An on-orbit image quality monitoring method and system for an automated optical remote sensing satellite

Abstract

This invention discloses an automated method and system for monitoring the quality of on-orbit images from optical remote sensing satellites, belonging to the field of on-orbit quality evaluation of remote sensing satellite image data. The image quality monitoring method proposed in this invention comprehensively employs multiple evaluation indicators, including on-orbit MTF and on-orbit SNR, to more comprehensively reflect satellite image quality issues. In the selection of edge feature regions, this invention utilizes the LSD algorithm and threshold-based feature set screening, and effectively filters edge features based on the edge signal-to-noise ratio. Theoretical analysis and verification with actual on-orbit image data demonstrate that using the edge ROI regions selected by the proposed method for on-orbit MTF evaluation yields better convergence and accuracy that meets the requirements for satellite image quality monitoring. The method proposed in this invention can be implemented based on remote sensing images of natural ground features, comprehensively utilizes multiple evaluation indicators, has a high degree of automation, and enables large-scale management of on-orbit satellite image quality issues.

Description

Technical Field

[0001] This invention relates to the field of on-orbit quality assessment of remote sensing satellite image data, and specifically to an automated method and system for on-orbit image quality monitoring of optical remote sensing satellites. Background Technology

[0002] After long-term operation, the image quality of onboard cameras on remote sensing satellites typically changes due to factors such as variations in the space environment, decreased platform stability, and component aging. This often manifests as blurring and distortion in the remote sensing images. Therefore, to accurately assess the on-orbit operational status of remote sensing satellites, it is necessary to periodically monitor the image quality of the onboard cameras.

[0003] During the on-orbit operation and management phase of a satellite, two methods are typically used for image quality monitoring: one is based on visual interpretation, and the other is based on objective parameter calculations. Visual interpretation involves satellite operators judging the degree of blurring and distortion in on-orbit images with their own eyes. This method is relatively intuitive and more in line with the biological interpretation characteristics of the human eye. However, in recent years, with advancements in satellite manufacturing and processing technologies, the blurring and distortion caused by short-term image quality degradation of onboard cameras are usually small in magnitude. Such changes are difficult to detect visually in the early stages. Once the human eye can clearly distinguish image quality degradation in remote sensing images, it may affect the normal execution of the satellite's observation mission. Objective parameter calculation methods are currently used to evaluate the sharpness of spaceborne camera remote sensing images. The most widely used metrics are the on-orbit modulation transfer function (MTF) and the on-orbit signal-to-noise ratio (SNR). The MTF characterizes the imaging system's response to signals of different spatial frequencies. Currently, the edge-edge method is often used, extracting a suitable edge region from the remote sensing image and then calculating its MTF along the orbit and perpendicular to the orbit. The on-orbit SNR reflects the relationship between the effective signal of the spaceborne camera and image noise under certain lighting and ground reflectivity conditions. A large, uniform area is selected from the remote sensing image acquired under specific conditions, and the system SNR is derived using information such as the image mean square error.

[0004] Currently, image quality assessment for spaceborne cameras is mostly based on remote sensing images of artificial targets. In recent years, my country has successively built dedicated calibration sites for remote sensing satellites, such as the Baotou Target Range, Songshan Target Range, and Yuxi Target Range. These sites have large areas of black and white targets, which can be used to calculate and evaluate the on-orbit MTF of spaceborne cameras, and can also be used to calculate on-orbit MTF and SNR. In recent years, researchers have also been exploring automatic MTF assessment methods based on remote sensing images of natural ground features. For example, Wang et al. from the Institute of Optics, Electronics and Information Technology, Chinese Academy of Sciences, used the Hough transform algorithm to extract edge features and verified it using IKONOS satellite data; Cenci et al. from Serco, Italy, proposed a statistically based semi-automatic MTF measurement method and verified it using Landsat 8 OLI-L1T data; Wang Yuhao et al. (CN114399688A) from the Resource Remote Sensing Satellite Application Center proposed a method and device for selecting edge regions, using a prior knowledge base to extract and restore the edge regions of ground features.

[0005] The above analysis reveals that existing methods for monitoring the quality of on-orbit images from optical remote sensing satellites share common problems, requiring human intervention in the monitoring loop: First, visual interpretation methods are inherently subjective, with different operators potentially yielding different conclusions, the validity of which depends on their experience. Second, methods based on manual targets for objective parameter calculation rely heavily on calibration sites, severely limiting the number of effective samples. Third, current methods for monitoring natural ground features in remote sensing images largely depend on operators examining the images to select suitable edge and uniform regions, especially for multispectral or hyperspectral images, requiring analysis and judgment band by band, making it difficult to maximize the extraction of effective data, and the repeatability of the results depends on the operator's experience. Fourth, methods by Cenci et al. and Wang Yuhao et al. both evaluate images by extracting single edge feature regions, failing to comprehensively reflect the impact of dark current noise, readout circuit noise, etc., making it difficult to provide a comprehensive evaluation of the image quality of the onboard camera.

[0006] Therefore, there is an urgent need for an automated method and system for monitoring the quality of on-orbit images of optical remote sensing satellites. This method can be carried out using remote sensing images of natural ground features and can comprehensively utilize evaluation indicators such as on-orbit MTF and on-orbit SNR to achieve automated monitoring of the quality of on-orbit images of remote sensing satellites. Summary of the Invention

[0007] The purpose of this invention is to address the technical problems of low automation and difficulty in obtaining effective samples in traditional remote sensing satellite on-orbit image quality monitoring, which makes it unsuitable for large-scale constellation image quality monitoring. This invention provides an automated method and system for monitoring the on-orbit image quality of optical remote sensing satellites. The method proposed in this invention can be carried out based on remote sensing images of natural ground features and comprehensively utilizes multiple evaluation indicators. The system proposed in this invention has a high degree of automation and can realize large-scale management of on-orbit satellite image quality issues.

[0008] To achieve the above objectives and solve the above technical problems, the technical solution of the present invention is as follows: An automated on-orbit image quality monitoring system for optical remote sensing satellites includes a mission planning subsystem, an image quality evaluation and analysis subsystem, and an operational early warning subsystem. The mission planning subsystem is responsible for formulating monitoring missions and has functions such as satellite orbit extrapolation, observation mission analysis and calculation, satellite control parameter generation, and management of a typical target library for monitoring missions. Monitoring missions are initiated in two ways: external triggering and internal triggering. External triggering is initiated by satellite operation and maintenance personnel. When operation and maintenance personnel determine that the satellite image quality is abnormal, they manually initiate a monitoring mission application. Internal triggering is initiated periodically by the mission planning subsystem, which uses the typical target library for monitoring missions to conduct routine monitoring of the satellite's on-orbit image quality. The image quality evaluation and analysis subsystem is used to complete the automated initial screening of regions of interest, the refined selection based on parameter evaluation, and the image quality evaluation calculation. The operational early warning subsystem is used to provide early warnings for image quality issues. The early warning logic is divided into two types: First, the evaluation results are compared with the standard values. The evaluation results given by the image quality analysis and evaluation subsystem are compared with the standard values ​​obtained during the satellite performance evaluation. If the evaluation results of five consecutive images are found to be lower than the standard values, an early warning will be issued. Second, based on the changes in statistical results in the same area, such as if the image quality analysis and evaluation subsystem shows a continuous downward trend in the evaluation results given for the same area over a period of time, an early warning will be issued. The alerts are sent to satellite maintenance personnel via dialog boxes, voice prompts, and animated alerts.

[0009] Furthermore, the typical target database for monitoring tasks consists of regions with characteristic information, such as large areas of farmland, Gobi desert, artificial calibration fields, airports, and coastlines. Elements such as the latitude and longitude of the center point, the latitude and longitude of the four corner points, the name, and the region type are added, modified, and deleted by the operation and maintenance personnel.

[0010] This invention provides an automated method for monitoring the quality of on-orbit images from optical remote sensing satellites. The method employs an automated on-orbit image quality monitoring system for optical remote sensing satellites and includes the following steps: Step 1: Monitoring Task Formulation Based on the requirements of external and internal triggers, the mission planning subsystem formulates the image quality monitoring task, sending the imaging task to the satellite for execution and the image quality analysis task to the image quality evaluation and analysis subsystem. Step 2, Satellite Imaging Step 3: After receiving satellite image data, automatically filter the region of interest. The image quality assessment and analysis subsystem performs satellite image quality analysis tasks. After receiving satellite image data, it selects level 0 image products without radiometric or geometric correction as the processing object. The steps for filtering regions of interest along ground features are as follows: 3.1 Extracting the image to be tested using the LSD line detection algorithm The linear features in the data are used to obtain the linear feature set L0; 3.2 Remove line features with a length of less than 15 pixels from L0 to obtain the filtered line feature set L1; 3.3 Remove straight line features from L1 that do not meet the edge angle threshold. The angle between the straight line and the horizontal or vertical direction of the image is between 4° and 8° to obtain the filtered straight line feature set L2. 3.4 Detect the pixel comparability within a range of at least 15 pixels on both sides of the straight line and the signal-to-noise ratio of the blade edge. SNR bright-dark It meets the following conditions: ， ， in, The average DN value for the high reflectivity region. The average DN value for the low reflectivity region. and It is the standard deviation of the mean values ​​for the high reflectivity region and the low reflectivity region; Based on the above conditions and the L2 linear feature set, the Region of Interest (ROI) set of the cutting edge that meets the MTF test conditions is selected. MTF Based on the angle ROI between the line and the horizontal direction of the image MTF水平 or the included angle of the vertical direction ROI MTF垂直 Screening of ROI regions along the edges of ground features; Step 4: Perform image quality assessment calculations on the selected images. Calculate the region set ROI MTF垂直 and ROI MTF水平 MTF value of each edge region: 4.1 Edge Detection In the edge image, the gray-scale transition points of each row are detected and used as the edge points of that row. The edge points of each row form the edge line. 4.2 Establishment of Edge Expansion Function For a row of the edge image, the edge expansion function is obtained by using the pixel index as the x-axis and the corresponding gray value as the y-axis. The edge expansion function of each row is interpolated, and the final edge expansion function is obtained by averaging and normalizing based on pixel alignment. 4.3 Establishment of Line Extension Function The formula for calculating the line spread function from the edge spread function is as follows:

[0011] In the formula, n is the pixel number; 4.4 Fourier Transform and Modulation Transfer Function Calculation The MTF is obtained by performing a Fourier transform on the line spread function and taking the modulus. The value corresponding to the Nyquist frequency point is the MTF. 4.5 Calculate the Region of Interest (ROI) SNR SNR values ​​for each region: Take ROI SNR In the rectangular region, the difference matrix is ​​obtained by subtracting the next row from the previous row. The signal-to-noise ratio of each column is calculated based on the difference matrix. Then, the average of multiple columns is calculated to obtain the region's signal-to-noise ratio. 4.5.1 Calculate the difference matrix

[0012] In the formula Let DN be the value in the i-th column and j-th row. The DN value in the (j+1)th row of the (i)th column.

[0013] 4.5.2 Calculate the noise for each column

[0014] In the formula, n is the number of rows in the image.

[0015] 4.5.3 Calculate the signal-to-noise ratio (SNR) for each column. i ) , in,

[0016] 4.5.4 Calculate the mean signal-to-noise ratio of multiple columns

[0017] In the formula, m is the number of columns in the image; ROI SNRThe SNR values ​​obtained from each region are taken as the arithmetic mean, which is used as the SNR value of the image under test. The product of MTF and SNR is used as the comprehensive image evaluation result of the image under test.

[0018] Step 5: Image quality problem warning The operational early warning subsystem is responsible for issuing early warnings for image quality issues.

[0019] Furthermore, in step 3.4, the steps for selecting farmland areas for ROI region screening are as follows: 1) Use region segmentation + OTSU algorithm for variable threshold processing of the image under test. Perform binarization segmentation to obtain a binarized image. ; 2) Using the erosion-expansion algorithm to... The data is processed, and closed regions are extracted to form a region set (ROI). MEAN ; 3) From ROI MEAN Regions with an average grayscale value not exceeding 25% of the image's full scale were selected as the test ROI set for SNR. SNR ,satisfy ， Where N is the camera quantization bit depth.

[0020] Furthermore, in step 4.4, the formula for calculating the Nyquist frequency point is as follows:

[0021] In the formula, n is the number of edge sampling points, and Δd is the interpolation interval of the edge spread function; ROI respectively MTF垂直 and ROI MTF水平 The MTF obtained from each region is calculated by taking the arithmetic mean. 垂直 and MTF 水平 The geometric mean of the two is used as the MTF value of the image under test.

[0022] The advantages of this invention compared to existing technologies are as follows: (1) This invention proposes an automated method for monitoring the quality of on-orbit images of optical remote sensing satellites. The method includes steps such as setting monitoring tasks, screening regions of interest, calculating image quality evaluation, and issuing early warnings for image quality problems.

[0023] (2) The present invention proposes an automated on-orbit image quality monitoring system for optical remote sensing satellites, which consists of a mission planning subsystem, an image quality evaluation and analysis subsystem, and an operational early warning subsystem.

[0024] (3) The automated image quality monitoring method proposed in this invention can effectively avoid the problems of difficulty in discovering image quality problems and difficulty in obtaining effective samples caused by human-in-the-loop. It can acquire feature information in panoramic remote sensing images at one time and is suitable for on-orbit image quality monitoring of large-scale constellations and wide-swath remote sensing satellites.

[0025] (4) The automated image quality monitoring system proposed in this invention can realize unmanned monitoring of satellite image quality through a full-link automated design scheme, which can effectively reduce the number of system maintenance personnel and has certain economic value. Attached Figure Description

[0026] Figure 1 A schematic diagram of the system composition and method flow of this invention; Figure 2 A schematic diagram of the ROI region screening results for the edge of the ground feature. Detailed Implementation

[0027] The implementation process of the present invention will be explained and described in detail below with reference to the accompanying drawings.

[0028] This invention proposes an automated method and system for monitoring the quality of on-orbit images of optical remote sensing satellites. The system includes steps such as monitoring task formulation, region of interest screening, image quality evaluation calculation, and image quality problem early warning. The system consists of a task planning subsystem, an image quality evaluation and analysis subsystem, and an operational early warning subsystem, which can realize the full automation of the monitoring process of optical remote sensing satellite image quality.

[0029] The closest prior art to this invention is a method and apparatus for selecting edge regions in remote sensing images (CN114399688A). This invention is a method and apparatus for extracting regions of interest (ROI) images containing edge images of ground features from single-band remote sensing images. This invention utilizes an edge region feature knowledge base and employs the Sobel operator and OTU algorithm to extract edge region ROI images.

[0030] Compared to that invention, the image quality monitoring system proposed in this invention has better comprehensiveness and systematicness, including the entire process of image quality monitoring task initiation, feature region selection, evaluation index calculation, and quality problem early warning, while CN114399688A only involves the single step of "feature region selection".

[0031] Meanwhile, the image quality monitoring method proposed in this invention comprehensively employs multiple evaluation indicators, including on-orbit MTF and on-orbit SNR, to more comprehensively reflect satellite image quality issues. In the selection of edge feature regions in this invention, the LSD algorithm and threshold-based feature set screening are used, and effective edge feature screening is performed based on the edge signal-to-noise ratio. Through theoretical analysis and verification with actual on-orbit image data, the edge ROI regions selected using the method proposed in this invention demonstrate better convergence and accuracy in on-orbit MTF evaluation, meeting the requirements for satellite image quality monitoring.

[0032] This invention provides an automated on-orbit image quality monitoring system for optical remote sensing satellites, comprising a mission planning subsystem, an image quality evaluation and analysis subsystem, and an operational early warning subsystem, the composition and process of which are as follows: Figure 1 As shown.

[0033] 1. Task Planning Subsystem The mission planning subsystem is mainly responsible for formulating monitoring missions and has functions such as satellite orbit simulation, observation mission analysis and calculation, satellite control parameter generation, and management of typical target database for monitoring missions.

[0034] Monitoring tasks can be initiated in two ways: external triggering and internal triggering. External triggering is initiated by satellite operation and maintenance personnel. When the operation and maintenance personnel determine that the satellite image quality is abnormal, they can manually initiate a monitoring task application. Internal triggering is initiated periodically by the task planning subsystem, which uses the typical target library of monitoring tasks to carry out routine monitoring of the satellite's on-orbit image quality.

[0035] The typical target database for monitoring tasks consists of regions with characteristic information, such as large areas of farmland, Gobi desert, artificial calibration fields, airports, and coastlines. Elements such as the latitude and longitude of the center point, the latitude and longitude of the four corner points, the name, and the region type can be added, modified, and deleted by operation and maintenance personnel.

[0036] 2. Image Quality Evaluation and Analysis Subsystem The image quality evaluation and analysis subsystem mainly completes the automated initial screening of regions of interest, the refined selection based on parameter evaluation, and the image quality evaluation calculation.

[0037] (1) Automated filtering of Region of Interest (ROI) After receiving satellite image data, the uncorrected (unradiographically and geometrically corrected) Level 0 image products are selected as the processing targets. An automated initial screening of Regions of Interest (ROIs) is performed. The screening steps for ROIs along the edges of ground features are as follows: 1) Extract the image to be tested using the LSD line detection algorithm. The linear features in the data are used to obtain the linear feature set L0; 2) Remove line features with a length of less than 15 pixels from L0 to obtain the filtered line feature set L1; 3) Remove straight line features from L1 that do not meet the edge angle threshold. The angle between the straight line and the horizontal or vertical direction of the image is between 4° and 8° to obtain the filtered straight line feature set L2. 4) Theoretical and experimental analysis shows that on-orbit MTF evaluation based on the edge-side method is not affected by the contrast of the edge-side region. However, to suppress the influence of random noise, the comparability and signal-to-noise ratio of the edge-side region must meet certain requirements. Therefore, the comparability of pixels within a range of at least 15 pixels on both sides of the detection line and the edge-side signal-to-noise ratio are required. SNR bright-dark It meets the following conditions: ， ， in, The average DN value for the high reflectivity region. The average DN value for the low reflectivity region. and It is the standard deviation of the mean values ​​for the high reflectivity region and the low reflectivity region.

[0038] Based on the above conditions and the L2 linear feature set, the ROI region set that meets the MTF test conditions is selected. MTF ROIs are categorized based on the angle between the line and the horizontal or vertical direction of the image. MTF垂直 and ROI MTF水平 .

[0039] Taking Jilin-1 satellite remote sensing image data as an example, we selected farmland areas for ROI (Region of Interest) screening, and the results are as follows: Figure 2 As shown.

[0040] The steps for screening ROI regions with uniform features are as follows: 1) Use region segmentation + OTSU algorithm for variable threshold processing of the image under test. Perform binarization segmentation to obtain a binarized image. .

[0041] 2) Using the erosion-expansion algorithm to... The data is processed, and closed regions are extracted to form a region set (ROI). MEAN .

[0042] 3) From ROI MEAN Regions with an average grayscale value not exceeding 25% of the image's full scale were selected as the test ROI set for SNR. SNR ,satisfy ， Where N is the camera quantization bit depth.

[0043] (2) Image quality assessment calculation Calculate the region set ROI MTF垂直 and ROI MTF水平 MTF value of each edge region: 1) Edge detection In the edge image, the gray-scale transition points of each row are detected and used as the edge points of that row. The edge points of each row form the edge line.

[0044] 2) Edge Spread Function (ESF) Establishment For a given row of the edge image, the edge expansion function is obtained by plotting the pixel index as the x-axis and the corresponding gray value as the y-axis. The edge expansion functions for each row are interpolated, and then averaged and normalized based on (sub)pixel alignment to obtain the final edge expansion function.

[0045] 3) Establishment of Line Spread Function (LSF) The formula for calculating the line spread function from the edge spread function is as follows:

[0046] In the formula, n is the pixel number. 4) Fourier transform and modulation transfer function (MTF) calculation The MTF is obtained by performing a Fourier transform on the line spread function and taking its modulus. The value corresponding to the Nyquist frequency is the value of the line spread function. The formula for determining the Nyquist frequency is as follows:

[0047] In the formula, n is the number of edge sampling points, and Δd is the interpolation interval of the edge spread function.

[0048] ROI respectively MTF垂直 and ROI MTF水平 The MTF obtained from each region is calculated by taking the arithmetic mean. 垂直 and MTF 水平 The geometric mean of the two is used as the MTF value of the image under test.

[0049] Calculate the region set ROI SNR SNR values ​​for each region: Take ROI SNR For a rectangular region, subtract the next row from the previous row to obtain a difference matrix. Calculate the signal-to-noise ratio of each column based on the difference matrix, and then calculate the average of multiple columns to obtain the region's signal-to-noise ratio.

[0050] 1) Calculate the difference matrix

[0051] In the formula Let DN be the value in the i-th column and j-th row. The DN value in the (j+1)th row of the (i)th column.

[0052] 2) Calculate the noise for each column.

[0053] In the formula, n is the number of rows in the image.

[0054] 3) Calculate the signal-to-noise ratio (SNR) for each column. i ) , in,

[0055] 4) Calculate the mean signal-to-noise ratio of multiple columns.

[0056] In the formula, m is the number of columns in the image.

[0057] ROI SNR The SNR values ​​obtained from each region are taken as the arithmetic mean and used as the SNR value of the image under test.

[0058] The product of MTF and SNR is used as the comprehensive image evaluation result of the image under test.

[0059]

[0060] 3. Operational Early Warning Subsystem The operational early warning subsystem mainly performs early warning of image quality problems. The early warning logic is divided into two types: First, the evaluation results are compared with the standard values. The evaluation results given by the image quality analysis and evaluation subsystem are compared with the standard values ​​obtained during the satellite performance evaluation. If the evaluation results of five consecutive images are found to be lower than the standard values, an early warning will be issued. Second, based on the changes in statistical results in the same area, such as if the image quality analysis and evaluation subsystem shows a continuous downward trend in the evaluation results given for the same area over a period of time, an early warning will be issued.

[0061] The warning methods include dialog boxes, voice prompts, and animated alerts, which send information about abnormal issues to satellite operation and maintenance personnel.

[0062] The above description, in conjunction with specific embodiments, provides a further detailed explanation of the present invention. It should not be construed that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, various simple deductions or substitutions can be made without departing from the concept of the present invention, and all such modifications and substitutions should be considered within the scope of protection of the present invention.

Claims

1. An automated method for monitoring the quality of on-orbit images from optical remote sensing satellites, characterized in that, This is achieved using an automated on-orbit image quality monitoring system for optical remote sensing satellites, including the following steps: Step 1: Monitoring Task Formulation Based on the requirements of external and internal triggers, the mission planning subsystem formulates the image quality monitoring task, sending the imaging task to the satellite for execution and the image quality analysis task to the image quality evaluation and analysis subsystem. Step 2, Satellite Imaging Step 3: After receiving satellite image data, automatically filter the region of interest. The image quality assessment and analysis subsystem performs satellite image quality analysis tasks. After receiving satellite image data, it selects level 0 image products without radiometric or geometric correction as the processing object. The steps for filtering regions of interest along ground features are as follows: 3.1 Extracting the image to be tested using the LSD line detection algorithm The linear features in the data are used to obtain the linear feature set L0; 3.2 Remove line features with a length of less than 15 pixels from L0 to obtain the filtered line feature set L1; 3.3 Remove straight line features from L1 that do not meet the edge angle threshold. The angle between the straight line and the horizontal or vertical direction of the image is between 4° and 8° to obtain the filtered straight line feature set L2. 3.4 Detect the pixel comparability within a range of at least 15 pixels on both sides of the straight line and the signal-to-noise ratio of the blade edge. SNR bright-dark It meets the following conditions: ， ， in, The average DN value for the high reflectivity region. The average DN value for the low reflectivity region. and It is the standard deviation of the mean values ​​for the high reflectivity region and the low reflectivity region; Based on the above conditions and the L2 linear feature set, the Region of Interest (ROI) set of the cutting edge that meets the MTF test conditions is selected. MTF Based on the angle ROI between the line and the horizontal direction of the image MTF水平 or the included angle of the vertical direction ROI MTF垂直 Screening of ROI regions along the edges of ground features; Step 4: Perform image quality assessment calculations on the selected images. Calculate the region set ROI MTF垂直 and ROI MTF水平 MTF value of each edge region: 4.1 Edge Detection In the edge image, the gray-scale transition points of each row are detected and used as the edge points of that row. The edge points of each row form the edge line. 4.2 Establishment of Edge Expansion Function For a row of the edge image, the edge expansion function is obtained by using the pixel index as the x-axis and the corresponding gray value as the y-axis. The edge expansion function of each row is interpolated, and the final edge expansion function is obtained by averaging and normalizing based on pixel alignment. 4.3 Establishment of Line Extension Function The formula for calculating the line spread function from the edge spread function is as follows: In the formula, n is the pixel number; 4.4 Fourier Transform and Modulation Transfer Function Calculation The MTF is obtained by performing a Fourier transform on the line spread function and taking its modulus; the value corresponding to the Nyquist frequency point is the Nyquist frequency point. The formula for determining the Nyquist frequency point is as follows: In the formula, n is the number of edge sampling points, and Δd is the interpolation interval of the edge spread function; ROI respectively MTF垂直 and ROI MTF水平 The MTF obtained from each region is calculated by taking the arithmetic mean. 垂直 and MTF 水平 The geometric mean of the two is used as the MTF value of the image under test; 4.5 Calculate the Region of Interest (ROI) SNR SNR values ​​for each region: Take ROI SNR In the rectangular region, the difference matrix is ​​obtained by subtracting the next row from the previous row. The signal-to-noise ratio of each column is calculated based on the difference matrix. Then, the average of multiple columns is calculated to obtain the region's signal-to-noise ratio. 4.5.1 Calculate the difference matrix In the formula Let DN be the value in the i-th column and j-th row. The DN value in the (j+1)th row of the i-th column; 4.5.2 Calculate the noise for each column In the formula, n is the number of rows in the image; 4.5.3 Calculate the signal-to-noise ratio (SNR) for each column. i ， 4.5.4 Calculate the mean signal-to-noise ratio of multiple columns In the formula, m is the number of columns in the image; ROI SNR The SNR values ​​obtained from each region are taken as the arithmetic mean, which is used as the SNR value of the image under test. The product of MTF and SNR is used as the comprehensive image evaluation result of the image under test. Step 5: Image quality problem warning The operational early warning subsystem is responsible for issuing early warnings for image quality issues.

2. The automated optical remote sensing satellite on-orbit image quality monitoring method according to claim 1, characterized in that, The automated optical remote sensing satellite on-orbit image quality monitoring system includes a mission planning subsystem, an image quality evaluation and analysis subsystem, and an operational early warning subsystem. The mission planning subsystem is responsible for formulating monitoring missions and has functions such as satellite orbit extrapolation, observation mission analysis and calculation, satellite control parameter generation, and management of a typical target library for monitoring missions. Monitoring missions are initiated in two ways: external triggering and internal triggering. External triggering is initiated by satellite operation and maintenance personnel. When operation and maintenance personnel determine that the satellite image quality is abnormal, they manually initiate a monitoring mission application. Internal triggering is initiated periodically by the mission planning subsystem, which uses the typical target library for monitoring missions to conduct routine monitoring of the satellite's on-orbit image quality. The image quality evaluation and analysis subsystem is used to complete the automated initial screening of regions of interest, the refined selection based on parameter evaluation, and the image quality evaluation calculation. The operational early warning subsystem is used to provide early warnings for image quality issues. The early warning logic is divided into two types: First, the evaluation results are compared with the standard values. The evaluation results given by the image quality analysis and evaluation subsystem are compared with the standard values ​​obtained during the satellite performance evaluation. If the evaluation results of five consecutive images are found to be lower than the standard values, an early warning will be issued. Second, based on the changes in statistical results in the same area, if the image quality analysis and evaluation subsystem shows a continuous downward trend in the evaluation results given for the same area over a period of time, an early warning will be issued. The alerts are sent to satellite maintenance personnel via dialog boxes, voice prompts, and animated alerts to indicate any abnormal issues.

3. The automated optical remote sensing satellite on-orbit image quality monitoring method according to claim 2, characterized in that, The typical target database for the monitoring task consists of regions with characteristic information, such as large areas of farmland, Gobi desert, artificial calibration fields, airports, and coastlines. The latitude and longitude of the center point, the latitude and longitude of the four corner points, the name, and the region type elements are added, modified, and deleted by the operation and maintenance personnel.

4. The automated on-orbit image quality monitoring method for optical remote sensing satellites according to claim 3, characterized in that, In step 3, the steps for selecting farmland areas for ROI (Region of Interest) screening are as follows: 1) Use region segmentation and OTSU algorithm for variable thresholding of the image under test. Perform binarization segmentation to obtain a binarized image. ; 2) Using the erosion-expansion algorithm to... The data is processed, and closed regions are extracted to form a region set (ROI). MEAN ; 3) From ROI MEAN Regions with an average grayscale value not exceeding 25% of the image's full scale were selected as the test ROI set for SNR. SNR ,satisfy ， Where N is the camera quantization bit depth.

Images