Hyperspectral image system geometric correction method based on MPI parallel computing

Abstract

The invention belongs to the technical field of hyperspectral image system geometric correction, and discloses a hyperspectral image system geometric correction method based on MP I parallel computing. The method comprises the following specific steps: S1, analyzing input parameters; s2, explaining an imaging model; s3, selecting an image blocking strategy; s4, block parameter calculation is carried out; s5, parallel block calculation is carried out; s6, executing geometric correction; and S7, outputting a result. The MPI parallel computing geometric correction algorithm has the remarkable advantages that by means of an efficient task distribution mechanism, the processing speed is greatly increased, complex tasks are decomposed to multiple nodes for synchronous processing, the bottleneck of a traditional algorithm is overcome, and the modern image data processing requirement is met; the algorithm is good in expansibility, can flexibly adapt to tasks of different scales and various image data, and also has universality and practicability. And various advantages are integrated, the practical application value is achieved in multiple industries, and the wide market prospect is manifested along with technical popularization and application.

Description

Technical Field

[0001] The present invention belongs to the technical field of geometric correction of hyperspectral imaging systems, and specifically relates to a geometric correction method for hyperspectral imaging systems based on MPI parallel computing. Background Technique

[0002] Hyperspectral satellite remote sensing has significant advantages. Its image spectral resolution is high, and it can obtain fine spectral features of ground objects and has wide applications. However, there are geometric deformation problems when acquiring images. The reasons include factors such as satellite platforms, sensors, and terrain, which affect the accuracy and usability of the images. The collinearity condition equation is the core. Establishing a strict imaging model requires multiple parameters. When there are no sensor parameters, a general model can be used. The RPC model is based on the collinearity condition, has universality and confidentiality, and is the main form of remote sensing satellite products. Using the model parameters for geometric transformation can correct the geometric distortion caused by shooting, convert the pixel position from the image coordinate system to the geographic coordinate system. After the transformation, interpolation processing is often required to estimate the gray value of the pixels in the corrected image to ensure the smoothness of the output image, thereby improving the quality and application value of hyperspectral satellite remote sensing images.

[0003] In today's era, the image resolution continues to climb and the data volume has increased explosively, which poses a severe test to traditional algorithms. Traditional algorithms have not effectively tapped the potential of modern computer multi-core processors and distributed computing environments, lacking the key element of parallel computing ability. Especially for traditional geometric correction algorithms, when dealing with large-scale image data, the speed and efficiency are becoming increasingly lagging, and it is difficult to adapt to the growing data processing requirements. There is an urgent need for new algorithms or technical improvements to make full use of hardware resources, improve processing efficiency, and meet the urgent needs of practical applications. Summary of the Invention

[0004] The purpose of the present invention is to provide a geometric correction method for hyperspectral imaging systems based on MPI parallel computing to solve the problems raised in the above background technique.

[0005] To achieve the above purpose, the present invention provides the following technical solutions: The specific steps of a geometric correction method for hyperspectral imaging systems based on MPI parallel computing are as follows:

[0006] S1: Parse input parameters: The 0th process is responsible for parsing the input parameters of the system geometric correction, including parsing the input RPC or RPB file. According to the parsed input image and RPC parameters, calculate the size of the corrected image and the six-parameter information, create the result image, and broadcast the input parameters to other processes;

[0007] S2: Explanation of Imaging Model: The strict imaging model of satellite sensors is used to express the conversion relationship between image pixel coordinates and geodetic longitude and latitude coordinates. The commonly used inverse calculation model represents the mapping from pixel coordinates and given elevation to geodetic longitude and latitude coordinates. Through the inverse calculation model of RPC and the input image information, the six parameters of the geometrically corrected image can be calculated;

[0008] S3: Selection of Image Blocking Strategy: After extensive data testing, the processing speed of block-by-row processing is better than other blocking methods. Therefore, the present invention adopts the block-by-row strategy;

[0009] S4: Calculation of Block Parameters: Each process calculates the starting position of the image block it processes, the size of the block processed at one time, and the total size of the blocks processed by the process according to the image size and the total number of MPI processes;

[0010] S5: Parallel Block Calculation: All processes further decompose and calculate the blocks based on the data blocking information in Step 3 to reduce the memory usage, and read the corresponding data blocks according to the blocking information;

[0011] S6: Perform Geometric Correction: Perform systematic geometric correction on the hyperspectral image according to the functional chain of "constructing RPC coordinate conversion relationship - finding the position relationship of the output points in the input image - image resampling";

[0012] S7: Result Output: Each process writes the calculation result into the output file according to the starting position and data block size calculated by itself.

[0013] Preferably, the parsing of input parameters in S1 means that Process 0 undertakes the key starting task in the geometric correction process. First, parse the input parameters, covering RPC or RPB files, thereby constructing the mapping relationship between image row and column coordinates and geodetic longitude and latitude coordinates. Based on this mapping and the original image information, accurately calculate the size of the corrected image and the six key parameter data, and successfully create the corrected result image. After completing these preliminary preparations, Process 0 broadcasts the geometric correction input parameters including the input image, RPC, and output image to other processes, laying a foundation for the smooth progress of the subsequent overall geometric correction work.

[0014] Preferably, the specific steps of the imaging model explanation in S2 are as follows:

[0015] Step 1: Foundation of Imaging Model: Clarify the role of the strict imaging model of satellite sensors, that is, it is used to establish the conversion relationship between image pixel coordinates and geodetic longitude and latitude coordinates. Among them, the inverse calculation model can realize the mapping from pixel coordinates and given elevation to geodetic longitude and latitude coordinates, such as (φ,λ) = F(x,y);

[0016] Step 2: Data Information Integration: Collect and organize various types of information contained in the RPC inversion model and the input image to prepare for subsequent calculations. This information covers the relevant data in the original parameter RPC file of the image and is the key basis for calculating the six parameters of the geometrically corrected image;

[0017] Step 3: Six-Parameter Calculation: Accurately calculate the six parameters of the geometrically corrected image, including the number of rows and columns and the key index of spatial resolution, according to (φ,λ)=F(x,y), so as to provide basic data support for subsequent geometric correction work, ensure that the image can be accurately corrected and processed to meet the needs of actual applications.

[0018] Preferably, the image block division strategy selection in S3 refers to the comparative analysis of the performance of various block processing methods under different scenarios and different data scales.

[0019] Preferably, the block parameter calculation in S4 means that each process processes a certain number of consecutive data blocks in rows. To reduce the temporal usage of memory resources, each process divides the data blocks into consecutive rows for processing. To improve the resource utilization rate and make the processing time of each process approximately equal, the image is divided into blocks by rows according to the number of processes.

[0020] Preferably, the parallel block calculation in S5 is based on the data block information determined in Step 3, and further carries out fine-grained block decomposition calculations. In this way, the memory occupancy during the data processing is effectively reduced, and the resource utilization efficiency is optimized. According to the accurate block information, the process can accurately read the corresponding data blocks, ensuring the accuracy and integrity of data acquisition, laying a solid foundation for subsequent high-quality processing work, and ensuring the efficient and stable operation of the entire process.

[0021] Preferably, the specific steps for performing geometric correction in S6 are as follows:

[0022] Step 1: Initial Position Calculation: Calculate the initial position of each pixel of the output image in the input image according to the mapping relationship between the image row-column coordinates and the geodetic longitude and latitude coordinates. This step establishes a preliminary connection between the input and output images and determines the approximate corresponding positions of the pixels;

[0023] Step 2: Non-Integer Coordinate Judgment: Judge whether the position coordinates of the pixels of the output image in the input image are integers. Usually, these coordinates are not integers, as shown in the appendix Figure 3 shown. This characteristic determines the necessity of subsequent processing and points the way for further operations;

[0024] Step 3: Interpolation and smoothing processing: Due to the existence of non-integer coordinates, to ensure the smoothness of the output image, a spatial interpolation processing method is adopted. Through this process, the pixel values of each pixel in the corrected image are accurately calculated, so that the output image can achieve an ideal effect in terms of visual and data quality and meet the requirements of practical applications.

[0025] Preferably, the result output in S7 refers to that each process accurately writes to the corresponding output file according to the calculated starting position and data block size of itself, ensuring the integrity of the entire processing result.

[0026] The beneficial effects of the present invention are as follows:

[0027] 1. Through the MPI parallel computing geometric correction algorithm, with its efficient MPI task allocation mechanism, the present invention gives full play to the parallel computing ability, greatly improving the geometric correction processing speed. For large-scale image data, the processing time can be significantly reduced. It decomposes complex tasks to multiple computing nodes for synchronous processing, with much higher efficiency compared to the traditional serial method. This parallel processing mode effectively overcomes the bottleneck of traditional algorithms in processing high-resolution and large-volume images, enabling the geometric correction work to be completed quickly and efficiently, bringing a qualitative leap to the data processing efficiency and meeting the urgent needs of modern image data processing.

[0028] 2. The algorithm of the present invention has good scalability and can be flexibly adjusted according to the actual scale of computing resources. When the number of computing nodes increases, the processing ability increases linearly. This means that whether it is a small data processing task or a large-scale image processing project, it can be well adapted to easily handle workloads of different magnitudes. At the same time, it has a wide range of adaptability. Whether it is satellite images from space, images obtained by aerial photography, or professional images in the medical field and other various types of image data, as well as diverse geometric correction requirements, they can all be effectively processed by this algorithm, demonstrating its versatility and practicality in different fields.

[0029] 3. The MPI parallel computing geometric correction algorithm of the present invention contributes an innovative solution to the field of image and data processing. It combines advantages such as high efficiency, accuracy, and scalability. It can not only ensure the accuracy of correction during the processing, but also flexibly adjust resource utilization according to the actual situation to meet the changing business needs. In many industries relying on image data processing, such as remote sensing mapping, medical diagnosis, geographic information systems, etc., it has practical application value that cannot be ignored, and its broad market prospect will be further demonstrated with the continuous popularization and application of the technology. Description of the Drawings

[0030] Figure 1 It is a flowchart of the geometric correction method for the hyperspectral image system based on MPI parallel computing of the present invention;

[0031] Figure 2 Schematic diagram of the parallel (block) system geometric correction process based on the MPI message passing model adopted by the present invention;

[0032] Figure 3 Schematic diagram of the geometric transformation mapping relationship before and after the geometric correction of the remote sensing image system;

[0033] Figure 4 Schematic diagram of the process for determining the range of the system geometric correction image;

[0034] Figure 5 Schematic diagram of the parallel geometric correction block strategy for hyperspectral images of the present invention;

[0035] Figure 6 Parallel execution time diagram of the system geometric correction of the embodiment of the present invention;

[0036] Figure 7 Acceleration ratio diagram of the parallel execution of the system geometric correction of the embodiment of the present invention;

[0037] Figure 8 Diagram showing the relationship between the execution time, acceleration ratio and efficiency of the system geometric correction of the hyperspectral data of XIGUANG-02 satellite of the present invention and the number of processes. Detailed implementation manners

[0038] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0039] As Figures 1 to 8 shown, the embodiment of the present invention provides a method for geometric correction of a hyperspectral image system based on MPI parallel computing, and the specific steps are as follows:

[0040] S1: Analyze input parameters: The 0th process is responsible for analyzing the input parameters of the system geometric correction, including analyzing the input RPC or RPB file, calculating the size of the corrected image and the six-parameter information based on the analyzed input image and RPC parameters, creating the result image, and broadcasting the input parameters to other processes;

[0041] S2: Imaging model description: The strict imaging model of the satellite sensor is used to express the conversion relationship between the image pixel coordinates and the geodetic longitude and latitude coordinates. The commonly used inverse calculation model represents the mapping from the pixel coordinates and the given elevation to the geodetic longitude and latitude coordinates. The six parameters of the geometrically corrected image can be calculated through the inverse calculation model of RPC and the input image information;

[0042] S3: Selection of image block division strategy: Through a large number of data tests, the processing speed of row-based block division is better than other block division methods. Therefore, the present invention adopts the row-based block division strategy;

[0043] S4: Calculation of block parameters: Each process calculates the starting position of the image block it processes, the size of the block processed at one time, and the total size of the blocks processed by the process based on the image size and the total number of MPI processes;

[0044] S5: Parallel block calculation: All processes further decompose and calculate the block division based on the data block division information in step 3 to reduce the memory usage, and read the corresponding data blocks according to the block division information;

[0045] S6: Perform geometric correction: Perform systematic geometric correction on the hyperspectral image according to the functional chain of "construction of RPC coordinate conversion relationship - finding the position relationship of the output points in the input image - image resampling";

[0046] S7: Result output: Each process writes the calculation result into the output file according to the starting position and data block size calculated by itself.

[0047] Construct the mapping relationship between the image row-column coordinates (x, y) and the geodetic longitude and latitude coordinates according to the parsed RPC parameters between:

[0048]

[0049] In the formula, (X, Y) represents the normalized row and column values, and (P, L, H) represents the normalized latitude, longitude, and elevation; the 20 coefficients of the cubic polynomial are expressed as:

[0050]

[0051] Among them, C 1 … C 20 represents LINE_NUM_COEF i , LINE_NUM_COEF i , LINE_NUM_COEF i and LINE_NUM_COEF i this set of coefficients; in this rational polynomial function model, the distortion caused by optical projection is represented by a first-order polynomial, the distortion caused by the earth's curvature, atmospheric refraction, and lens distortion is represented by a second-order polynomial, and other unknown distortions are represented by a third-order polynomial. The conversion relationships between the row-column values and the normalized row-column values, and between the geographic coordinates and the normalized geographic coordinates are as follows:

[0052]

[0053] L = (λ - LONG_OFF) / LONG_SCALE

[0054] H = (h - HEIGHT_OFF) / HEIGHT_SCALE

[0055] X = (x - LINE_OFF) / LINE_SCALE

[0056] Y = (y - SAMP_OFF) / SAMP_SCALE

[0057] Among them, the parsing of input parameters in S1 refers to that Process 0 undertakes the key starting tasks in the geometric correction process. First, the input parameters are parsed, covering RPC or RPB files, thereby constructing the mapping relationship between the image row-column coordinates and the geodetic longitude and latitude coordinates. Based on this mapping and the original image information, the size of the corrected image and the six key parameters are accurately calculated, and the corrected result image is successfully created. After completing these preliminary preparations, Process 0 broadcasts the geometric correction input parameters including the input image, RPC, and output image to other processes, laying a foundation for the smooth progress of the subsequent overall geometric correction work.

[0058] Process 0 plays a key role at the initial stage of geometric correction. By parsing input parameters such as RPC or RPB files, it constructs the image coordinate mapping relationship, calculates the key data of the corrected image based on this and the original image information, and creates the result image. After completing these preparatory works, Process 0 broadcasts the parameters including the input image, RPC, and output image to other processes, thus laying a solid foundation for the smooth development of the subsequent geometric correction process and ensuring the orderly progress of the entire correction work.

[0059] Among them, the specific steps of the imaging model description in S2 are as follows:

[0060] Step 1: Foundation of the imaging model: Clarify the role of the strict imaging model of the satellite sensor, which is used to establish the conversion relationship between the image pixel coordinates and the geodetic longitude and latitude coordinates. The inverse calculation model can realize the mapping from pixel coordinates and a given elevation to geodetic longitude and latitude coordinates, such as (φ, λ) = F(x, y);

[0061] Step 2: Integration of data information: Collect and organize various information included in the RPC inverse calculation model and the input image to prepare for subsequent calculations. This information covers the original parameters of the image and the relevant data in the RPC file, which are the key basis for calculating the six parameters of the corrected image;

[0062] Step 3: Calculation of six parameters: Accurately calculate the six parameters of the geometrically corrected image according to (φ, λ) = F(x, y), including the number of rows and columns and the key index of spatial resolution, thereby providing basic data support for the subsequent geometric correction work, ensuring that the image can be accurately corrected and processed to meet the needs of practical applications.

[0063] First is the basis of the imaging model. It is necessary to clearly understand its role in establishing the image coordinate conversion relationship and the inverse calculation model. Next is the integration of data information, collecting various information of the RPC inverse calculation model and the input image, which is the key basis for calculation. Finally is the calculation of six parameters. Relying on the previous steps, accurately calculate six parameters such as the row and column numbers and spatial resolution of the corrected image, providing basic data for subsequent correction, meeting the requirements of accurate correction and processing of the image in practical applications, and ensuring the image quality and usability.

[0064] Among them, the selection of the image block division strategy in S3 refers to the comparative analysis of the performance of various block processing methods under different scenarios and different data scales.

[0065] Suppose the image has a total of N rows and is divided among n processes for processing; then, process 0 processes the data block from row 0 to N / n, process 1 processes the data block from row N / n to 2*N / n, and process n - 1 processes the data block from row (n - 1)*N / n to N; each process can be further divided according to the usage of resources inside.

[0066] Among them, the calculation of block parameters in S4 means that each process processes a certain number of consecutive rows of data blocks. To reduce the temporary usage of memory resources, each process is divided into consecutive-row data blocks for processing inside. To improve the resource utilization rate and make the processing time of each process approximately equal, the image is divided by rows according to the number of processes.

[0067] Each process is assigned to process a certain number of consecutive rows of data blocks. To minimize the occupation of memory resources to the greatest extent, it is further divided into consecutive-row data blocks inside each process for fine processing. At the same time, to fully improve the resource utilization rate and ensure that the processing time of each process can be approximately equal, the method of dividing the image by rows according to the number of processes is adopted. This method makes the entire processing flow more efficient and stable, and the resource allocation is more reasonable.

[0068] Among them, the parallel block calculation in S5 is based on the data block information determined in step three, and further conducts fine block decomposition calculation. In this way, the memory occupation during the data processing is effectively reduced, and the resource utilization efficiency is optimized. According to the accurate block information, the process can accurately read each corresponding data block, ensuring the accuracy and integrity of data acquisition, laying a solid foundation for subsequent high-quality processing work, and ensuring the efficient and stable operation of the entire process.

[0069] Deeply carry out fine block decomposition, which reduces memory occupancy and improves resource efficiency. With accurate block information, the process can accurately read the corresponding data blocks, ensuring the accuracy and integrity of the data, laying a solid foundation for subsequent processing, and strongly promoting the efficient and stable continuous progress of the entire process, thus achieving high-performance data processing effects.

[0070] Among them, the specific steps of performing geometric correction in S6 are as follows:

[0071] Step 1: Initial position calculation: According to the mapping relationship between the image row-column coordinates and the geodetic longitude and latitude coordinates, calculate the initial position of each pixel in the output image in the input image. This step establishes a preliminary connection between the input and output images and determines the approximate corresponding positions of the pixels.

[0072] Step 2: Non-integer coordinate judgment: Judge whether the position coordinates of the pixels in the output image in the input image are integers. Usually, these coordinates are not integers, as shown in the appendix. This characteristic determines the necessity of subsequent processing and indicates the direction for further operations. Figure 3 As shown, this characteristic determines the necessity of subsequent processing and indicates the direction for further operations.

[0073] Step 3: Interpolation smoothing processing: Due to the existence of non-integer coordinates, to ensure the smoothness of the output image, a spatial interpolation processing method is adopted. Through this process, the pixel values of each pixel in the corrected image are accurately calculated, enabling the output image to achieve an ideal effect in terms of visual and data quality and meeting the requirements of practical applications.

[0074] The time complexity of serial calculation, the calculation formula is:

[0075] T s = O(V)

[0076] The time complexity of parallel calculation, the calculation formula is:

[0077] T p = O(V / P)

[0078] The speedup ratio, the calculation formula is:

[0079] S p = T s / T p

[0080] The efficiency, the calculation formula is:

[0081] E p = S p / P

[0082] Among them, V is the scale of the image, P is the number of processes. The speedup is proportional to the number of processes. The efficiency is used to measure the utilization of multiple processors during calculation and is usually less than or equal to 1. The closer it is to 1, the higher the utilization efficiency of P processors. To verify the speed of the geometric correction of the hyperspectral image system provided by the present invention, taking the geometric correction of the hyperspectral data (150 bands) of the Xiguang-02 satellite of Zhongke Xiguang Aerospace Technology Co., Ltd. as an example, the geometric correction is performed using the method for geometric correction of the hyperspectral image system based on MPI parallel computing provided by the present invention and compared with serial geometric correction. The execution time, speedup, and efficiency of the parallel system geometric correction of the Xiguang-02 satellite image (2000×2200) are shown in Figure 8 . It can be seen from the table that the speedup is approximately proportional to the number of processes. However, as the number of processes increases, the increment of the speedup continuously decreases. The reason is that as the number of processes increases, the proportion of the algorithm calculation time gradually becomes smaller, while the proportion of the process communication time is continuously increasing, resulting in the decreasing growth rate of the speedup. In addition, process 0 needs to parse the input parameters and create the output file, and at this time, other processes are in a waiting state. As the number of processes increases, the proportion of the algorithm calculation time gradually becomes smaller, and the time proportion used by process 0 alone increases, which also leads to the decreasing growth rate of the speedup. (Appendix Figure 6 is the graph of the change in the parallel execution time of the system geometric correction with the number of processes in the example, and Appendix Figure 7 is the parallel speedup graph of the system geometric correction in the example.) By comparing and analyzing the speeds of parallel and serial system geometric corrections, it is proved that the speed of the parallel geometric correction of the hyperspectral remote sensing image system provided by the present invention is significantly better than that of serial geometric correction.

[0083] Among them, the result output in S7 refers to that each process accurately writes to the corresponding output file according to its own calculated starting position and data block size, ensuring the integrity and accuracy of the entire processing result and providing reliable data support for subsequent analysis and application.

[0084] According to the starting position carefully calculated in advance and the accurately measured data block size, the processing results of their respective responsible parts are methodically and accurately written line by line and pixel by pixel into a specific output file. This process not only ensures that the processing results are complete without omission, but also ensures the high accuracy of the data, providing a solid and reliable data foundation for subsequent in-depth analysis, practical applications, etc., and successfully concluding the entire processing process.

[0085] It should be noted that in this text, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprising", "including" or any other variant thereof are intended to cover non-exclusive inclusion, such that a process, method, article or device comprising a series of elements not only includes those elements but also includes other elements not expressly listed, or elements inherent to such process, method, article or device.

[0086] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A hyperspectral imaging system geometric correction method based on MPI parallel computing, characterized by: The specific steps of the hyperspectral imaging system geometric correction method based on MPI parallel computing are as follows: S1: Parsing input parameters: Process 0 is responsible for parsing the input parameters of the system geometry correction, including parsing the input RPC or RPB file, calculating the size of the corrected image and six parameter information based on the parsed input image and RPC parameters, creating the result image, and broadcasting the input parameters to other processes; S2: Imaging model description: The strict imaging model of satellite sensors is used to express the conversion relationship between image pixel coordinates and geodetic longitude and latitude coordinates. The commonly used inverse calculation model represents the mapping from pixel coordinates and given elevation to geodetic longitude and latitude coordinates. The six parameters of the geometrically corrected image can be calculated through the RPC inverse calculation model and input image information. S3: Image segmentation strategy selection: After a large amount of data testing, the processing speed of segmentation by row is better than other segmentation methods, so the present invention adopts the strategy of segmentation by row; S4: Block parameter calculation: Each process calculates the starting position of its own processed image block, the size of a single processing block, and the total size of the process processing block based on the image size and the total number of MPI processes; S5: Parallel block calculation: All processes further divide the calculation into blocks based on the data block information in step 3 to reduce memory usage, and read the corresponding data blocks according to the block information; S6: Execute geometric correction: According to the function chain of "constructing the RPC coordinate transformation relationship - finding the position relationship of the output point in the input image - image resampling", perform systematic geometric correction on the hyperspectral image; S7: Result output: Each process writes the calculation result into the output file according to its own calculation starting position and data block size.

2. The hyperspectral imaging system geometric correction method based on MPI parallel computing according to claim 1, characterized in that: Parsing the input parameters in S1 refers to the key starting task that process 0 undertakes in the geometric correction process. First, the input parameters are parsed, including the RPC or RPB file, to build a mapping relationship between the image row and column coordinates and the geodetic longitude and latitude coordinates. Based on this mapping and the original image information, the size of the corrected image and the six-parameter key data are accurately calculated, and the correction result image is successfully created. After completing these preliminary preparations, process 0 will broadcast the geometric correction input parameters including the input image, RPC and output image to other processes, laying the foundation for the smooth progress of the subsequent overall geometric correction work.

3. The hyperspectral imaging system geometric correction method based on MPI parallel computing according to claim 1, characterized in that: The specific steps of the imaging model described in S2 are as follows: Step 1: Imaging model foundation: clarify the role of the strict imaging model of the satellite sensor, that is, to establish the conversion relationship between the image pixel coordinates and the geodetic longitude and latitude coordinates. The inverse model can realize the mapping from the pixel coordinates and the given elevation to the geodetic longitude and latitude coordinates, such as (φ,λ) = F(x,y); Step 2: Data information integration: Collect and organize the RPC inverse model and various information contained in the input image to prepare for subsequent calculations. This information covers the relevant data in the RPC file of the original parameters of the image, which is the key basis for calculating the six parameters of the image after geometric correction; Step 3: Calculation of six parameters: Based on (φ, λ) = F(x, y), the six parameters of the geometrically corrected image are accurately calculated, including the number of rows and columns and the key indicators of spatial resolution, so as to provide basic data support for subsequent geometric correction work and ensure that the image can be accurately corrected and processed to meet the needs of practical applications.

4. The hyperspectral imaging system geometric correction method based on MPI parallel computing according to claim 1, characterized in that: The image segmentation strategy selection in S3 refers to the comparative analysis of the performance of various segmentation processing methods in different scenarios and different data scales.

5. The hyperspectral imaging system geometric correction method based on MPI parallel computing according to claim 1, characterized in that: The block parameter calculation in S4 refers to each process processing a certain number of continuous rows of data blocks. In order to reduce the amount of memory resources used, each process is divided into continuous rows of data blocks for processing. In order to improve resource utilization and make the processing time of each process roughly equal, the image is divided into rows according to the number of processes.

6. The hyperspectral imaging system geometric correction method based on MPI parallel computing according to claim 1, characterized in that: The parallel block calculation in S5 refers to further carrying out detailed block decomposition calculation based on the data block information determined in step three. In this way, the memory usage in the data processing process is effectively reduced, and the resource utilization efficiency is optimized. Based on the precise block information, the process can accurately read the corresponding data blocks, ensuring the accuracy and completeness of data acquisition, laying a solid foundation for subsequent high-quality processing work, and ensuring the efficient and stable operation of the entire process.

7. The hyperspectral imaging system geometric correction method based on MPI parallel computing according to claim 1, characterized in that: The specific steps of performing geometric correction in S6 are as follows: Step 1: Initial position calculation: Based on the mapping relationship between the image row and column coordinates and the geodetic longitude and latitude coordinates, the initial position of each pixel of the output image in the input image is calculated. This step establishes the initial connection between the input and output images and determines the approximate corresponding positions of the pixels. Step 2: Non-integer coordinate determination: Determine whether the position coordinates of the pixels of the output image in the input image are integers. Usually, these coordinates are not integers, as shown in FIG3 . This feature determines the necessity of subsequent processing and indicates the direction for further operations. Step 3: Interpolation and smoothing: Due to the existence of non-integer coordinates, in order to ensure the smoothness of the output image, a spatial interpolation method is used. Through this process, the pixel value of each pixel in the corrected image is accurately calculated, so that the output image can achieve ideal effects in terms of vision and data quality, meeting the needs of practical applications.

8. The hyperspectral imaging system geometric correction method based on MPI parallel computing according to claim 1, characterized in that: The result output in S7 refers to that each process accurately writes the starting position and data block size calculated by itself into the corresponding output file, ensuring that the integrity and accuracy of the entire processing result are effectively preserved, providing reliable data support for subsequent analysis and application.

Images