A layered privacy protection and communication efficient compressed sensing data encoding method

Abstract

The application claims a layered privacy protection and communication efficient compressed sensing data coding method, belonging to the technical field of computer vision, and specifically comprising the following steps: obtaining sensitive information and non-sensitive information of a face image of two images; after dividing sensitive areas and non-sensitive areas of the two image data, using two blur vectors to blur the sensitive areas respectively to obtain blurred sensitive areas; using a compressed sensing algorithm to sample the blurred sensitive areas and the non-sensitive areas to obtain measurement values of the two images; fusing the obtained measurement values of the two images to obtain fused measurement values; the compressed sensing algorithm comprises a problem definition step, an encoding step, a controllable operation step and a decoding step. The left zero matrix of the application is helpful to realize efficient communication data coding. Meanwhile, semi-authorized users and authorized users can respectively recover signals of different qualities from the two signals.

Description

Technical Field

[0001] This invention belongs to the field of computer vision technology, specifically to a data encoding method. Background Technology

[0002] In intelligent surveillance applications, with technological advancements and increasing demand, sensors collect vast amounts of data daily in buildings and public areas. This data includes not only visual information such as images and videos, but also various sensor data, such as temperature, humidity, and light intensity. This data is crucial for monitoring and managing these areas, but it also presents challenges related to data privacy and communication. First, the large volume of data collection and processing raises privacy concerns. Personal privacy is a critical concern in intelligent surveillance applications. Improper use or leakage of this data could threaten personal privacy. Therefore, protecting the privacy of this data is a critical issue. Furthermore, the large amounts of data collected by sensors require transmission and processing. Without compression, transmission and processing consume significant bandwidth and computing resources. Fortunately, compressed sensing technology offers a preliminary method for protecting data privacy and can effectively reduce data size and storage requirements. Compressed sensing has significant application value in intelligent surveillance, providing initial protection for data privacy and alleviating the communication burden. However, to achieve higher levels of privacy protection and communication efficiency, further research and application of other technologies and methods are needed.

[0003] Regarding privacy concerns, data is typically considered to consist of a privacy domain and a public domain. The privacy domain contains sensitive information, while the public domain contains non-sensitive information. The goal of privacy-preserving data encoding design is to encrypt sensitive areas of data without compromising non-sensitive areas. Traditional strong cryptographic methods (AES and DES) offer strong security, but these methods are expensive. More importantly, encrypting the entire data is neither useful nor necessary in practical multimedia applications. Therefore, only sensitive areas deemed to carry privacy information need protection, while the entire data can be compressed and encrypted simultaneously using compression sensing. This layered privacy protection corresponds to different decoding operations, resulting in layered recovery. Currently, multi-level privacy protection schemes have been implemented, where semi-authorized users and authorized users at the decoding end recover data with different privacy levels, thus achieving layered recovery. Semi-authorized users can only access partial data or have lower access levels, while authorized users have higher access levels and can access more complete and sensitive data. Through this layered recovery strategy, we can better manage the access permissions of different users, ensuring data security and privacy. However, current multi-level privacy protection schemes suffer from high computational complexity in recovery. Therefore, this patent designs a novel encoding operation to alleviate this problem. Regarding communication, existing schemes, apart from compressed sensing, have not further improved communication efficiency. To fill this gap, this patent uses a left-zeroing matrix to obtain data of various recovery qualities from measurements through controllable operations.

[0004] Specifically, the two signals are processed together, and two ambiguity vectors are used to encrypt the sensitive regions of each signal while preserving the non-sensitive regions. The resulting ambiguous signals are then compressed, encrypted, and further fused to obtain the measurement values. For both semi-licensed and licensed users, the left-zeroing matrix can recover different quality versions of the two signals from the fused measurement values, which further enables efficient data encoding for communication. The contributions of this patent are as follows:

[0005] (1) Before performing compressed sensing, we blurred sensitive areas to ensure data privacy. This blurring effectively hides sensitive information and prevents unauthorized access and use. Since sensitive areas are blurred while non-sensitive areas are preserved, we can consider a layered recovery strategy at the decoding end.

[0006] (2) Both semi-authorized and authorized users can use the left nullification matrix to recover different quality versions of the two signals. This means that we can provide different quality versions of data according to the needs and permission levels of different users. Summary of the Invention

[0007] This invention aims to solve the problems of the prior art mentioned above. It proposes a layered, privacy-preserving, and communication-efficient compressed sensing data encoding method. The technical solution of this invention is as follows:

[0008] A layered, privacy-preserving, and communication-efficient compressed sensing data encoding method includes the following steps:

[0009] Extract sensitive and non-sensitive information from the facial images of two images;

[0010] After separating the sensitive and non-sensitive regions from the two image data, the sensitive regions are blurred using two blur vectors to obtain the blurred sensitive regions.

[0011] The compressed sensing algorithm is used to sample the blurred sensitive and non-sensitive regions to obtain the measurement values ​​of two images; the measurement values ​​of the two images are fused to obtain the fused measurement value; the compressed sensing algorithm includes a problem definition step, an encoding step, a controllable operation step, and a decoding step.

[0012] Furthermore, the problem definition steps specifically include: assuming that signals x1 and x2 are both composed of sensitive and non-sensitive regions; and that the lengths of signals x1 and x2 are N, i.e. The length of the set of real numbers is N. Furthermore, the expressions for x1 and x2, which can both be represented by orthogonal sums, are as follows:

[0013]

[0014]

[0015] in Indicates orthogonal sum; and This represents the sensitive regions of signals x1 and x2; The length of the set of real numbers is Λ 1_s , The length of the set of real numbers is . and x 2_s By using the index set Λ that is not corresponding 1_s and Λ 2_s The result obtained after setting the coefficients of index x1 and x2 to zero and clearing them; and These are the non-sensitive regions of signals x1 and x2, respectively. and The non-zero coefficients are respectively indexed by the index set Λ 1_n and Λ 2_n Obviously, |Λ 1_s |+|Λ 1_n |=N and|Λ 2_s |+|Λ 2_n|=N, where N represents the lengths of signals x1 and x2.

[0016] Furthermore, the step of using two blur vectors to blur the sensitive region to obtain the blurred sensitive region specifically includes:

[0017] Ambiguous signals and It is obtained through the following formula:

[0018]

[0019]

[0020] in the formula and It is a fuzzy vector that perturbs the indexed set Λ. 1_s and Λ 2_s The sensitive regions of the indexed signals x1 and x2; and All elements are '1', and they are respectively indexed by the set Λ 1_n and Λ 2_n Index; Operator It represents the Hadamadaji, and and

[0021] Furthermore, the encoding step specifically includes:

[0022] Encoding operations utilize fuzzy vector r 1_s and r 2_n The sensitive regions of signals x1 and x2 are blurred respectively. 1_s and x 2_s The non-sensitive regions x1 and x2 of signals x2 1_n and x 2_n It remains unchanged; then the signal becomes blurred. and Simultaneously sampled, compressed, and encrypted, the resulting measurement results are presented as follows:

[0023]

[0024]

[0025] in and These are fuzzy signals. and The measured value, and It is a matrix The sub-column matrix, where M and N represent the dimensions of the measurement and the signal, respectively, with M < N, and their columns are indexed by the set Λ. 1_s and Λ 1_n Index and Λ 1_s +Λ 1_n =N; similarly, and It is a matrix The sub-column matrix, and their columns are respectively indexed by the set Λ 2_s and Λ 2_n Index and Λ 2_s +Λ 2_n =N; Chaotic sequence and It is through the initial values ​​(u1, v1, c) of the chaotic system 1_0 ),(u2,v2,c 2_0 ),(u3,v3,c 3_0 ) and (u4,v4,c 4_0 The matrix A1 and A2 are generated by cascaded chaotic systems, where M < L ≤ 2M, and L represents the number of rows in matrices A1 and A2. Furthermore, these chaotic systems are rearranged and orthogonalized to obtain matrices. and Matrix B1, B2, A1, and A2 are respectively obtained through and The obtained fuzzy vector r 1_s and r 2_s It is obtained by performing the Hadamard product between multiple chaotic sequences.

[0026] Furthermore, the controllable computation steps specifically include:

[0027] Before decoding, perform controllable operations; given and The left-zeroing matrices F1 and F2 can achieve controllable recovery from the fused measurement value y, and the set of row subsets of the left-zeroing matrices F1 and F2 is represented as Ω1={F 11 ,...,F 1k ,...,F 1m} and Ω2={F 21 ,...,F 2k ,...,F 2m}, 1≤k≤m, C represents permutations and combinations. Left-zeroed matrices F1 and F2 belong to sets Ω1 and Ω2; F 1k A1 = 0 and F 2k A2 = 0, Where C·S·log(N / S)≤D≤(LM);

[0028]

[0029]

[0030] Furthermore, the decoding step specifically includes:

[0031] Semi-licensed and licensed users receive the fused measurement value y, and use controllable operations to select the signal to be recovered and its quality; using the controllable factor F... 2k To restore the corresponding semi-authorized users and authorized users And x1:

[0032]

[0033] In the above formula s1 is x1 in the orthogonal transformation basis For semi-authorized users, the key (u1, v1, c) 1_0 ) and keys (u3, v3, c 3_0 ) is accessible, i.e., F 2k A1 and B1 are available; since semi-authorized users cannot access the fuzzy vector r1, they can, according to... recover ∈ represents error. This represents the recovered sparse signal. Next, the result of minimizing the l1 norm is given. It can be used to obtain the estimated value of the accompanying fuzzy vector signal x1. It can be accessed via get;

[0034] Next, the result of minimizing the l1 norm. It can be used to obtain the estimated value of the accompanying fuzzy vector signal x1. It can be accessed via get;

[0035] Authorized users possess all keys (u1, v1, c) 1_0 ), (u3,v3,c 3_0 ), (u5,v5,c 5_0 ), (u6,v6,c 6_0 ) and (u7,v7,c 7_0 That is, authorized users can obtain A1, B1, F 2k and r1:

[0036]

[0037] Authorized users can use the l1 norm minimization strategy Restore s1; then signal x1 can pass through Obtain; similarly, x2 can be obtained by left-zeroing the matrix F 1k get.

[0038] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements a layered privacy-preserving and communication-efficient compressed sensing data encoding method as described in any one of the claims.

[0039] A non-transitory computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements a layered privacy-preserving and communication-efficient compressed sensing data encoding method as described in any one of the claims.

[0040] A computer program product includes a computer program that, when executed by a processor, implements a layered privacy-preserving and communication-efficient compressed sensing data encoding method as described in any one of the claims.

[0041] The advantages and beneficial effects of this invention are as follows:

[0042] 1. Obtaining Sensitive and Non-Sensitive Information: Sensitive and non-sensitive information about the face image is extracted from the two images using image processing techniques. This step provides the foundational data for subsequent privacy protection and compressed sensing.

[0043] 2. Separate sensitive and non-sensitive regions: Analyze the two image datasets to determine the sensitive and non-sensitive regions. This process may involve using facial recognition technology or image segmentation algorithms.

[0044] 3. Sensitive Area Blur: Two blur vectors are used to blur sensitive areas. The purpose of this step is to obfuscate sensitive information to protect personal privacy. At the same time, the blurred sensitive areas still retain certain image features.

[0045] 4. Compressed Sensing Sampling: Compressed sensing algorithms are used to sample blurred sensitive and non-sensitive areas, obtaining measurements from two images. This step aims to reduce data volume, decrease transmission bandwidth, and ensure data privacy.

[0046] 5. Merge Measurements: The measurements from the two images are merged to obtain a fused measurement. This step helps integrate the information from the two images for further processing or transmission.

[0047] 6. Design of Compressed Sensing Algorithm: This method covers the entire process of compressed sensing algorithm, including problem definition, encoding, controllable computation, and decoding steps. This algorithm is applied to sampling sensitive and non-sensitive areas, thereby reducing data dimensionality while ensuring privacy.

[0048] 7. Application of the Left-Nullification Matrix: The left-nullification matrix mentioned in the invention facilitates efficient data encoding for communication. It allows semi-licensed and licensed users to recover signals of different qualities from two signals. This means that the technology can flexibly meet the signal recovery quality needs of different users while ensuring communication efficiency.

[0049] The ingenuity of this invention lies in its integration of technologies from multiple fields, including image processing, privacy protection, and compressed sensing. Through steps such as blurring sensitive information, compressed sensing sampling, and information fusion, it protects the privacy of sensitive information while maintaining high communication efficiency and enabling flexible responses to different user needs. Its benefits include privacy protection, reduced bandwidth requirements for data transmission, and continued high communication efficiency and flexibility. This method achieves a good balance between information security and communication efficiency, which is of great significance for protecting personal privacy and providing flexible data recovery capabilities. Attached Figure Description

[0050] Figure 1 This is a general framework diagram of the preferred embodiment of the method provided by the present invention. Detailed Implementation

[0051] The technical solutions of the embodiments of the present invention will be clearly and thoroughly described below with reference to the accompanying drawings. The described embodiments are merely some embodiments of the present invention.

[0052] The technical solution of the present invention to solve the above-mentioned technical problems is:

[0053] As attached Figure 1 As shown, a layered privacy-preserving and communication-efficient compressed sensing data encoding method includes the following steps:

[0054] (1) Use facial recognition technology to determine the sensitive and non-sensitive information of the face images in the two images;

[0055] (2) After separating the sensitive and non-sensitive regions of the two image data, the sensitive regions are blurred by two blur vectors to obtain the blurred sensitive regions.

[0056] (3) Compressed sensing is used to sample the blurred sensitive and non-sensitive areas to obtain the measurement values ​​of the two images;

[0057] (4) In order to further improve the transmission efficiency of communication coded data, the two measured values ​​are fused to obtain a fused measured value;

[0058] (5) During recovery, first determine which signal the user wants to recover, and then determine the quality of the recovered signal. This is determined by the controllability factor F. 1k and F 2k It's confirmed.

[0059] A layered, privacy-preserving, and communication-efficient compressed sensing data encoding method is proposed. Compressed sensing-based data encoding is achieved through encoding, controllable computation, and decoding.

[0060] This invention is mainly divided into four modules. The first module is problem definition, which aims to describe the problem in more detail and clearly so that subsequent modules can better understand and process it. The second module is encoding, which represents the raw data as a shorter encoded sequence to reduce the need for storage space or transmission bandwidth, thereby achieving data compression. This encoded sequence requires less storage space or transmission bandwidth, thus saving resources and improving efficiency. The third module is controllable computation, which aims to recover data signals of different qualities according to different client permissions. The fourth module is decoding, which is the process of restoring the encoded data to its original form for subsequent processing and utilization.

[0061] First, this patent defines the problem. Assume a signal... These signals are all composed of sensitive and non-sensitive regions. Furthermore, they can all be represented by orthogonal sums. Specifically, sensitive and non-sensitive regions can be represented as specific parts of an orthogonal sum. In this way, we can more accurately describe the characteristics and structure of the signals, thus providing a more accurate basis for subsequent processing and analysis.

[0062]

[0063]

[0064] in It represents an orthogonal sum. and This represents the sensitive regions of signals x1 and x2. It's worth noting that x... 1_s and x 2_s By using the index set Λ that is not corresponding 1_s and Λ 2_s This is obtained by setting the coefficients of index x1 and x2 to zero and clearing them. Similarly, and These are the non-sensitive regions of signals x1 and x2, respectively. and The non-zero coefficients are respectively indexed by the index set Λ 1_n and Λ 2_n Obviously, |Λ 1_s |+|Λ 1_n |=N and|Λ 2_s |+|Λ 2_n |=N.

[0065] Let's take an example to better understand orthogonal summation. A signal x1 = (100, 80, 55, 75, 95, 110) can be divided into sensitive regions x... 1_s Non-sensitive area x 1_n If x 1_s =(100,80,110) and Λ 1_n = (1, 2, 6), then x 2_s = (80, 55, 75, 95) and Λ 2_n = (3,4,5).

[0066] Ambiguous signals and It is obtained through the following formula:

[0067]

[0068]

[0069] In the above formula and It is a fuzzy vector that perturbs the indexed set Λ. 1_s and Λ 2_s The sensitive regions of the indexed signals x1 and x2. and All elements are '1', and they are respectively indexed by the set Λ 1_n and Λ 2_n Index. Operators It represents the Hadamadaji, and and

[0070] Definition 1: The encoding operation E(·), controllable operation C(·), and dual-layer decoding operations D1(·) and D2(·) involved in this patent.

[0071] Encoding operation E(·): To protect sensitive data from leakage, obfuscation vectors are used to obfuscate these regions. However, simply obfuscating sensitive regions is insufficient. To further protect data security, compressed sensing technology is introduced. Compressed sensing not only compresses data, reducing storage and transmission requirements, but also encrypts both sensitive and non-sensitive areas of the data simultaneously. By combining obfuscation vectors and compressed sensing, the security of both sensitive and non-sensitive data areas can be protected more effectively.

[0072] E(x1,x2,r1,r2)=E(x1,r1)+E(x2,r2)=y1+y2=y (5)

[0073] Where y1=E(x1,r1) and y2=E(x2,r2) represent the measured values ​​of signals x1 and x2, respectively. y=E(x1,x2,r1,r2) are the fused measured values.

[0074] Controllable operation C(·): C(·) = F×(·) determines which signal, x1 or x2, is recovered and the quality of its recovery.

[0075]

[0076] in This represents a controllable measurement value. When F 1k When = F, the above formula can be written as F 1k (y1+y2)=F 1k E(x2,r2). Similarly, when F 2k =F, F 2k (y1+y2)=F 2k E(x1,r1). F 1k and F 2k This will be described in detail in the section on controllable operations later.

[0077] Two-layer decoding operations D1(·), D2(·): Assuming the recovered signal x1, the decoding operations D1(·), D2(·) for semi-licensed users and licensed users are described as follows:

[0078]

[0079] ‖D2(F2E(x1,r1)-x1‖2≤κ‖e‖2 (8) In the above formula, e represents additive perturbation noise.

[0080] Next, the three operations described above will be described in detail: encoding operation E(·), controllable operation C(·), and two-layer decoding operations D1(·) and D2(·).

[0081] A. Encoding operation E(·)

[0082] The encoding operation proposed in this patent utilizes the fuzzy vector r 1_s and r 2_n The sensitive regions of signals x1 and x2 are blurred respectively. 1_s and x 2_s The non-sensitive regions x1 and x2 of signals x2 1_n and x 2_n It remains unchanged, as shown in formulas (3) and (4) above. Then the blurred signal (fuzzy signal) is... and Simultaneously sampled, compressed, and encrypted, the resulting measurement results are presented as follows:

[0083]

[0084]

[0085] in and These are fuzzy signals. and The measured value, and It is a matrix The sub-column matrix, and their columns are respectively indexed by the set Λ 1_s and Λ 1_n Index and Λ 1_s +Λ 1_n =N. Similarly, and It is a matrix The sub-column matrix, and their columns are respectively indexed by the set Λ 2_s and Λ 2_n Index and Λ 2_s +Λ 2_n =N. Chaotic sequence and It is through the initial values ​​(u1, v1, c) 1_0 ),(u2,v2,c 2_0 ),(u3,v3,c 3_0 ) and (u4,v4,c 4_0 This is generated by cascaded chaotic systems. Furthermore, these chaotic systems, after rearrangement and orthogonalization, yield matrices. and Matrix B1, B2, A1, and A2 are respectively obtained through and The obtained fuzzy vector r 1_s and r 2_s It is obtained by performing the Hadamard product between multiple chaotic sequences. Clearly, the more chaotic sequences in the Hadamard product, the worse the quality of visual recovery of sensitive areas for semi-authorized users. This patent uses initial values ​​of (u5, v5, c). 5_0 ), (u6,v6,c 6_0 ), (u7,v7,c 7_0 ), and(u 10 ,v 10 ,c 10_0 The fuzzy vector r is obtained by using the Hadamard product among the six chaotic sequences. 1_s and r 2_s .

[0086] To further reduce the transmission volume, the measured values ​​y1 and y2 are fused to obtain the fused measured value. Right now:

[0087]

[0088] The fused measurement value y can be transmitted to the user for decoding via a wireless communication channel.

[0089] B. Controllable operation C(·)

[0090] Before decoding, controllable operations are performed to further improve communication efficiency by increasing the amount of data transmitted. Given... and The left-zeroing matrices F1 and F2 can achieve controllable recovery from the fused measurement value y, and the set of row subsets of the left-zeroing matrices F1 and F2 can be represented as Ω1={F 11 ,...,F 1k ,...,F 1m}(1≤k≤m) and Ω2={F 21 ,...,F 2k ,...,F 2m In the first two formulas, m can be expressed as m = ∑i = 1 L-M CL-M i It is worth noting that the left-zeroed matrices F1 and F2 belong to sets Ω1 and Ω2. Since the row submatrices of left-zeroed matrices F1 and F2 with the same number of rows have the same recovery quality, and there are at most (LM) recovery qualities, it is clear that F... 1k A1 = 0 and Where C·S·log(N / S)≤D≤(LM).

[0091]

[0092]

[0093] C. Two-layer decoding operations D1(·), D2(·)

[0094] Once both the semi-licensed and licensed users receive the fused measurement value y, controllable computation is used to select the signal to be recovered and its quality. Here, a controllable factor F is used. 2k To restore the corresponding semi-authorized users and authorized users And x1:

[0095]

[0096] In the above formula s1 is x1 in the orthogonal transformation basis For semi-authorized users, the key (u1, v1, c) 1_0 ) and keys (u3, v3, c 3_0 ) is accessible, i.e., F 2kA1 and B1 are available. Since semi-authorized users cannot access the fuzzy vector r1, they can, according to... recover Next, the result of minimizing the l1 norm. It can be used to obtain the estimated value of the accompanying fuzzy vector signal x1. It can be accessed via get.

[0097] Authorized users possess all keys (u1, v1, c) 1_0 ), (u3,v3,c 3_0 ), (u5,v5,c 5_0 ), (u6,v6,c 6_0 ) and (u7,v7,c 7_0 That is, authorized users can obtain A1, B1, F 2k and r1:

[0098]

[0099] Authorized users can use the l1 norm minimization strategy Restore s1. Then signal x1 can pass through. To obtain. Similarly, x2 can be obtained by left-zeroing the matrix F 1k get.

[0100] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, a computer can be, for example, a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email device, game console, tablet computer, wearable device, or any combination of these devices.

[0101] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, 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, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0102] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0103] The above embodiments should be understood as illustrative only and not as limiting the scope of protection of the present invention. After reading the description of the present invention, those skilled in the art can make various alterations or modifications to the present invention, and these equivalent changes and modifications also fall within the scope defined by the claims of the present invention.

Claims

1. A layered privacy protection and communication efficient compressive sensing data encoding method, characterized in that, The method comprises the following steps: obtaining sensitive information and non-sensitive information of a face image of two images; after separating sensitive areas and non-sensitive areas of the two images, blurring the sensitive areas by using two blur vectors to obtain blurred sensitive areas; using a compressive sensing algorithm to sample the blurred sensitive areas and the non-sensitive areas to obtain measurement values of the two images; fusing the obtained measurement values of the two images to obtain fused measurement values; the compressive sensing algorithm comprises a problem definition step, an encoding step, a controllable operation step and a decoding step; the encoding step specifically comprises: The encoding operation makes use of a blurring vector and respectively to blur the signal and in the sensitive regions and , the blurring vector and being used for the non-sensitive regions, while the non-sensitive regions and of the signal and are left unchanged; the resulting blurred signals and are then simultaneously sampled, compressed and encrypted, the resulting measurements being represented as follows: ； ； r1, r2 denote the total blur vector, the blur vector and for sensitive areas, wherein and are the measured values of the blur signal and , and are subcolumn matrices of the matrix , denote the dimensions of the measured values and the signal, respectively, and their columns are indexed by the index sets and , respectively, and ; Similarly, and are sub-column matrices of matrix , and their columns are indexed by sets and respectively, and ; chaotic sequences , , and are generated by a cascaded chaotic system with initial values and , and the chaotic system is rearranged and orthogonalized to obtain matrix , , and ; matrices , , and are obtained by , , and respectively, and sensitive region fuzzy vectors and are obtained by performing Hadamard product between the above-mentioned chaotic sequences, , and L represents the number of rows of matrix . 2.The layered privacy protection and communication efficient compressed sensing data encoding method of claim 1, wherein, The problem defining step specifically includes assuming that the signals and are composed of sensitive regions and non-sensitive regions; the lengths of the signals and are N, i.e. , , represent the length of the real set N, and and can be represented by orthogonal sums. 3.The layered privacy preserving and communication efficient compressive sensing data encoding method of claim 2, wherein, The and can be expressed by the expressions of the orthogonal sum and are specifically: (1) (2) wherein denotes the orthogonal sum; and denotes the signal and sensitive region of the signal denotes the length of the real set , denotes the length of the real set ; and are obtained by zeroing and removing the coefficients of the signal and indexed by the index set and ; and are the non-sensitive regions of the signals and , and non-zero coefficients of the signals and are indexed by the index set and , wherein N denotes the length of the signals and .

4. The layered privacy-preserving and communication-efficient compressive sensing data encoding method of claim 3, wherein, the blurring of the sensitive areas by using two blur vectors to obtain blurred sensitive areas specifically comprises: The obtained blur signal and is obtained by the following equation: In the formula and is the sensitive region blur vector, which perturbs the indexed set and indexed signal and of the sensitive region; and are all '1', and they are indexed by the set and respectively; the operator denotes the Hadamard product, and and .

5. The layered privacy preserving and communication efficient compressive sensing data encoding method of claim 1, wherein, the controllable operation step specifically comprises: Controllable operations are performed before decoding; given a left-annihilating matrix and ; and Controllable recovery from fused measurements is possible, and and The set of row subsets of and , , C is a permutation, the left-annihilating matrix and belongs to the set and ; and where ; S represents sparsity, and D represents the number of rows of a left null matrix.

6. The layered privacy-preserving and communication-efficient compressive sensing data encoding method of claim 5, wherein, the decoding step specifically comprises: Semi-licensed users and licensed users receive fused measurement values , with controllable operation to select signals and their quality to restore; using controllable factors to restore and : ； In the above formula , , yes In orthogonal transformation basis sparse representation; Represents a fuzzy vector, and for a semi-authorized user, the initial value of the chaotic system. and It is accessible, that is... and It is available; semi-authorized users are subject to recover ; Indicates error. This represents the recovered sparse signal, followed by the recovered sparse signal. It can be used to obtain the accompanying fuzzy vector The estimated value It can be accessed through get; The authorized user has the initial value of the chaotic system , , the key , and , i.e. the authorized user can obtain and : ； Authorized users can Norm minimization strategy recover Then the signal It is possible Obtain; similarly, and The matrix can be left-zeroed. get.

7. An electronic device, comprising: The computer program is executed by the processor to realize the layered privacy protection and communication efficient compressive sensing data encoding method of any one of claims 1 to 6.

8. A non-transitory computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to realize the layered privacy protection and communication efficient compressive sensing data encoding method of any one of claims 1 to 6.

9. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to realize the layered privacy protection and communication efficient compressive sensing data encoding method of any one of claims 1 to 6.

Images