A processing method for generating picture data for a patrol process

Abstract

The application belongs to the field of industrial internet, and particularly relates to a processing method for picture data generated in a patrol inspection process during power station operation and maintenance, comprising the following steps: an edge device collects image data, image data of all channels of the image is disassembled to obtain an optimal window matrix size; scanning disassembly is performed using the optimal window matrix size, arithmetic coding is performed according to the appearance order of each element in the scanned matrix set; the already compressed data is segmented and encrypted; RSA decryption is performed and comparison verification is performed; decompression is performed; post-processing is performed. The method scans each channel numerical matrix of picture data generated in the patrol inspection process to obtain a set of a series of different matrices capable of expressing the original matrix through splicing, the appearance frequency of each element in the set is counted, an arithmetic coding mode is used to find a real number to express the original matrix, and lossless compression of data and data decompression in the subsequent reverse process are realized.

Description

Technical Field

[0001] This application belongs to the field of industrial internet, specifically involving a method for processing image data generated during the inspection process in power plant operation and maintenance. Background Technology

[0002] With the rapid development of 5G applications, although most power plants have achieved comprehensive 5G signal coverage in their management interfaces, in the production interface, the weak penetration of 5G signals makes it difficult to achieve signal coverage in some special physical locations. Therefore, data communication in these scenarios requires effective data compression before transmission to improve information transmission efficiency. However, for some critical and sensitive data, lossy compression methods often lead to information loss. Meanwhile, the increasingly widespread application of wireless network communication technology in power plant operation and maintenance scenarios also presents serious challenges to data security in the transmission of daily inspection and maintenance data.

[0003] Power plant operation and maintenance includes the operation and maintenance of power plants and substations. The networks used during inspections have low coverage, poor signal quality, and unstable connection status. The image data generated during power plant inspections often have a narrow color gamut and limited color richness. Since this type of image data is mainly used to record the appearance of equipment and facilities, existing technologies for transferring unstructured data, especially image data, generated during inspections between edge devices and data centers mainly rely on data encryption.

[0004] Existing patent applications, such as Chinese invention patent CN114928392A, published on August 19, 2022, entitled "A Real-time Data Feedback Method and System for Automatic Inspection by UAVs Based on 5G," disclose a method for real-time data feedback for automatic inspection by UAVs based on 5G, comprising the following steps: S1, UAV conducts inspection and acquires real-time inspection data; S2, the inspection data is encrypted; S3, the encrypted data and key are transmitted to the cloud platform via 5G, and the cloud platform decrypts the data to obtain the original inspection data; S4, the original inspection data is transmitted to the measurement and control analysis terminal via the 5G intranet to obtain measurement and control data, and the cloud platform feeds back the measurement and control data to the UAV, which then performs targeted inspections based on the data. However, the aforementioned patent does not design the necessary data compression / decompression process and method for efficient data transmission between edge devices and the data platform. Summary of the Invention

[0005] This application aims to solve the above-mentioned problems existing in the prior art, and hereby proposes a method for processing image data generated during the inspection process.

[0006] To achieve the above objectives, the technical solution of the present invention is as follows: A method for processing image data generated during an inspection process includes the following steps: Step S1. The edge device acquires image data and decomposes the image data for all channels of the image using window matrices of multiple sizes to obtain the optimal window matrix size; Image data decomposition refers to processing data through an m x n window matrix. The process of performing layered scanning on each channel of the original image, where m and n are the number of rows and columns of the matrix, will decompose the original image into multiple non-overlapping sub-matrices after the process is completed. For images where the number of rows / columns of the original image is not an integer multiple of the number of rows or columns of the window matrix, each channel is filled with 0 values ​​up to the last row or last column. The number of rows that need to be padded with zeros in the row direction is The number of columns that need to be padded with zeros in the column direction is n. .

[0007] Using window matrix When scanning each channel of the original image in layers, the sliding offset in the row direction is m, and the sliding offset in the column direction is n.

[0008] During the scan, if the original image's single-channel matrix has A rows and B columns, then the matrix... The number of scans in the column direction of a single channel of the image is The number of scans in the row direction is i. The number of matrices decomposed from this channel is i. .

[0009] Using matrices After all channels of the original image have been scanned layer by layer, the set of matrices decomposed by each scan is denoted as . , Generate a set by taking the unique value of all elements. and to Count each element to generate And record i, j and ;in After scanning all channels of an image, a series of dimensions are obtained, along with E. m×n A set of identical matrices; To The set of all elements in the set that have a unique value; for The count of each element in the array.

[0010] For any 2,B / 2+1] 2, A / 2+1], using a matrix During layered scanning, record the matrix of each window. The corresponding i, j and A set is formed after multiple scans. ,Pick In any pair of i, j and j that makes the following expression true and maximizes its value The corresponding m and n are taken as the optimal window matrix sizes: .

[0011] Step S2. Use the optimal window matrix size to scan and decompose each channel after zero padding, take the unique value of the scanned matrix set, calculate the probability and cumulative probability of each unique value, and perform arithmetic encoding according to the order of appearance of each element in the scanned matrix set.

[0012] By scanning and decomposing each channel after zero-padding using the optimal window matrix size, we can obtain the optimal result under these conditions. , and .pass Computable The probability values ​​of each element :

[0013] Define a set And calculate its cumulative probability and denot it as ,Right now:

[0014] It is the set of scanned matrices, for The first element in, find its position in The probability value corresponding to Find two consecutive ones in Q. and Make ;Will According to the proportion of each element in Q The proportion is used to redistribute Q and denoted as ,right The second element in, find its in The probability value corresponding to Make ;right Repeat the above operation for each element in the array, and after the operation is completed for the last element, the range can be obtained. Find the real number with the fewest decimal places within this interval, extract its decimal part, convert it to a string, and denot it as [string name missing]. This is the complete compressed data.

[0015] Step S3. Encrypt the compressed data from step S2 in segments.

[0016] Obtain the compressed complete data Afterwards, record The number of decimal places in the string, which is also the length of the string, is and RSA Segmented encryption can be performed, for example, when using RSA for segmented encryption. It needs to be done according to each One character per data partition Split into several plaintext data partitions, The number of binary bits in the modulus of the key pair. The string length is not When the value is a multiple of an integer, use the character "0". Add the last part; after encryption using the RSA public key, a ciphertext data partition is formed, denoted as . This represents the string length count, where n is the number of data partitions. For each plaintext data partition, its hash function value is calculated to generate a set of hash values, denoted as . and calculate The hash value of the string formed by concatenating all elements in the string is denoted as . ;Record Let m be the number of channels during scanning, a, the number of rows i and columns j of all channels scanned.

[0017] Except for the private key in the key pair required for encryption, which is issued offline, the data transmitted from the edge device's data transmission service instance to the data platform includes a set of data from the edge side. , , , and Metadata composed of, and n groups of and , Sector data as a whole: Step S4. The segmented encrypted data from step S3 is transmitted over the network; Step S5. When the server receives sector data, it uses the offline-issued private key to perform RSA decryption on the received ciphertext sector data. After decryption, it calculates the hash function value of the decrypted plaintext and compares it with the plaintext hash function value of the sector sent by the client. If the verification fails, the client is asked to retransmit the sector. If it passes, the plaintext data of the sector is temporarily stored in the position corresponding to its sector sequence number. The number of sectors that have been temporarily stored is determined by the m value in the received metadata. If so, the hash function values ​​of all the temporarily stored sectors are concatenated in order and the hash value is calculated and compared with the hash function value in the metadata. If the verification fails, the client is asked to retransmit. If it passes, decompression is performed. Step S6. Decompress After the data in each sector is decrypted and the hash function value passes verification, the data from each sector is sequentially concatenated. After concatenation, all characters from the 0th to the kth position (inclusive) of the concatenation interface string are taken and converted into a number with an integer part of 0 and a fractional part consisting of the k characters, denoted as S. Then, the metadata is used... Restoring Cumulative Probability And find two consecutive ones within them. Make ,but The first in The term is the first matrix that has been restored. This process is repeated until the number of restored matrices reaches a certain threshold. Stop at this time.

[0018] The restored matrix is ​​spliced ​​together according to the number of channels 'a' during scanning, the number of rows 'i' and the number of columns 'j' in each channel, to complete the lossless decompression process of the compressed data; Step S7. Post-processing: Crop the image.

[0019] The advantages of this application are: 1. This method scans the numerical matrix of each channel of the image data generated during the inspection process to obtain a set of distinct matrices that can be concatenated to represent the original matrix. By statistically analyzing the frequency of each element in this set, an arithmetic encoding method is used to find a real number to represent the original matrix, thereby achieving lossless data compression and subsequent data decompression in the reverse process.

[0020] 2. This method achieves lossless compression and decompression of image data. When network connectivity is unstable or performance is low, it enables breakpoint resumption of data transmission by independently transmitting data from multiple sectors and linking data sending service instances of edge devices with data receiving service instances of the data middle platform. When deploying multiple data sending service instances and multiple data receiving service instances, it can also realize multiple concurrent sector data transmission processes. Not only can each concurrent process achieve breakpoint resumption of data transmission, but it also maximizes the utilization of network bandwidth. Attached Figure Description

[0021] Figure 1 This is the overall flowchart of this application.

[0022] Figure 2 This is a schematic diagram of the matrix scanning process for a single channel of an image.

[0023] Figure 3 This is a diagram of the arithmetic encoding process.

[0024] Figure 4 This is a diagram of the data to be transmitted.

[0025] Figure 5 This refers to the lossless compression process of sector data transmission and restoration.

[0026] Figure 6 This is a schematic diagram of the matrix after zero-padding the R layer in an RGB three-channel image.

[0027] Figure 7 To use one A schematic diagram of image decomposition of the R layer using a window matrix. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of the embodiments of the invention clearer, the technical solutions of the embodiments of the invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the invention, not all embodiments. The components of the embodiments of the invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0029] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0030] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0031] In the description of this invention, it should be noted that the terms "upper," "vertical," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of this invention is in use, or the orientation or positional relationship commonly understood by those skilled in the art. They are only used for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this invention. In addition, the terms "first," "second," etc., are only used to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0032] This invention presents an aircraft surface feature segmentation method based on contour constraint optimization, which uses a deep learning network to segment targets in an image. The method learns feature information from the image through a feature extraction backbone network, then fits the outer contour constraint of the target based on this feature information, initially segmenting the target in the image. Finally, the target contour constraint is used to optimize the segmentation result, achieving high-precision segmentation of each target instance in the image.

[0033] Example 1 The power station includes a power plant and a substation; the lossless compression method refers to arithmetic encoding; the inspection process refers to the process by which the power station inspects and checks its electromechanical equipment; the network that can be used in this process has low coverage, poor signal quality, and unstable connection status; the network refers to the wireless network between the data platform and the edge devices; the edge devices refer to mobile devices such as mobile phones, tablets, or PDAs; the images generated during the inspection process refer to the original images collected and generated by the edge devices, which are uncompressed original image data, and the images are stored in RGB(A) multi-channel format; the data platform refers to the hardware and software platform that undertakes data aggregation, data verification and decompression, and data storage functions.

[0034] Image data generated during power plant inspections often has a narrow color gamut and limited color richness. Since this type of image data is primarily used to record the appearance of equipment and facilities, its requirements for image resolution are not high. However, to accurately reproduce the on-site conditions and lay the groundwork for future machine vision-based intelligent inspections, this image data should be original, uncompressed images. Furthermore, to achieve efficient image data transmission between edge devices and the data center, lossless compression of the original image data is required on the edge device side before transmission, and decompression of the received data on the data center side after transmission.

[0035] First, the original image is decomposed by scanning to form multiple data sectors. Then, each data sector is arithmetic encoded. The arithmetic encoding process is a compression and initial encryption process. Another encryption process is to perform segmented RSA encryption on the compressed sector data.

[0036] Example 2 A method for processing image data generated during an inspection process includes the following steps: Step S1. The edge device acquires image data and decomposes the image data for all channels of the image using window matrices of multiple sizes to obtain the optimal window matrix size; Image data decomposition refers to processing data through an m x n window matrix. The process of performing layered scanning on each channel of the original image, where m and n are the number of rows and columns of the matrix, results in the original image being decomposed into multiple non-overlapping sub-matrices. For images where the number of rows / columns of the original image is not an integer multiple of the number of rows or columns of the window matrix, each channel is padded with 0 values ​​up to the last row or column. Figure 6 As shown, assuming the original image is an RGB three-channel image, each channel matrix has A rows and B columns. The matrix after zero-padding in layer R is shown. The number of rows that need zero-padding in the row direction is... The number of columns that need to be padded with zeros in the column direction is n. .

[0037] Using window matrix When scanning the channels of the original image layer by layer—for example, RGB has three color channels (red + green + blue) and RGBA has four channels (red + green + blue + opacity)—the sliding offset in the row direction is m, and the sliding offset in the column direction is n. This also means that all elements in the image use matrices. It will only be scanned once during a scan; such as Figure 7 As shown, using one When using a window matrix to decompose an image of an R layer, it will not decompose into a matrix. .

[0038] During the scan, assuming the original image's single-channel matrix has A rows and B columns, then the matrix... The number of scans in the column direction of a single channel of the image is The number of scans in the row direction is i. The number of matrices decomposed from this channel is i. .

[0039] Using matrices After all channels of the original image have been scanned layer by layer, the set of matrices decomposed by each scan is denoted as . , Generate a set by taking the unique value of all elements. and to Count each element to generate And record i, j and ;in After scanning all channels of an image, a series of dimensions are obtained, along with E. m×n A set of identical matrices; To The set of all elements in the set that have a unique value; for The count of each element in the array.

[0040] For any 2,B / 2+1] 2, A / 2+1], using a matrix During layered scanning, record the matrix of each window. The corresponding i, j and A set is formed after multiple scans. ,Pick In any pair of i, j and j that makes the following expression true and maximizes its value The corresponding m and n are taken as the optimal window matrix sizes: .

[0041] Step S2. Use the optimal window matrix size to scan and decompose each channel after zero padding, take the unique value of the scanned matrix set, calculate the probability and cumulative probability of each unique value, and perform arithmetic encoding according to the order of appearance of each element in the scanned matrix set.

[0042] By scanning and decomposing each channel after zero-padding using the optimal window matrix size, we can obtain the optimal result under these conditions. , and .pass Computable The probability values ​​of each element :

[0043] Define a set And calculate its cumulative probability and denot it as ,Right now:

[0044] It is the set of scanned matrices, for The first element in, find its position in The probability value corresponding to Find two consecutive ones in Q. and Make ;Will According to the proportion of each element in Q The proportion is used to redistribute Q and denoted as ,right The second element in, find its in The probability value corresponding to Make ;right Repeat the above operation for each element in the array, and after the operation is completed for the last element, the range can be obtained. Find the real number with the fewest decimal places within this interval, extract its decimal part, convert it to a string (e.g., 0.123456789 becomes "123456789"), and record it as... This is the complete compressed data.

[0045] Step S3. Encrypt the compressed data from step S2 in segments.

[0046] Obtain the compressed complete data Afterwards, record The number of decimal places in the string, which is also the length of the string, is and using RSA to Segmented encryption, for example, when using RSA for segmented encryption, means... It needs to be done according to each One character per data partition It is split into several plaintext data partitions. The string length is not When the value is a multiple of an integer, use the character "0". Add the last part; after encryption using the RSA public key, a ciphertext data partition is formed, denoted as . This represents the string length count, where n is the number of data partitions, i.e., the ciphertext sector data that needs to be transmitted from the client to the server in multiple transmissions; for each plaintext data partition, its hash function value is calculated to generate several sets of hash values, denoted as . and calculate The hash value of the string formed by concatenating all elements in the string is denoted as . ;Record Let m be the number of channels during scanning, a, the number of rows i and columns j of all channels scanned.

[0047] Except for the private key in the key pair required for encryption, which is issued offline, the data transmitted from the edge device's data transmission service instance to the data platform includes a set of data from the edge side. , , , and Metadata composed of, and n groups of and , Sector data as a whole: Step S4. The segmented encrypted data from step S3 is transmitted over the network; Step S5. When the server receives sector data, it uses the offline-issued private key to perform RSA decryption on the received ciphertext sector data. After decryption, it calculates the hash function value of the decrypted plaintext and compares it with the plaintext hash function value of the sector sent by the client. If the verification fails, the client is asked to retransmit the sector. If it passes, the plaintext data of the sector is temporarily stored in the position corresponding to its sector sequence number. The number of sectors that have been temporarily stored is determined by the m value in the received metadata. If so, the hash function values ​​of all the temporarily stored sectors are concatenated in order and the hash value is calculated and compared with the hash function value in the metadata. If the verification fails, the client is asked to retransmit. If it passes, decompression is performed. Specifically, after the data sending service instance in the edge device detects data to be transmitted in the edge device's database, it calls the data receiving service instance to transmit compressed and encrypted sector data from the edge side to the data platform. Upon receiving the data, the data receiving service instance on the data platform first processes the received sector data... use The process involves verification. If verification passes, the sector data is stored in the database, and the edge-side data sending service instance is invoked to delete the original sector data. If sector data verification fails, the process exits. If sector data verification passes, the process checks if the number of sector data in the local database is m. If so, the verification values ​​of all sector data are obtained, and these values ​​are combined into a string and then... Perform a comparison check. If the check passes, use the private key to decrypt the data in each sector. If the check fails, exit. Step S6. Decompress After the data in each sector is decrypted and the hash function value passes verification, the data from each sector is sequentially concatenated. After concatenation, all characters from the 0th to the kth position (inclusive) of the concatenation interface string are taken and converted into a number with an integer part of 0 and a fractional part consisting of the k characters, denoted as S. Then, the metadata is used... Restoring Cumulative Probability And find two consecutive ones within them. Make ,but The first in The term is the first matrix that has been restored. This process is repeated until the number of restored matrices reaches a certain threshold. Stop at this time.

[0048] The restored matrix is ​​spliced ​​together according to the number of channels 'a' during scanning, the number of rows 'i' and the number of columns 'j' in each channel, to complete the lossless decompression process of the compressed data; Step S7. Post-processing Since the original image may have been padded with zeros at the end of rows or columns of the channel matrix in step S1, it may be necessary to crop the image accordingly.

[0049] This application uses arithmetic coding to achieve lossless image compression. The difference between arithmetic coding and other entropy coding methods is that other entropy coding methods usually divide the input message into single characters or multiple strings and then encode each symbol. Arithmetic coding encodes the message as a whole into a decimal between 0 ≤ n < 1, which represents all the compressed information.

[0050] As an optimization in this embodiment, the decomposition in step S1 uses a window matrix, which is repeated multiple times with different window matrix sizes each time, until an optimal window matrix is ​​determined. Step S2 involves using the parameters of this optimal window matrix to decompose the image and obtain the decomposed matrix. However, since step S2 has already been executed once in step S1, in practical use, it is only necessary to retrieve the result of the decomposition using this window matrix in step S1; it is not necessary to perform the decomposition again using the parameters of the optimal window matrix.

[0051] Example 3 Step 1. Scan and decompose all channels of the original image using a window matrix to form multiple data sectors. The set of these sectors may contain duplicate values. Count how many of these duplicate values ​​there are.

[0052] Step 2. Perform statistics on all data sectors generated in the previous step, and formulate the parameters required for the compression process based on the statistical results. The parameters required for the compression process obtained in this step are treated as metadata and are transmitted only once between the transmitting and receiving ends.

[0053] Step 3. Based on the parameters required for the compression process obtained in the previous step, perform lossless compression on each sector separately.

[0054] Step 4. After compression, the data formed for a sector includes: ① compressed data, ② number of significant bits of data, ③ checksum of compressed data, ④ total number of sectors, and ⑤ checksum of original data.

[0055] Step 5. Transmit each sector separately (the transmission process is described in another pending patent).

[0056] Step 6. After the receiving end obtains the sector data, it uses "the compressed data of the sector" and "the checksum of the compressed data of the sector" to verify it. If the verification passes, it decompresses (also decrypts) and saves the data to the database; otherwise, it requires the sending end to retransmit the data.

[0057] Step 7. Decompress (decrypt).

[0058] The lossless compression process is also an encryption process.

Claims

1. A method for processing image data generated during an inspection process, characterized in that, Includes the following steps: Step S1. The edge device acquires image data and decomposes the image data for all channels of the image using window matrices of multiple sizes to obtain the optimal window matrix size. After this process, the original image will be decomposed into multiple non-overlapping sub-matrices. For any 2,B / 2+1] 2, A / 2+1], using a matrix During layered scanning, record the matrix of each window. The corresponding i, j and After multiple scans, a set is formed. ,Pick In any pair of i, j and j that makes the following expression true and maximizes its value The corresponding m and n are taken as the optimal window matrix sizes: ； The original image's single-channel matrix has A rows and B columns, m and n are the number of rows and columns of the window matrix, i is the number of scans in the row direction, and j is the number of scans in the column direction. This is the count of each element in the set after all elements in the matrix set obtained from each scan have a unique value; Step S2. Use the optimal window matrix size to scan and decompose each channel after zero padding, take the unique value of the scanned matrix set, calculate the probability and cumulative probability of each unique value, and perform arithmetic encoding according to the order of appearance of each element in the scanned matrix set. Step S3. Encrypt the compressed data from step S2 in segments; Step S4. The segmented encrypted data from step S3 is transmitted over the network; Step S5. When the server receives sector data, it uses the offline-issued private key to perform RSA decryption on the received ciphertext sector data; After decryption, the hash function value of the decrypted plaintext is calculated and compared with the hash function value of the plaintext of that sector sent by the client. If the verification fails, the client is asked to retransmit this sector. If it passes, the plaintext data of that sector is temporarily stored in the position corresponding to its sector sequence number. The number of sectors that have been temporarily stored is determined by the m value in the received metadata. If so, the hash function values ​​of all the temporarily stored sectors are concatenated in order and the hash value is calculated and compared with the hash function value in the metadata. If the verification fails, the client is asked to retransmit. If it passes, decompression is performed. Step S6. Decompress; Step S7. Post-processing.

2. The method for processing image data generated during the inspection process according to claim 1, characterized in that, In step S1, image data decomposition refers to processing the data through an m x n window matrix. The process of performing layered scanning on each channel of the original image involves filling each channel with 0 values ​​up to the last row or column for images whose number of rows / columns is not an integer multiple of the number of rows or columns of the window matrix.

3. The method for processing image data generated during the inspection process according to claim 2, characterized in that, The number of rows that need to be padded with zeros in the row direction is The number of columns that need to be padded with zeros in the column direction is n. .

4. A method for processing image data generated during an inspection process according to claim 2, characterized in that, Using window matrix When scanning each channel of the original image in layers, the sliding offset in the row direction is m, and the sliding offset in the column direction is n; during the scan, the matrix... The number of scans in the column direction of a single channel of the image is The number of scans in the row direction is i. The number of matrices decomposed from this channel is i. .

5. A method for processing image data generated during an inspection process according to claim 4, characterized in that, Using matrices After all channels of the original image have been scanned layer by layer, the set of matrices decomposed by each scan is denoted as . , Generate a set by taking the unique value of all elements. and to Count each element to generate And record i, j and ;in After scanning all channels of an image, a series of dimensions are obtained, along with E. m×n A set of identical matrices; To The set of all elements in the set that have taken a unique value; for The count of each element in the array.

6. The method for processing image data generated during the inspection process according to claim 1, characterized in that, In step S2, after scanning and decomposing each zero-padded channel using the optimal window matrix size, the optimal condition can be obtained. , and ,pass Computable The probability values ​​of each element in the equation : Define a set And calculate its cumulative probability and denot it as ,Right now: It is the set of scanned matrices, for The first element in, find its position in The probability value corresponding to Find two consecutive ones in Q. and Make ;Will According to the proportion of each element in Q The proportion is used to redistribute Q and denoted as ,right The second element in, find its position in The probability value corresponding to Make ;right Repeat the above operation for each element in the array, and after the operation is completed for the last element, the range can be obtained. Find the real number with the fewest decimal places within this interval, extract its decimal part, convert it to a string, and denot it as [string name missing]. This is the complete compressed data.

7. A method for processing image data generated during an inspection process according to claim 6, characterized in that, In step S3, the compressed complete data is obtained. Afterwards, record The number of decimal places in the string, which is also the length of the string, is and using RSA to Perform segmented encryption. It needs to be done according to each One character per data partition Split into several plaintext data partitions, The number of binary bits in the modulus of the key pair. The string length is not When the integer is a multiple of an integer, use the character "0". Add the last part; after encryption using the RSA public key, a ciphertext data partition is formed, denoted as . , The set formed after segmented encryption , , This represents the string length count, where n is the number of data partitions. For each plaintext data partition, its hash function value is calculated to generate a set of hash values, denoted as . and calculate The hash value of the string formed by concatenating all elements in the string is denoted as . ;Record Let m be the number of channels during scanning, a, the number of rows i and columns j of all channels scanned.

8. A method for processing image data generated during an inspection process according to claim 7, characterized in that, Except for the private key in the key pair required for encryption, which is issued offline, the data transmitted from the edge device's data transmission service instance to the data platform includes a set of data from the edge side. , , , and Metadata composed of, and n groups of and , The sector data that makes up the data.

9. A method for processing image data generated during an inspection process according to claim 8, characterized in that, In step S6, after the data in each sector is decrypted and the hash function value is verified, the data in each sector is sequentially concatenated. After concatenation, all characters from the 0th to the kth position of the concatenation interface string are taken and converted into a number whose integer part is 0 and whose fractional part is the fraction of the k characters, denoted as S. The metadata is then used. Restoring Cumulative Probability And find two consecutive ones within them. Make ,but The first in The term is the first matrix that has been restored. This process is repeated until the number of restored matrices reaches a certain threshold. Stop at this time; The restored matrix is ​​then stitched together according to the number of channels 'a' during scanning, the number of rows 'i' and columns 'j' of each channel scan to complete the lossless decompression process of the compressed data.

10. A method for processing image data generated during an inspection process according to claim 1, characterized in that, Step S7 specifically involves cropping the image.

Images