generation method specifically includes: processing the second deep feature data set by a third key to obtain an original complete feature; the third key includes a second random number, a third random number and a preset inverse matrix; For the second depth feature data set, convert all second depth feature data vectors into a second target dimension by using the second random number and the third random number to obtain a second target feature vector; the second target dimension is the dimension of the second depth feature data vector in the second depth feature data set plus 3; The second target feature vector is encrypted by a preset inverse matrix to obtain the original complete feature; the number of rows and columns of the preset inverse matrix is ​​the same as the dimension size of the second depth feature data vector in the inverse depth feature data set.

7. A privacy protection method for a fully secret database according to claim 1, characterized in that: The method is applied to cloud servers: Obtain the query fragment hash code, and search a preset database for an encrypted data set corresponding to the query fragment hash code; the encrypted data set includes encrypted data, a complete hash code, and complete original features; The query fragment feature is obtained, and the target distance between the query fragment feature and the complete original feature corresponding to each encrypted data in the encrypted data set is calculated by Euclidean distance, the encrypted data corresponding to the minimum target distance is determined and recorded as the target encrypted data, and the target encrypted data is sent to the client.

8. A privacy protection device for a fully secret database, characterized in that: The device is applied to the client and includes: A first deep feature generation module, configured to extract a deep feature vector of the query fragment data through a preset model to obtain a first deep feature data set; A first deep feature processing module, configured to obtain a query fragment hash code and a query fragment feature corresponding to the query fragment data by using a hash generation method and a query feature generation method for the first deep feature data set; A query module, used for uploading the query fragment hash code and the query fragment feature to a cloud server; The cloud server searches for an encrypted data set corresponding to the query segment hash code in a preset database according to the query segment hash code, determines target encrypted data according to the query segment features and the complete original features corresponding to the encrypted audio data for each encrypted data, and sends the target encrypted data to the client; the target encrypted data is the original data encrypted by using the first key; The query fragment data is a fragment in the original data; The decryption module is used to obtain the target encrypted data and decrypt the target encrypted data using the first key to obtain the original data.

9. A privacy protection device for a fully secret database, characterized in that: The device is applied to a cloud server: An encrypted data set search module is used to obtain the query fragment hash code and search for an encrypted data set corresponding to the query fragment hash code in a preset database; the encrypted data set includes encrypted data, a complete hash code and complete original features; The target encrypted data determination module is used to obtain the query fragment feature, calculate the target distance between the query fragment feature and the complete original feature corresponding to each encrypted data in the encrypted data set through Euclidean distance, determine the encrypted data corresponding to the minimum target distance as the target encrypted data, and send the target encrypted data to the client.

Citation Information

Patent Citations

  • Privacy information retrieval system and privacy information retrieval method

    CN114036565A

  • Matching quantity hidden encryption multi-mapping method and system based on layered structure

    CN117972742A