Satellite-borne optical push-broom image on-orbit correction method and device, storage medium and equipment

Abstract

The application discloses a kind of on-orbit correction method, device, storage medium and equipment of spaceborne optical push-broom image, it is related to remote sensing image processing technical field, improves push-broom image on-orbit correction efficiency and correction precision.It includes: obtaining the push-broom image to be corrected obtained by the ground coverage area of spaceborne optical shooting device shooting;Determine the relative radiometric calibration parameter of corresponding detector according to the pixel gray value detected by each detector, to carry out relative radiometric correction to image;Based on pseudo-invariant calibration field, obtain calibration image square window area, obtain the ideal pixel value of each band of calibration image square window area, determine absolute radiometric calibration parameter based on ideal pixel value and corresponding actual pixel value, carry out absolute radiometric correction to image after relative radiometric correction;According to the feature point matching relationship between key area in ground coverage area and push-broom image after absolute radiometric correction, carry out geometric correction to image after absolute radiometric correction.The application is suitable for push-broom image correction scene.

Description

Technical Field

[0001] This invention relates to the field of remote sensing image processing technology, and in particular to an on-orbit correction method, apparatus, storage medium and device for spaceborne optical pushbroom images. Background Technology

[0002] Spaceborne optical pushbroom cameras are one of the most commonly used imaging payloads on Earth observation remote sensing satellites. Pushbroom imaging uses a linear CCD / CMOS (Charge-coupled Device / Complementary Metal-Oxide Semiconductor) sensor to acquire images by scanning line by line along the satellite's flight path. Each line of the image corresponds to a different satellite position, attitude angle, and imaging time. Due to factors such as changes in sensor characteristics, space environment interference, differences in atmospheric conditions, and terrain undulations during the imaging process, pushbroom images are distorted, thus requiring correction.

[0003] Currently, on-orbit calibration and pushbroom image correction are typically achieved by mounting hardware such as solar diffusers and integrating spheres on satellites. However, due to constraints in size, weight, and power consumption, such hardware is difficult to deploy on small satellites, limiting its applicability. Furthermore, the calibration hardware itself experiences degradation, requiring continuous accuracy calibration, which is time-consuming and labor-intensive. Incorrect calibration will affect the accuracy of pushbroom image correction. Moreover, the calibration frequency is limited by the hardware's lifespan, necessitating frequent hardware replacements, thus increasing calibration costs and consequently, pushbroom image correction costs. Summary of the Invention

[0004] This invention provides an on-orbit correction method, apparatus, storage medium, and device for spaceborne optical pushbroom images, which mainly improves the correction efficiency and accuracy of spaceborne optical pushbroom images and saves correction costs.

[0005] According to a first aspect of the present invention, an on-orbit correction method for spaceborne optical pushbroom images is provided, comprising: Acquire pushbroom images of the ground coverage area captured by the spaceborne optical imaging device, which are to be corrected. Identify the calibration reference data source in the ground coverage area, determine the calibration reference image area in the pushbroom image based on the calibration reference data source, determine the current relative radiometric calibration parameter of the corresponding probe based on the pixel gray value detected by each probe in the calibration reference image area, and perform relative radiometric correction on the pushbroom image element by element based on the current relative radiometric calibration parameter. Based on the geographical range of the preset pseudo-invariant calibration field, the projection area of ​​the preset pseudo-invariant calibration field in the pushbroom image after relative radiometric correction is located to obtain the calibration image window area. The ideal pixel gray values ​​of each band in the calibration image window area corresponding to the preset pseudo-invariant calibration field are obtained. Based on the ideal pixel gray values ​​and the actual pixel gray values ​​of each band in the calibration image window area in the pushbroom image, the current absolute radiometric calibration parameters of the corresponding band are determined, and the absolute radiometric correction of the pushbroom image after relative radiometric correction is performed. The location of a key area within the ground coverage region in the absolute radiometrically corrected pushbroom image is identified as a registration window region. Within the registration window region, geometric correction is performed on the absolute radiometrically corrected pushbroom image based on the feature point matching relationship between the key area and the absolute radiometrically corrected pushbroom image.

[0006] Optionally, identifying the calibration reference data source in the ground coverage area includes: The overlapping area between the ground coverage area and the preset uniform desert area is determined. If the area of ​​the overlapping area is less than the preset area threshold, the calibration reference data source is determined to be a cloud data source. If the area of ​​the overlapping area is greater than or equal to the preset area threshold, the calibration reference data source is determined to be a desert data source.

[0007] Optionally, determining a calibration reference image region in the pushbroom image based on the calibration reference data source includes: When the calibration reference data source is the cloud data source, the historical relative radiometric calibration parameters of each probe element are determined; Based on the historical relative radiometric calibration parameters, the mean pixel gray value and standard deviation of each probe element in the cloud region of the pushbroom image are calculated row by row. Based on the standard deviation of pixel gray value, the intra-row pixel gray value variation amplitude of the corresponding row pixels is determined. Based on the mean pixel gray value, the inter-row pixel gray value variation amplitude of adjacent rows pixels is determined. The cloud region in which the intra-row pixel gray value variation amplitude and the inter-row pixel gray value variation amplitude of a consecutive preset number of rows are both less than the corresponding preset amplitude threshold is taken as the calibration reference image region. When the calibration reference data source is the desert data source, the projection area of ​​the preset uniform desert region in the push-broom image is used as the calibration reference image region.

[0008] Optionally, the current relative radiometric calibration parameters of the corresponding detector element are determined based on the pixel grayscale values ​​detected by each detector element in the calibration reference image region, including: Determine the average pixel grayscale value detected by all detectors in the calibration reference image region: ,in, The average pixel grayscale value common to all probes, The total number of detector elements in the calibration reference image area, For the first The average gray value of pixels detected by each probe in the calibration reference image region; The current relative radiometric correction gain for each detector element is determined based on the average pixel grayscale value: ,in, For the first The current relative radiation correction gain of each detector element; The current relative fused radiometric correction gain for each probe is determined based on the current relative radiometric correction gain: ,in, For the first The current relative fusion radiation correction gain of each detector, For smoothing coefficient, For the first Historical relative radiation correction gain of each detector element; The current relative convergent radiometric correction bias for each element is determined based on the current relative radiometric correction bias: ,in, For the first The current relative fusion radiometric correction bias of each probe, For the first The current relative radiometric correction bias of each detector element is set to zero. For the first Historical relative radiation correction bias of each detector element; The current relative radiometric calibration parameters of the corresponding detector element are determined based on the current relative fused radiometric correction gain and the current relative fused radiometric correction bias of each detector element. Specifically, the current relative radiometric calibration parameters for each detector element are determined based on its current relative fused radiometric correction gain and current relative fused radiometric correction bias, including: If the calibration reference image region is a cloud region, the historical relative fused radiometric correction gain determined most recently based on the projection region of a uniform desert is obtained. If the gain difference between the current relative fused radiometric correction gain and the historical relative fused radiometric correction gain exceeds a preset gain threshold, the historical relative radiometric calibration parameter determined most recently based on the projection region of a uniform desert is used as the current relative radiometric calibration parameter of the corresponding probe. If the gain difference does not exceed the preset gain threshold, the current relative fused radiometric correction gain and the current relative fused radiometric correction offset of each probe are used as the current relative radiometric calibration parameter of the corresponding probe.

[0009] Optionally, the method further includes: If the key area does not exist in the ground coverage area, multiple virtual control points are selected in the push-broom image after absolute radiometric correction based on digital elevation information. The geometric correction coefficient of the current pixel coordinate is determined based on the pixel coordinate and ground physical coordinate of each virtual control point. The geometric correction of the push-broom image after absolute radiometric correction is performed based on the geometric correction coefficient of the current pixel coordinate.

[0010] Optionally, multiple virtual control points are selected in the absolute radiometrically corrected pushbroom image based on digital elevation information, including: The elevation range is determined based on the digital elevation information of the ground coverage area, the number of elevation layers in the ground coverage area is determined based on the elevation range, and the ground coverage area is divided into multiple elevation layers based on the number of elevation layers. The absolute radiometrically corrected pushbroom image is divided into multiple grids. For each grid, an elevation coverage area belonging to an arbitrary elevation layer is determined in the grid, and multiple candidate virtual control points are determined in the elevation coverage area. Select multiple virtual control points from the multiple candidate virtual control points corresponding to each grid.

[0011] Optionally, multiple virtual control points are selected from the multiple candidate virtual control points corresponding to each grid, including: For each candidate virtual control point, a rectangular window is determined with the ground projection position of the candidate virtual control point as the center, the ground slope angle within the rectangular window is determined, and the slope factor of the candidate virtual control point is determined based on the ground slope angle. Determine the elevation standard deviation of multiple elevation values ​​within the rectangular window, and determine the elevation difference anomaly factor of the candidate virtual control point based on the elevation standard deviation; Determine the physical coordinates of the ground projection position of the candidate virtual control point and its image coordinates in the corresponding grid. Based on the line-of-sight vector pointing from the image coordinates to the physical coordinates, determine the slope normal vector of the ground slope where the ground projection position is located. Determine the vector angle between the line-of-sight vector and the slope normal vector. Based on the vector angle, determine the line-of-sight angle factor of the candidate virtual control point. The distance between the candidate virtual control point and the image center of the pushbroom image after absolute radiometric correction is determined, the diagonal length of the pushbroom image after absolute radiometric correction is determined, and the image-side position factor of the candidate virtual control point is determined based on the ratio of the distance to the diagonal length. The evaluation scores of the candidate virtual control points are determined based on the slope factor, the height difference anomaly factor, the line-of-sight-slope angle factor, and the image position factor. Multiple virtual control points are selected from each candidate virtual control point based on the evaluation scores.

[0012] Optionally, obtaining the ideal pixel grayscale values ​​of each band in the calibration image window region corresponding to the preset pseudo-invariant calibration field includes: Determine the band pixel reference information for each band of the preset pseudo-invariant calibration field, wherein the band pixel reference information includes reference apparent reflectance, solar zenith angle, and Earth-Sun distance factor; Based on the band pixel reference information, determine the ideal pixel grayscale value of each band in the calibration image window region: ,in, For band Ideal pixel grayscale value For band Reference apparent reflectance, For the solar zenith angle, Earth-Sun distance factor For band Radiative calibration conversion coefficient; The current absolute radiometric calibration parameters for the corresponding bands are determined based on the ideal pixel grayscale value and the actual pixel grayscale value of each band in the calibration image window region of the pushbroom image, including: Determine the current absolute radiation correction gain for the band: ,in, For band The current absolute radiation correction gain, For band The average actual pixel grayscale value in the calibration image window area, For band Historical absolute radiation correction bias; Determine the current absolute radiometric correction bias for the band: ,in, For band The current absolute radiation correction bias; Determine the current absolute fused radiative correction gain for the band: ,in, For band The current absolute fusion radiative correction gain, For the moving average smoothing coefficient, For band Historical absolute radiation correction gain; Determine the current absolute fused radiative correction bias for the band: ,in, For band The current absolute fusion radiative correction bias, For band Historical absolute radiation correction bias; The current absolute radiometric calibration parameters for the corresponding band are determined based on the current absolute fused radiometric correction gain and the current absolute fused radiometric correction bias.

[0013] According to a second aspect of the present invention, an on-orbit correction device for spaceborne optical pushbroom images is provided, comprising: The acquisition unit is used to acquire pushbroom images of the ground coverage area to be corrected obtained by the spaceborne optical imaging device. The relative radiometric correction unit is used to identify the calibration reference data source in the ground coverage area, determine the calibration reference image area in the pushbroom image based on the calibration reference data source, determine the current relative radiometric calibration parameter of the corresponding probe based on the pixel gray value detected by each probe in the calibration reference image area, and perform relative radiometric correction on the pushbroom image element by element based on the current relative radiometric calibration parameter. An absolute radiometric correction unit is used to locate the projection area of ​​the preset pseudo-invariant calibration field in the pushbroom image after relative radiometric correction, based on the geographical range of the preset pseudo-invariant calibration field, to obtain the calibration image window area, obtain the ideal pixel gray value of each band in the calibration image window area corresponding to the preset pseudo-invariant calibration field, and determine the current absolute radiometric calibration parameters of the corresponding band based on the ideal pixel gray value and the actual pixel gray value of each band in the calibration image window area in the pushbroom image, and perform absolute radiometric correction on the pushbroom image after relative radiometric correction. The geometric correction unit is used to identify the position of the key area in the ground coverage area within the pushbroom image after absolute radiometric correction as a registration window area. Within the registration window area, geometric correction is performed on the pushbroom image after absolute radiometric correction based on the feature point matching relationship between the key area and the pushbroom image after absolute radiometric correction.

[0014] According to a third aspect of the present invention, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the above-described on-orbit correction method for spaceborne optical pushbroom images.

[0015] According to a fourth aspect of the present invention, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the above-described on-orbit correction method for spaceborne optical pushbroom images.

[0016] The present invention provides an on-orbit correction method, apparatus, storage medium, and device for spaceborne optical pushbroom images. Compared with the current method of correcting pushbroom images by relying on hardware such as solar diffusers and integrating spheres on satellites, the present invention corrects pushbroom images by using existing reference image data sources, preset pseudo-invariant calibration fields, and key areas on the satellite. That is, it only uses the imaging data stored on the satellite to correct the pushbroom images, without relying on calibration hardware equipment. This can save image correction resources and improve the correction efficiency and accuracy of pushbroom images. At the same time, the present invention can directly correct pushbroom images at the satellite end, avoiding the time and bandwidth consumption of transmitting pushbroom images to the ground for correction. Specifically, this invention eliminates image stripes or non-uniformity caused by inconsistent responses of different detectors in spaceborne optical imaging devices by obtaining a calibration reference data source, determining relative radiometric calibration parameters for each detector element based on this data source, and performing correction. This ensures the relative radiometric consistency within the image. By utilizing the deviation between the known ideal pixel gray values ​​and the actual pixel gray values ​​in the image from each band of the pseudo-invariant calibration field, the current absolute radiometric calibration parameters are solved band by band. Utilizing the long-term stable physical property of the pseudo-invariant field's spectral characteristics, a traceable true radiometric reference is provided for the image gray values. Using the projection area of ​​the pseudo-invariant calibration field as the calibration image window, the correction range is precisely constrained within the effective area, ensuring calibration accuracy and improving the precision of absolute radiometric correction. By using the feature point matching relationship between the key area and the pushbroom image to solve the geometric transformation relationship, sub-pixel-level high-precision registration can be achieved, fully utilizing the accuracy advantages of external reference data. Attached Figure Description

[0017] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings: Figure 1 This invention provides a flowchart of an on-orbit correction method for spaceborne optical pushbroom images according to an embodiment of the present invention. Figure 2 This invention provides a flowchart of another on-orbit correction method for spaceborne optical pushbroom images according to an embodiment of the present invention. Figure 3 This diagram illustrates the structure of an on-orbit correction device for spaceborne optical pushbroom images provided in an embodiment of the present invention. Figure 4 This invention provides a schematic diagram of another on-orbit correction device for spaceborne optical pushbroom images. Figure 5 A schematic diagram of the physical structure of a computer device provided in an embodiment of the present invention is shown. Detailed Implementation

[0018] The present invention will be described in detail below with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in the present application can be combined with each other.

[0019] Currently, the method of correcting pushbroom images by using hardware such as solar diffusers and integrating spheres mounted on satellites has limitations. The calibration hardware itself is subject to degradation, requiring continuous accuracy calibration, which is time-consuming and labor-intensive. If the accuracy calibration is incorrect, it will affect the correction accuracy of the pushbroom images. Furthermore, the calibration frequency is limited by the lifespan of the hardware, requiring constant hardware replacement, which increases the calibration cost.

[0020] To address the aforementioned problems, embodiments of the present invention provide an on-orbit correction method for spaceborne optical pushbroom images, such as... Figure 1 As shown, the method includes: 101. Obtain pushbroom images of the ground coverage area captured by the spaceborne optical imaging device, which are to be corrected.

[0021] In this embodiment of the invention, the spaceborne optical imaging device, such as the spaceborne camera, acquires pushbroom images to be corrected during its on-orbit operation via pushbroom imaging, and simultaneously records the corresponding metadata, such as satellite position, satellite velocity, attitude angle, camera in-situ orientation elements, and pixel grayscale values ​​for each band. The on-board storage system contains multiple databases, namely, a detector relative radiometric correction coefficient table database, a uniform scene database (i.e., a preset on-board uniform desert database), a PICS (Pseudo Invariant Calibration Sites) database (i.e., a preset pseudo invariant calibration field database), a key area base map database, and a global DEM (Digital Elevation Model) database. The detector relative radiometric correction coefficient table stores the relative radiometric correction coefficients for each band and detector in historical images. These coefficients can be obtained through laboratory pre-calibration and updated through relative radiometric calibration during in-orbit operation. The uniform scene database stores the geographical extent of several large-area uniform desert hinterlands around the world, which is used to trigger the judgment of in-orbit relative radiometric calibration. The PICS database stores the geographical extent and multispectral reference reflectance data of several pseudo-invariant calibration fields. The key area base map library stores base maps of several key areas containing geometric information, such as compressed orthophotos and geographic coordinate mapping tables. The global DEM database stores global digital elevation model data.

[0022] 102. Identify the calibration reference data source in the ground coverage area, determine the calibration reference image area in the pushbroom image based on the calibration reference data source, determine the current relative radiometric calibration parameter of the corresponding probe based on the pixel gray value detected by each probe in the calibration reference image area, and perform relative radiometric correction on the pushbroom image element by element based on the current relative radiometric calibration parameter.

[0023] The calibration reference data sources include, but are not limited to, uniform desert data sources and uniform cloud data sources. If the calibration reference data source is a uniform desert data source, then the calibration reference image area is a desert area; if the calibration reference data source is a uniform cloud data source, then the calibration reference image area is a cloud area.

[0024] In this embodiment of the invention, a reference data source for probe relative radiometric calibration is determined based on information such as the type of ground cover area in the pushbroom image and a uniform scene database. This data source can be a uniform ground cover area, such as a pseudo-invariant target area like a uniform desert, or a uniform cloud area. The area in the data source has a corresponding high-quality orthorectified image area stored on the satellite, which is free from radiometric and geometric distortions. That is, the surface radiometric characteristics of the area are stable, there is no cloud cover, the geographic coordinates and image-side coordinates of the uniform ground cover area are clearly known, and the surface cover of the area is uniform and the terrain is flat. Then, the overlapping area between the ground cover area of ​​the pushbroom image and the image in the calibration reference data source is determined, and then the window area of ​​the overlapping area in the pushbroom image is calculated as the calibration reference image area. Furthermore, the average pixel grayscale value of each detector element in the calibration reference image area is statistically analyzed, and the total number of detector elements is determined. Finally, the current relative gain of each detector element is determined based on the average pixel grayscale value and the total number of detector elements. Then, the current relative radiometric calibration parameters of each detector element are determined by combining the historical relative gains and biases of the detector elements stored on-board. Finally, relative radiometric correction is performed element-by-element on the pushbroom image to be calibrated according to the following formula:

[0025] in, For the exploration The corresponding push-broom image number Linear correction of pixel grayscale values For the current relative radiation calibration parameters, the probe element Current relative gain, For the exploration The corresponding push-broom image number Pixel grayscale values ​​before line correction For the current relative radiation calibration parameters, the probe element The current bias, during image on-orbit correction. It can be approximated as 0.

[0026] This invention utilizes on-board stored image data sources to perform relative radiometric correction on pushbroom images. This avoids the resource consumption of external hardware and the correction errors and time wasted on external hardware calibration. Consequently, this invention improves the relative radiometric correction efficiency and accuracy of pushbroom images while saving correction costs. Furthermore, by acquiring a calibration reference data source and determining and correcting the relative radiometric calibration parameters element by element based on this data source, this invention eliminates image stripes or non-uniformity caused by inconsistent responses among different elements in the pushbroom image acquisition device, ensuring the relative radiometric consistency within the image.

[0027] 103. Based on the geographical range of the preset pseudo-invariant calibration field, locate the projection area of ​​the preset pseudo-invariant calibration field in the pushbroom image after relative radiometric correction to obtain the calibration image window area. Obtain the ideal pixel gray values ​​of each band in the calibration image window area corresponding to the preset pseudo-invariant calibration field. Based on the ideal pixel gray values ​​and the actual pixel gray values ​​of each band in the calibration image window area in the pushbroom image, determine the current absolute radiometric calibration parameters of the corresponding band and perform absolute radiometric correction on the pushbroom image after relative radiometric correction.

[0028] The database of preset pseudo-invariant calibration fields stores multiple preset pseudo-invariant calibration fields, as well as their geographical range (latitude and longitude), reference apparent reflectance in each multispectral band, solar zenith angle range, and available seasonal windows. A pseudo-invariant calibration field is a specific geographical area whose surface reflectance and other physical properties remain relatively stable over a long period, minimally affected by seasonal and weather changes, and whose radiation characteristics do not change. This includes, but is not limited to, specific bare land, large bodies of water, and vast desert areas. It also stores band pixel reference information, including but not limited to reference apparent reflectance, solar zenith angle, and Earth-Sun distance factor.

[0029] For embodiments of the present invention, absolute radiometric correction needs to be performed on the pushbroom image after relative radiometric correction. Based on the ground coverage of the pushbroom image, its intersection with the geographical range of all pseudo-invariant calibration fields in the preset pseudo-invariant calibration field database is calculated one by one. If the area of ​​the overlapping region with any pseudo-invariant calibration field is greater than a preset threshold set according to actual needs, it is determined that the pushbroom image covers the pseudo-invariant calibration field, triggering this absolute radiometric calibration update. If the area of ​​the overlapping region with the geographical range of all pseudo-invariant calibration fields is less than or equal to the preset threshold, this absolute radiometric calibration update is not triggered, and the historical absolute radiometric calibration parameters stored in the onboard database are directly used to perform absolute radiometric correction on the pushbroom image. If this absolute radiometric calibration update is triggered, the area of ​​the calibration field overlap region on the pushbroom image is calculated as the calibration image window region. This calibration image window region contains the start row, end row, start probe, and end probe of the image. Next, cloud detection is performed on the calibration image window area. Pixels identified as clouds can be removed using spectral thresholding or spatial texture methods. Pixels that are not removed are considered valid pixels. The number of valid pixels is counted. If the number of valid pixels is too small or too large (i.e., the number of valid pixels is less than the first threshold or greater than the second threshold, or the first threshold is less than the second threshold), the determination of the current absolute radiometric calibration parameters is prohibited. Instead, the historical absolute radiometric calibration parameters for this pushbroom camera stored in the onboard database are directly used to perform absolute radiometric correction on the image. Conversely, if the number of effective pixels is greater than or equal to the first threshold and less than or equal to the second threshold, the update of the absolute radiometric calibration parameters is triggered. During the update process, the ideal pixel grayscale values ​​of different band images within the calibration image window area are first determined based on the band pixel reference information, and the actual pixel grayscale values ​​of different band images within the calibration image window area are also determined. Based on the ideal and actual pixel grayscale values, the current absolute radiometric gain for the corresponding band is determined. Then, the current absolute radiometric calibration parameters are determined by combining the historical absolute radiometric gains stored on-board for this pushbroom camera. These current absolute radiometric calibration parameters include the current absolute radiometric calibration gain and the current absolute radiometric calibration bias. Finally, the image is subjected to absolute radiometric correction using the current absolute radiometric calibration parameters, specifically according to the following formula:

[0030] in, For the bands in the pushbroom image Pixel grayscale values ​​after absolute radiometric correction of the image For band Current absolute radiometric calibration gain of the image, For band Pixel grayscale values ​​before absolute radiometric correction of the image For band The current absolute radiometric calibration bias of the image.

[0031] This invention utilizes a pre-stored pseudo-invariant calibration field database on-board to perform absolute radiometric correction on pushbroom images. This avoids the resource consumption of external hardware and the correction errors and time wasted on external hardware calibration. Consequently, this invention improves the efficiency and accuracy of absolute radiometric correction for pushbroom images while saving correction costs. Furthermore, this invention solves for the current absolute radiometric calibration parameters band by band by band by utilizing the deviation between the known ideal pixel grayscale values ​​and the actual pixel grayscale values ​​in the image from each band of the pseudo-invariant calibration field. Leveraging the long-term stable physical properties of the pseudo-invariant field's spectral characteristics, it provides a traceable true radiometric reference for the image grayscale values. By using the calibration projection area as the calibration image window, the correction range is precisely constrained within the effective area, ensuring calibration accuracy and improving the precision of absolute radiometric correction.

[0032] 104. Identify the location of key areas in the ground coverage area within the pushbroom image after absolute radiometric correction as the registration window area. Within the registration window area, perform geometric correction on the pushbroom image after absolute radiometric correction based on the feature point matching relationship between the key areas and the pushbroom image.

[0033] In this embodiment of the invention, a base map library for key areas is maintained on-board. The base maps in the library can be selected from high-precision orthophoto products of key areas, with geometric positioning accuracy meeting the requirements. The base maps are accompanied by complete geographic coordinate mapping information, such as an image-object coordinate mapping table, so that each pixel of the base map has a clear latitude and longitude correspondence. The base map library for key areas stores the geographic range (represented by a latitude and longitude bounded rectangle) of several preset key areas and the corresponding base maps containing geometric information. Each pixel on the base map has precise geographic coordinates and clearly defines the base map projection method and ellipsoid parameters used. At the same time, the base map is an orthorectified image, that is, it contains a high-precision digital elevation model, and each pixel on it represents the actual position when viewed vertically downward from directly above. Key areas refer to specific ground coverage areas with high-precision prior geographic information and distinct ground features, suitable for high-precision image registration and geometric correction. These key areas contain standard geometric information, including geographic coordinates and image-side coordinates from pushbroom cameras. The base map of a key area includes easily identifiable ground features such as road intersections, right angles of buildings, and dams. Key areas include, but are not limited to, specific ports, airports, and border crossings. The base maps in this key area base map library are uploaded to the satellite from ground stations via the telemetry, tracking, and command (TT&C) link. During onboard storage, to save space, the base maps are stored in a compressed format, retaining only the bands required for geometric correction. Each key area stores its optimal base map, and the geographic coordinate mapping table is stored independently, recording the base map data. Figure 4Angular latitude and longitude and pixel resolution. The base maps in this base map library are constantly updated. For example, an image with absolute radiometric correction may be used to replace the corresponding area base map in the base map library. Furthermore, if updates cause the total storage capacity of the base map library to exceed a preset limit, the oldest base map backup will be discarded according to the least used strategy, and the latest version will be retained.

[0034] Specifically, geometric correction needs to be performed on the image after absolute radiometric correction. During geometric correction, the overlap between the ground cover area and the geographical extent of each key area is calculated. If the overlap of any key area is greater than a threshold set according to actual needs, a key area is determined to exist in the pushbroom image. If the overlap with all key areas is less than or equal to the threshold, no key area is determined to exist in the pushbroom image. If multiple key areas exist in the image, the key area with the highest overlap is selected for subsequent geometric correction. During the geometric correction of the image based on key areas, the latitude and longitude coordinates of the four corners of the key area base map are inversely calculated to the image-side coordinate system of the pushbroom image to obtain the corresponding area of ​​the key area on the pushbroom image as the registration window area. For example, for the key area base map... Figure 4 For each corner point, the object-side coordinates are converted to the image-side coordinates to obtain four image-side corner points in the pushbroom image. The bounding rectangles of these four corners serve as the registration window area, narrowing the registration range from the entire image to within this window and saving registration time. Next, feature points are extracted from the key area base map and the pushbroom image within the registration window. The similarity between each feature point in the key area base map and each feature point in the pushbroom image is determined. Feature points are matched based on similarity, with the most similar feature points being selected as feature point pairs. Accuracy checks are then performed on the feature point pairs, eliminating incorrectly matched pairs. Based on the feature point pairs' coordinates (including ground features from the key area base map and pixel coordinates from the pushbroom image), the least squares method is used to estimate the geometric transformation relationship from the pushbroom image to the key area base map. This geometric transformation relationship is then used to perform geometric transformation correction on the image. For example, an affine transformation model or a perspective transformation model can be used to determine the geometric transformation relationship. The process of using an affine transformation model for geometric correction is shown below:

[0035] in,( , ) represents the geometrically corrected pixel coordinates in the push-broom image. For the pixel coordinates in the push-broom image before geometric correction, , , , , and These are the affine transformation parameters in the affine transformation model, which is determined based on the feature point pairs' coordinates of the ground objects and pixel coordinates.

[0036] The process of geometrically correcting an image using a perspective transformation model is shown below:

[0037] in, , , , , , , , and These are the perspective transformation parameters in the perspective transformation model, which is determined based on the feature point pairs' coordinates and pixel coordinates.

[0038] Therefore, geometric correction can be performed on the pushbroom image using any of the methods described above. The geometric transformation relationships of the image, as well as the pixel coordinates before and after correction, can then be stored on-board for reference in subsequent geometric corrections.

[0039] In another embodiment of the present invention, when determining the key region base map and the feature point pairs of the push-broom image within the registration window, if the number of feature point pairs is greater than a preset threshold set according to actual needs, the aforementioned geometric correction process is performed. If the number of feature point pairs is less than or equal to the preset threshold, it indicates that the key region base map and the push-broom image differ too much, and the process of using the key region base map for geometric correction is abandoned. Virtual control points are used to perform geometric correction on the image. This avoids the problem of low correction accuracy caused by using a base map with excessive differences as a reference for geometric correction.

[0040] This invention adaptively selects a geometric correction path by determining whether a key area with precise location and orthophoto exists. When a key area exists, the geometric transformation relationship is solved by using the key area and feature point pairs, which can achieve sub-pixel-level high-precision registration and make full use of the accuracy advantage of external reference image data. When no key area exists, the geometric correction is completed by selecting virtual control points based solely on digital elevation information without relying on any external reference image, ensuring that geometric correction will be carried out smoothly in any scenario.

[0041] The present invention provides an on-orbit correction method for spaceborne optical pushbroom images. Compared with the current method of correcting pushbroom images by using hardware such as solar diffusers and integrating spheres on satellites, the present invention corrects pushbroom images by using existing reference image data sources, preset pseudo-invariant calibration fields, and key areas on the satellite. That is, it only uses the imaging data stored on the satellite to correct the pushbroom images, without relying on calibration hardware equipment. This saves image correction resources and improves the correction efficiency and accuracy of pushbroom images. At the same time, the present invention can directly correct pushbroom images at the satellite end, avoiding the time and bandwidth consumption of transmitting pushbroom images to the ground for correction. Specifically, this invention eliminates image stripes or non-uniformity caused by inconsistent responses of different detectors in spaceborne optical imaging devices by obtaining a calibration reference data source, determining relative radiometric calibration parameters for each detector element based on this data source, and performing correction. This ensures the relative radiometric consistency within the image. By utilizing the deviation between the known ideal pixel gray values ​​and the actual pixel gray values ​​in the image from each band of the pseudo-invariant calibration field, the current absolute radiometric calibration parameters are solved band by band. Utilizing the long-term stable physical property of the pseudo-invariant field's spectral characteristics, a traceable true radiometric reference is provided for the image gray values. Using the projection area of ​​the pseudo-invariant calibration field as the calibration image window, the correction range is precisely constrained within the effective area, ensuring calibration accuracy and improving the precision of absolute radiometric correction. By using the feature point matching relationship between the key area and the pushbroom image to solve the geometric transformation relationship, sub-pixel-level high-precision registration can be achieved, fully utilizing the accuracy advantages of external reference data.

[0042] Furthermore, to better illustrate the above-described process of on-orbit correction of spaceborne optical pushbroom images, as a refinement and extension of the above embodiments, this invention provides another method for on-orbit correction of spaceborne optical pushbroom images, such as... Figure 2 As shown, the method includes: 201. Obtain pushbroom images of the ground coverage area captured by the spaceborne optical imaging device, which are to be corrected.

[0043] Specifically, the images captured by the optical pushbroom camera on board the satellite are used as the pushbroom images to be calibrated.

[0044] 202. Determine the overlapping area between the ground coverage area and the preset uniform desert area. If the area of ​​the overlapping area is less than the preset area threshold, the calibration reference data source is determined to be the cloud data source. If the area of ​​the overlapping area is greater than or equal to the preset area threshold, the calibration reference data source is determined to be the desert data source.

[0045] Among them, the band emissivity and reflectivity of the uniform desert region in the preset on-board uniform desert database are relatively stable and there are no defects such as radiation distortion. At the same time, the band emissivity and reflectivity of the cloud region are relatively stable and there are no defects such as radiation distortion. The preset area threshold is a value set according to actual needs.

[0046] In this embodiment of the invention, if the overlapping area is less than or equal to a preset area threshold, and the number of days since the last successful relative radiometric calibration update exceeds a preset number of days, the cloud region in the cloud data source needs to be used to perform absolute radiometric correction on the pushbroom image. If the number of days since the last successful relative radiometric calibration update does not exceed the preset number of days, the calibration parameters from the last successful relative radiometric calibration can be directly used to perform relative radiometric correction on the image. If the overlapping area is greater than the preset area threshold, the desert region is used to perform relative radiometric correction on the image.

[0047] 203. Based on the calibration reference data source, determine the calibration reference image region in the pushbroom image. Based on the pixel grayscale value detected by each probe in the calibration reference image region, determine the current relative radiometric calibration parameter of the corresponding probe. Based on the current relative radiometric calibration parameter, perform relative radiometric correction on the pushbroom image element by element.

[0048] In this embodiment of the invention, if the calibration reference data source is a cloud data source, it is necessary to determine the calibration reference image region based on the cloud data source. Therefore, step 203 specifically includes: when the calibration reference data source is the cloud data source, determining the historical relative radiometric calibration parameters for each probe; calculating the mean pixel gray level and standard deviation of pixel gray level for each probe row by row in the cloud region of the pushbroom image based on the historical relative radiometric calibration parameters; determining the intra-row pixel gray level variation amplitude based on the pixel gray level standard deviation; determining the inter-row pixel gray level variation amplitude based on the mean pixel gray level; and taking the cloud region where the intra-row pixel gray level variation amplitude and inter-row pixel gray level variation amplitude of a consecutive preset number of rows are both less than the corresponding preset amplitude threshold as the calibration reference image region.

[0049] The historical relative radiometric calibration parameters for each probe element include historical relative radiometric correction gain and historical relative radiometric correction bias. Specifically, firstly, cloud regions are identified in the pushbroom image, and these cloud regions are randomly or equally divided into multiple cloud sub-regions as candidate cloud sub-regions. For each row of images in each candidate sub-region, the mean pixel grayscale value and standard deviation of each probe element in that row are calculated based on the historical relative radiometric calibration parameters. For each candidate cloud sub-region, a cloud sub-region is selected where the variation amplitude of the standard deviation of pixel grayscale within consecutive rows is less than a certain value set according to actual needs, and the variation amplitude of the mean pixel grayscale value between adjacent rows is also less than a certain value. This uniform cloud sub-region is then used as the calibration reference image region. This embodiment of the invention, through the dual constraints of the variation amplitude of pixel grayscale within rows and the variation amplitude of pixel grayscale between rows, can accurately select cloud sub-regions with extremely uniform optical characteristics as calibration references, avoiding the situation where incorrect calibration parameters are determined due to using clouds with uneven internal texture and thickness as calibration references.

[0050] In another embodiment of the present invention, when the calibration reference data source is a desert data source, it is necessary to determine the calibration reference image region based on the desert data source. Based on this, step 203 specifically includes: when the calibration reference data source is the desert data source, the projection area of ​​the preset uniform desert region in the push-broom image is taken as the calibration reference image region.

[0051] Specifically, since the satellite-borne optical pushbroom image to be calibrated (image-side coordinates) and the preset uniform desert region (object-side geographic coordinates) are in different coordinate systems, they are first transformed into a unified spatial reference coordinate system. For example, using satellite orbital parameters, attitude data, and digital elevation models, the pushbroom image is precisely projected into the geographic coordinate system through geometric models such as collinearity equations, or the uniform desert region is projected into the pixel coordinate system of the pushbroom image. Then, based on the consistency of the two coordinate systems, a spatial overlay operation is performed on the ground coverage area of ​​the pushbroom image and the boundary of the uniform desert region. By performing an intersection operation, the spatially overlapping portion is extracted as the desert overlap region. Finally, the desert overlap region is back-mapped into the original pushbroom image to obtain the mapped region as the calibration reference image region. Then, relative radiometric correction of the pushbroom image is performed using the calibration reference image region. Since the radiometric characteristics of the desert region are stable, using the desert region for radiometric correction can improve the accuracy of the radiometric correction.

[0052] Furthermore, after determining the calibration reference image region, it is necessary to determine the relative radiometric calibration parameters based on this calibration reference image region. Therefore, the method includes: determining the average pixel grayscale value detected by all detectors in the calibration reference image region. ,in, The average pixel grayscale value common to all probes, The total number of detector elements in the calibration reference image area, For the first The average pixel grayscale value detected by each detector in the calibration reference image region; the current relative radiometric correction gain of each detector is determined based on the average pixel grayscale value: ,in, For the first The current relative radiometric correction gain of each detector element; based on the current relative radiometric correction gain, the current relative fused radiometric correction gain of each detector element is determined: ,in, For the first The current relative fusion radiation correction gain of each detector, For smoothing coefficient, For the first The historical relative radiometric correction gain of each detector element; the current relative fused radiometric correction bias of each detector element is determined based on the current relative radiometric correction bias: ,in, For the first The current relative fusion radiometric correction bias of each probe, For the first The current relative radiometric correction bias of each detector element is set to zero. For the first The historical relative radiometric correction bias of each detector element is used; the current relative radiometric calibration parameters of the corresponding detector element are determined based on the current relative fused radiometric correction gain and the current relative fused radiometric correction bias of each detector element. Specifically, determining the current relative radiometric calibration parameters of the corresponding detector element based on the current relative fused radiometric correction gain and the current relative fused radiometric correction bias of each detector element includes: if the calibration reference image region is a cloud region, then the historical relative fused radiometric correction gain determined most recently based on a uniform desert projection region is obtained; if the gain difference between the current relative fused radiometric correction gain and the historical relative fused radiometric correction gain exceeds a preset gain threshold, then the historical relative radiometric calibration parameters determined most recently based on a uniform desert projection region are used as the current relative radiometric calibration parameters of the corresponding detector element; if the gain difference does not exceed the preset gain threshold, then the current relative fused radiometric correction gain and the current relative fused radiometric correction bias of each detector element are used as the current relative radiometric calibration parameters of the corresponding detector element.

[0053] The preset gain threshold is a value set according to actual needs. Specifically, each detector in the calibration reference image area has its detected pixel grayscale value. The average pixel grayscale value is obtained by averaging the pixel grayscale values ​​detected by all detectors. At the same time, for each detector, the average of the pixel grayscale values ​​detected by a certain detector in the calibration reference image area is calculated as the pixel grayscale average of that detector. Then, the current relative radiometric correction gain of each detector is determined based on the average pixel grayscale value and the pixel grayscale average. The current relative radiometric correction gain is then jointly analyzed with the historical relative radiometric correction gain of that detector stored in the on-board database. Based on the joint analysis result, the current relative fused radiometric correction gain of the corresponding detector is determined. The current relative radiometric correction bias is set to 0. Based on the current relative radiometric correction bias and the historical relative radiometric correction bias of that detector stored in the on-board database, a joint analysis is performed. Based on the joint analysis result, the current relative fused radiometric correction bias of that detector is determined. Next, the gain difference between the current relative fused radiometric correction gain and the most recent historical relative fused radiometric correction gain is determined. If the difference is too large, the current relative fused radiometric correction gain and bias are discarded, and the most recent historical relative fused radiometric correction gain and bias are used as the current relative radiometric calibration parameters. If the difference meets the requirements, the current relative fused radiometric correction gain and bias are used as the current relative radiometric calibration parameters.

[0054] In another embodiment of the present invention, it can be further determined whether the current calibration reference image region is a cloud region. If so, the historical relative fused radiometric correction gain determined most recently based on the desert projection region is obtained. For the same detector, it is determined whether the difference between the current relative fused radiometric correction gain and the historical relative fused radiometric correction gain is too large. If it is too large, the historical relative fused radiometric correction gain and offset are used as the current relative radiometric calibration parameters; otherwise, the current relative fused radiometric correction gain and the current relative fused radiometric correction offset are used as the current relative radiometric calibration parameters for the corresponding detector. Since pushbroom sensors (such as pushbroom cameras) are affected by the complex space environment during on-orbit operation, their response characteristics may change abruptly. By calculating the difference between the current gain and the historical reference gain, it is possible to capture drastic fluctuations in parameters caused by abnormal scenes or calculation errors. When the difference is too large, the current parameters are automatically discarded and the system reverts to historical reliable parameters, effectively avoiding the introduction of new stripe noise or radiometric distortion by erroneous calibration parameters and ensuring the accuracy of image radiometric correction. Furthermore, the retained current relative radiometric calibration parameters are stored in the on-board database for use as the basis for determining subsequent calibration parameters.

[0055] Finally, relative radiometric correction is performed on the pushbroom image based on the current relative radiometric calibration parameters. In another embodiment of the invention, if neither the cloud data source nor the desert data source exists, relative radiometric correction can be performed on the image based on historical calibration parameters stored on the satellite.

[0056] 204. Based on the geographical range of the preset pseudo-invariant calibration field, locate the projection area of ​​the preset pseudo-invariant calibration field in the pushbroom image after relative radiometric correction to obtain the calibration image window area. Obtain the ideal pixel gray values ​​of each band in the calibration image window area corresponding to the preset pseudo-invariant calibration field. Based on the ideal pixel gray values ​​and the actual pixel gray values ​​of each band in the calibration image window area of ​​the pushbroom image, determine the current absolute radiometric calibration parameters of the corresponding band and perform absolute radiometric correction on the pushbroom image after relative radiometric correction.

[0057] Specifically, the overlapping region of the calibration field is projected onto the pushbroom image to obtain the calibration image-side window region. Then, the ideal pixel grayscale values ​​of different bands within the calibration image-side window region need to be calculated. Based on this, the method includes: determining the band pixel reference information for each band of the preset pseudo-invariant calibration field, wherein the band pixel reference information includes reference apparent reflectance, solar zenith angle, and Earth-Sun distance factor; and determining the ideal pixel grayscale values ​​for each band of the calibration image-side window region based on the band pixel reference information. ,in, For band Ideal pixel grayscale value For band Reference apparent reflectance, For the solar zenith angle, Earth-Sun distance factor For band The radiation calibration conversion factor.

[0058] The ideal pixel grayscale value refers to the pixel grayscale value that an image captured under illumination conditions such as reference apparent reflectance, solar zenith angle, and Earth-Sun distance factor should possess. Specifically, information such as reference apparent reflectance, solar zenith angle, and Earth-Sun distance factor for different bands can be obtained from an on-board database. Based on this information, the ideal pixel grayscale value of the corresponding band image within the calibration image window area is determined. Further, after determining the ideal pixel grayscale value, the actual pixel grayscale value of the image within the calibration image window area is determined. Then, the current absolute radiometric calibration parameters for the corresponding band need to be determined based on the ideal and actual pixel grayscale values. Therefore, the method includes: determining the current absolute radiometric correction gain for the band. ,in, For band The current absolute radiation correction gain, For band The average actual pixel grayscale value in the calibration image window area, For band The historical absolute radiometric correction bias; determine the current absolute radiometric correction bias for the band: ,in, For band The current absolute radiometric correction bias; determine the current absolute fused radiometric correction gain for the band: ,in, For band The current absolute fusion radiation correction gain, For the moving average smoothing coefficient, For band Historical absolute radiometric correction gain; determine the current absolute fused radiometric correction bias for the band: ,in, For band The current absolute fusion radiative correction bias, For band The historical absolute radiometric correction bias; based on the current absolute fused radiometric correction gain and the current absolute fused radiometric correction bias, the current absolute radiometric calibration parameters for the corresponding band are determined.

[0059] Specifically, historical absolute radiometric correction offsets for images of different bands acquired by the same pushbroom camera are obtained from the onboard database. Then, based on the ideal pixel grayscale values, actual pixel grayscale values, and historical absolute radiometric correction offsets for the same band image, the current absolute radiometric correction gain and current absolute radiometric correction offset for the corresponding band are determined. Next, the current absolute fused radiometric correction gain and current absolute fused radiometric correction offset are determined through joint analysis of the historical absolute radiometric correction gain (historical absolute fused radiometric correction gain) and historical absolute radiometric correction offset (historical absolute fused radiometric correction offset). Then, for the same band, if the difference between the current absolute fused radiometric correction gain and the historical absolute fused radiometric correction gain is too large, the historical absolute fused radiometric correction gain and historical absolute fused radiometric correction offset are used as the current absolute radiometric calibration parameters; if the difference meets the requirements, the current absolute fused radiometric correction gain and current absolute fused radiometric correction offset are used as the current absolute radiometric calibration parameters. Finally, the determined current absolute radiometric calibration parameters are stored in the onboard database for use in subsequent absolute radiometric corrections. The on-board database supports time series trend analysis of calibration parameters, tracing of abnormal calibration events, and cross-validation of pseudo-invariant calibration fields.

[0060] This invention, by calculating the difference between the current gain and the historical gain, can capture drastic fluctuations in parameters caused by abnormal scenes or calculation errors. When the difference is too large, the current calibration parameters are automatically discarded and the system reverts to historical reliable calibration parameters. This effectively avoids the introduction of new strip noise or radiation distortion by incorrect calibration parameters, and ensures the accuracy of absolute radiometric correction of the image.

[0061] In another embodiment of the present invention, if the overlap area between the ground coverage area and the geographical range of the calibration field of the pseudo-invariant calibration field stored in the preset pseudo-invariant calibration field database is too small or there is no overlap, then the image can be absolutely radiometrically corrected based on the historical absolute radiometric calibration parameters stored on the satellite.

[0062] 205. Identify the location of key areas in the ground coverage area within the pushbroom image after absolute radiometric correction as the registration window area. Within the registration window area, perform geometric correction on the pushbroom image after absolute radiometric correction based on the feature point matching relationship between the key areas and the pushbroom image.

[0063] In this embodiment of the invention, it is first determined whether the ground coverage area of ​​the push-broom image contains a preset key area. If so, the corresponding area of ​​the key area in the push-broom image is extracted and set as the registration window area. Within the defined registration window area, an image matching algorithm is used to extract and determine the same points (i.e., feature point pairs) that have the same physical features between the key area and the push-broom image. A feature point pair is a combination of pixel coordinates that represents the same physical location and has significant features in the two images, such as the center of an intersection of the same overpass, the corner of the roof of the same landmark building, or a clear bend in the same river. (Example: Key area base map coordinates) Image coordinates within the registration window area This constitutes a set of feature point pairs. Based on the coordinates of the extracted multiple sets of feature point pairs, a mathematical model (such as a polynomial transformation model) is used to calculate the current geometric transformation parameters (such as translation, rotation, scaling, etc.) between the pushbroom image and the key area. Finally, the calculated current geometric transformation parameters are used to resample the spatial coordinates and perform geometric correction on the pushbroom image to eliminate local geometric distortions. Furthermore, the determined current geometric transformation parameters can be stored in the on-board database for subsequent geometric correction. This embodiment of the invention can also update the base map database of the key area based on the pushbroom image. For example, the average reprojection error of all correctly matched point pairs between the corrected image and the base map is calculated. If the error meets the requirements, it indicates that the geometric positioning accuracy of the current corrected image has reached or exceeded the accuracy of the base map, meeting the replacement conditions. The current corrected image is used to replace the corresponding base map in the base map database, and the geographic coordinate mapping table is updated simultaneously. A backup of the old base map is retained during replacement. If subsequent calibration reveals an anomaly in the new base map, a rollback can be performed. If the average reprojection error does not meet the threshold, the old base map is retained, and the current matching result is marked as unsatisfactory, without triggering replacement. This self-updating mechanism enables the base map library to gradually optimize itself as the satellite operates in orbit, continuously improving the timeliness of the base map, and eliminating the need for ground re-uploading, thus achieving fully autonomous maintenance of the on-board base map.

[0064] Furthermore, if there is no key area in the ground coverage area, the geometric correction method for the image includes: if there is no key area in the ground coverage area, then select multiple virtual control points in the push-broom image after absolute radiometric correction based on digital elevation information, determine the current pixel coordinate geometric correction coefficient based on the pixel coordinates and ground physical coordinates of each virtual control point, and perform geometric correction on the push-broom image after absolute radiometric correction based on the current pixel coordinate geometric correction coefficient.

[0065] Specifically, if there are no key areas in the ground coverage area, or if the number of feature point pairs is less than or equal to a preset threshold, then virtual control points are needed for geometric correction of the image. In the process of using virtual control points for geometric correction: firstly, virtual control points are selected in the pushbroom image based on the digital elevation model of the ground coverage area. The pixel coordinates and ground physical coordinates of each virtual control point are then determined. Pixel coordinates can be directly determined from the pushbroom image, while ground physical coordinates can be determined based on pixel coordinates and digital elevation model information, such as setting initial elevation values. The new elevation value is obtained by iteratively solving the collinearity equation and then looking it up in the digital elevation model. Iterate to (ε is the convergence threshold, such as 1m), and the final geographic latitude and longitude are obtained to determine the ground physical coordinates. After assignment, each virtual control point has pixel coordinates, ground physical coordinates, and elevation information. Based on this information, the geometric correction coefficient of each pixel coordinate in the image can be determined, and finally, the geometric correction of each pixel coordinate in the image is performed based on the geometric correction coefficient.

[0066] Furthermore, in order to perform geometric correction on the image based on virtual control points, virtual control points need to be selected in the pushbroom image first. Based on this, step 205 specifically includes: determining the elevation range based on the digital elevation information of the ground coverage area; determining the number of elevation layers in the ground coverage area based on the elevation range; performing elevation layering on the ground coverage area based on the number of elevation layers to obtain multiple elevation layers; dividing the absolute radiometrically corrected pushbroom image into multiple grids; for each grid, determining the elevation coverage area belonging to any elevation layer in the grid; determining multiple candidate virtual control points in the elevation coverage area; and selecting multiple virtual control points from the multiple candidate virtual control points corresponding to each grid. The method for selecting multiple virtual control points from candidate virtual control points includes: for each candidate virtual control point, determining a rectangular window centered on the ground projection position of the candidate virtual control point; determining the ground slope angle within the rectangular window; determining the slope factor of the candidate virtual control point based on the ground slope angle; determining the elevation standard deviation of multiple elevation values ​​within the rectangular window; determining the elevation difference anomaly factor of the candidate virtual control point based on the elevation standard deviation; determining the physical coordinates of the ground projection position of the candidate virtual control point and its image-side coordinates in the corresponding grid; and determining the slope of the ground slope where the ground projection position is located based on the line-of-sight vector from the image-side coordinates to the physical coordinates. The surface normal vector is used to determine the vector angle between the line-of-sight vector and the slope normal vector. Based on the vector angle, the line-of-sight angle factor of the candidate virtual control point is determined. The distance between the candidate virtual control point and the image center of the pushbroom image after absolute radiometric correction is determined. The diagonal length of the pushbroom image after absolute radiometric correction is determined. Based on the ratio of the distance to the diagonal length, the image-side position factor of the candidate virtual control point is determined. Based on the slope factor, the height difference anomaly factor, the line-of-sight angle factor, and the image-side position factor, the evaluation score of the candidate virtual control point is determined. Based on the evaluation score, multiple virtual control points are selected from each candidate virtual control point.

[0067] Specifically, using pre-stored global digital elevation model data on the satellite, the elevation distribution of the ground coverage area is statistically analyzed to determine the elevation range as shown below:

[0068] in, and These represent the maximum and minimum elevation values ​​within the ground coverage area of ​​the push-broom image, respectively. This represents the elevation range. The number of elevation layers is then determined based on the elevation range as follows:

[0069] in, For the number of elevation layers, The function is rounded up, with each layer divided into 500m intervals, a minimum of 3 layers and a maximum of 10 layers. This range is determined empirically because too few layers cannot effectively distinguish elevation differences, while too many layers result in too few candidate points per layer. The ground coverage area is then divided into multiple elevation layers according to the number of elevation layers. Simultaneously, the pushbroom image is divided into a regular grid of M rows × N columns. The number of grids is determined as follows: each grid covers approximately 1000 × 1000 pixels, resulting in a total of 100 to 300 virtual control points for the entire pushbroom image; the ratio of M to N is close to the aspect ratio of the image, making the grid nearly square; and both M and N are not less than 8 to ensure sufficient coverage. For example, for an 8000 × 8000 pixel pushbroom image, M = N = 8, meaning each grid covers approximately 1000 × 1000 pixels, totaling 64 grids, with at least one virtual control point selected for each grid. Furthermore, each grid is labeled with the elevation layer it covers according to the data elevation model information. A grid may contain multiple elevation layers, especially in areas with large topographic relief. Within each grid-elevation layer unit, that is, within a certain grid, the area belonging to a certain elevation layer is sampled, that is, candidate virtual control points are sampled. An equal interval sampling method can be used, such as 3×3 or 5×5 sampling points. Each sampling point is a candidate virtual control point, and each candidate virtual control point records its image coordinates and corresponding elevation.

[0070] Furthermore, after determining multiple candidate virtual control points, for each candidate virtual control point, a rectangular window is drawn centered on its ground projection position. The ground slope angle of the surface where the rectangular window is located is determined, and then the slope factor is determined according to the following formula:

[0071] in, For ground slope angle, Candidate virtual control points The slope factor, if If the slope factor is 0, then the slope factor is 0.

[0072] At the same time, determine the standard deviation of the elevation values ​​used within the rectangular window, and then determine the elevation difference anomaly factor according to the following formula:

[0073] in, Candidate virtual control points High differential factors Candidate virtual control points The standard deviation of the elevation values ​​used within the corresponding rectangular window. The elevation standard deviation threshold is set according to actual needs, such as It can be set to 30m.

[0074] Simultaneously, the ground physical coordinates of the candidate virtual control points and their image-side coordinates (pixel coordinates) in the push-broom image are determined. Based on these two sets of coordinates, a line-of-sight vector from the image-side to the object-side is constructed. Furthermore, the slope normal vector of the slope where the virtual control point's ground projection location is located is determined by combining local terrain data. The angle between the line-of-sight vector and the slope normal vector is calculated through vector operations, and this angle is used as the virtual control point. Line of sight - slope angle factor .

[0075] Simultaneously, the Euclidean distance between the candidate virtual control points and the center of the pushbroom image is determined, and the diagonal length of the pushbroom image is also determined. Then, the image-side location factor is determined using the following formula:

[0076] in, Candidate virtual control points Image location factor, Candidate virtual control points Compared with push-broom images Central European distance, For push-broom images The diagonal length can be taken as the maximum diagonal length.

[0077] Furthermore, based on actual needs, weight coefficients are determined for the slope factor, height difference anomaly factor, line-of-sight angle factor, and image-side location factor. These weight coefficients are then used to weight and sum the slope factor, height difference anomaly factor, line-of-sight angle factor, and image-side location factor to obtain the evaluation scores for the corresponding candidate virtual control points. Finally, virtual control points are selected based on these evaluation scores. During the selection process, the proportion of candidate virtual control points in the total number of candidate virtual control points for each elevation level is statistically analyzed. For the elevation level with the lowest proportion, the top n candidate points with the highest evaluation scores within that elevation level are added to the virtual control point set. This ensures that each elevation level has at least a preset number of control points (the preset number is set according to actual needs). This process prioritizes control point coverage in the elevation direction. Then, all M×N image-side grids are traversed. If a grid has no selected virtual control points, the control point with the highest evaluation score from all candidate virtual control points in that grid is added to the virtual control point set. This process ensures a uniform distribution of virtual control points in the image-side plane, meaning each grid has at least one control point. Next, a spatial redundancy check is performed on the virtual control point set. If two virtual control points simultaneously meet the following two conditions, they are merged into the one with the higher evaluation score: the planar distance between the two virtual control points is less than 0.5 times the corresponding grid side length, and the elevation difference between the two virtual control points is less than 0.3 times the corresponding elevation layer thickness. This process eliminates excessively dense virtual control points in space, avoids local overfitting, and reduces the computational load of subsequent geometric correction. This embodiment of the invention selects virtual control points by comprehensively analyzing slope factors, elevation anomaly factors, line-of-sight-slope angle factors, and image-side position factors. This allows for the selection of optimal control points in complex mountainous terrain that have gentle slopes, consistent elevations, lines of sight orthogonal to the slope, and are located within the effective image area, thereby significantly improving the correction accuracy of spaceborne optical pushbroom images.

[0078] Therefore, multiple virtual control points can be determined in the above manner. Then, using the image-space coordinates (pixel coordinates in the push-broom image) and the corresponding object-space coordinates (ground physical coordinates) of all virtual control points, an image-space correction polynomial is constructed. When constructing the polynomial, the image-space coordinates of the virtual control points ( row, col ) is the independent variable, and the ideal image coordinates corresponding to the object coordinates (E, N) are the inverse variables. row ideal , col ideal With as the dependent variable, the resulting polynomial is shown below:

[0079]

[0080] Substituting the pixel coordinates of multiple virtual control points and the ideal image-square coordinates (ideal pixel coordinates) corresponding to the ground physical coordinates into the above formula, the polynomial coefficients are obtained by least squares method. The polynomial coefficients are as follows: , , , , , , , , , , and Calculate the fitting residual for each virtual control point. If the residual of a control point exceeds three times the mean of the residuals of all control points, it is marked as an outlier and removed. The polynomial is then refitted. This process is repeated until there are no outliers or the number of control points still meets the minimum requirement. Finally, the pixel coordinates of each pixel in the pushbroom image are substituted into the image-side correction polynomial after solving for the polynomial coefficients to solve for the corrected pixel coordinates, thus obtaining the geometrically corrected pushbroom image.

[0081] It should be noted that the aforementioned radiometric and geometric correction processes can be efficiently implemented on a spaceborne heterogeneous parallel computing platform, enabling independent parallel correction for each detector and each waveband. During the correction process, relative radiometric correction, absolute radiometric correction, and geometric correction are performed sequentially; that is, relative radiometric correction is performed first, followed by absolute radiometric correction, and finally geometric correction. The final output is a standardized image with consistent radiometrics and accurate geometric positioning, directly serving subsequent intelligent applications such as onboard target detection, scene classification, and large-scale model inference.

[0082] According to another method for on-orbit correction of spaceborne optical pushbroom images provided by the present invention, compared with the current method of correcting pushbroom images by means of hardware such as solar diffusers and integrating spheres on the satellite, the present invention corrects pushbroom images by means of existing reference image data sources, preset pseudo-invariant calibration fields, key areas, etc. on the satellite. That is, it only uses the imaging data stored on the satellite to correct the pushbroom images, without relying on calibration hardware equipment, thereby saving image correction resources and improving the correction efficiency and accuracy of pushbroom images. At the same time, the present invention can directly correct pushbroom images at the satellite end, avoiding the time and bandwidth consumption of transmitting pushbroom images to the ground for correction. Specifically, this invention eliminates image stripes or non-uniformity caused by inconsistent responses of different detectors in spaceborne optical imaging devices by obtaining a calibration reference data source, determining relative radiometric calibration parameters for each detector element based on this data source, and performing correction. This ensures the relative radiometric consistency within the image. By utilizing the deviation between the known ideal pixel gray values ​​and the actual pixel gray values ​​in the image from each band of the pseudo-invariant calibration field, the current absolute radiometric calibration parameters are solved band by band. Utilizing the long-term stable physical property of the pseudo-invariant field's spectral characteristics, a traceable true radiometric reference is provided for the image gray values. Using the projection area of ​​the pseudo-invariant calibration field as the calibration image window, the correction range is precisely constrained within the effective area, ensuring calibration accuracy and improving the precision of absolute radiometric correction. By using the feature point matching relationship between the key area and the pushbroom image to solve the geometric transformation relationship, sub-pixel-level high-precision registration can be achieved, fully utilizing the accuracy advantages of external reference data.

[0083] Furthermore, as Figure 1 In specific implementation, embodiments of the present invention provide an on-orbit correction device for spaceborne optical pushbroom images, such as... Figure 3 As shown, the device includes: an acquisition unit 31, a relative radiation correction unit 32, an absolute radiation correction unit 33, and a geometric correction unit 34.

[0084] The acquisition unit 31 can be used to acquire pushbroom images of the ground coverage area captured by the spaceborne optical imaging device, which are to be corrected.

[0085] The relative radiometric correction unit 32 can be used to identify the calibration reference data source in the ground coverage area, determine the calibration reference image area in the pushbroom image based on the calibration reference data source, determine the current relative radiometric calibration parameter of the corresponding probe based on the pixel gray value detected by each probe in the calibration reference image area, and perform relative radiometric correction on the pushbroom image element by element based on the current relative radiometric calibration parameter.

[0086] The absolute radiometric correction unit 33 can be used to locate the projection area of ​​the preset pseudo-invariant calibration field in the pushbroom image after relative radiometric correction based on the geographical range of the preset pseudo-invariant calibration field, so as to obtain the calibration image window area, obtain the ideal pixel gray value of each band of the calibration image window area corresponding to the preset pseudo-invariant calibration field, and determine the current absolute radiometric calibration parameters of the corresponding band based on the ideal pixel gray value and the actual pixel gray value of each band of the calibration image window area in the pushbroom image, and perform absolute radiometric correction on the pushbroom image after relative radiometric correction.

[0087] The geometric correction unit 34 can be used to identify the position of the key area in the pushbroom image after absolute radiometric correction in the ground coverage area as a registration window area. Within the registration window area, geometric correction is performed on the pushbroom image after absolute radiometric correction based on the feature point matching relationship between the key area and the pushbroom image after absolute radiometric correction.

[0088] In specific application scenarios, in order to determine the calibration reference data source, the relative radiation correction unit 32 can be used to determine the overlapping area between the ground coverage area and the preset uniform desert area. If the area of ​​the overlapping area is less than the preset area threshold, the calibration reference data source is determined to be a cloud data source. If the area of ​​the overlapping area is greater than or equal to the preset area threshold, the calibration reference data source is determined to be a desert data source.

[0089] In specific application scenarios, in order to determine the calibration reference image region, such as Figure 4 As shown, the relative radiation correction unit 32 includes: a region division module 321, a region selection module 322, and a region determination module 323.

[0090] The region division module 321 can be used to determine the historical relative radiometric calibration parameters of each probe when the calibration reference data source is the cloud data source.

[0091] The region selection module 322 can be used to calculate the mean pixel gray level and standard deviation of each probe in the cloud region of the pushbroom image row by row based on the historical relative radiometric calibration parameters, determine the intra-row pixel gray level variation amplitude of the corresponding row pixels based on the pixel gray level standard deviation, determine the inter-row pixel gray level variation amplitude of adjacent row pixels based on the pixel gray level mean, and select the cloud region where the intra-row pixel gray level variation amplitude and inter-row pixel gray level variation amplitude of a consecutive preset number of rows are both less than the corresponding preset amplitude threshold as the calibration reference image region.

[0092] The region determination module 323 can be used to, when the calibration reference data source is the desert data source, use the projection area of ​​the preset uniform desert region in the push-broom image as the calibration reference image region.

[0093] In specific application scenarios, in order to determine the current relative radiometric calibration parameters of the corresponding probe, the relative radiometric correction unit 32 also includes a relative parameter determination module 324.

[0094] The relative parameter determination module 324 can be used to determine the average pixel grayscale value detected by all detectors in the calibration reference image region: ,in, The average pixel grayscale value common to all probes, The total number of detector elements in the calibration reference image area, For the first The average pixel grayscale value detected by each detector in the calibration reference image region; the current relative radiometric correction gain of each detector is determined based on the average pixel grayscale value: ,in, For the first The current relative radiometric correction gain of each detector element; based on the current relative radiometric correction gain, the current relative fused radiometric correction gain of each detector element is determined: ,in, For the first The current relative fusion radiation correction gain of each detector, For smoothing coefficient, For the first The historical relative radiometric correction gain of each detector element; the current relative fused radiometric correction bias of each detector element is determined based on the current relative radiometric correction bias: ,in, For the first The current relative fusion radiometric correction bias of each probe, For the first The current relative radiometric correction bias of each detector element is set to zero. For the first The historical relative radiometric correction bias of each detector element; the current relative radiometric calibration parameters of the corresponding detector element are determined based on the current relative fused radiometric correction gain and the current relative fused radiometric correction bias of each detector element.

[0095] In specific application scenarios, in order to determine the current relative radiometric calibration parameters, the relative parameter determination module 324 can be used to obtain the historical relative fused radiometric correction gain determined most recently based on the projection area of ​​a uniform desert if the calibration reference image area is a cloud area; if the gain difference between the current relative fused radiometric correction gain and the historical relative fused radiometric correction gain exceeds a preset gain threshold, then the historical relative radiometric calibration parameter determined most recently based on the projection area of ​​a uniform desert is used as the current relative radiometric calibration parameter of the corresponding probe; if the gain difference does not exceed the preset gain threshold, then the current relative fused radiometric correction gain and the current relative fused radiometric correction offset of each probe are used as the current relative radiometric calibration parameters of the corresponding probe.

[0096] In specific application scenarios, in order to perform geometric correction on the image based on virtual control points, the geometric correction unit 34 can also be used to select multiple virtual control points in the push-broom image after absolute radiometric correction based on digital elevation information if the key area does not exist in the ground coverage area, determine the current pixel coordinate geometric correction coefficient based on the pixel coordinates and ground physical coordinates of each virtual control point, and perform geometric correction on the push-broom image after absolute radiometric correction based on the current pixel coordinate geometric correction coefficient.

[0097] In specific application scenarios, in order to select multiple virtual control points, the geometric correction unit 34 includes an elevation layer determination module 341, a grid division module 342, and a control point selection module 343.

[0098] The elevation layer determination module 341 can be used to determine the elevation range based on the digital elevation information of the ground coverage area, determine the number of elevation layers in the ground coverage area based on the elevation range, and perform elevation layering on the ground coverage area based on the number of elevation layers to obtain multiple elevation layers.

[0099] The grid division module 342 can be used to divide the absolute radiometrically corrected pushbroom image into multiple grids. For each grid, an elevation coverage area belonging to an arbitrary elevation layer is determined in the grid, and multiple candidate virtual control points are determined in the elevation coverage area.

[0100] The control point selection module 343 can be used to select multiple virtual control points from multiple candidate virtual control points corresponding to each grid.

[0101] In specific application scenarios, to select multiple virtual control points, the control point selection module 343 can be used to: determine a rectangular window centered on the ground projection position of each candidate virtual control point; determine the ground slope angle within the rectangular window; determine the slope factor of the candidate virtual control point based on the ground slope angle; determine the elevation standard deviation of multiple elevation values ​​within the rectangular window; determine the elevation difference anomaly factor of the candidate virtual control point based on the elevation standard deviation; determine the physical coordinates of the ground projection position of the candidate virtual control point and its image-side coordinates in the corresponding grid; and determine the ground projection position's position based on the line-of-sight vector from the image-side coordinates to the physical coordinates. Based on the slope normal vector of the ground slope, determine the vector angle between the line-of-sight vector and the slope normal vector, and determine the line-of-sight angle factor of the candidate virtual control point based on the vector angle; determine the distance between the candidate virtual control point and the image center of the pushbroom image after absolute radiometric correction, determine the diagonal length of the pushbroom image after absolute radiometric correction, and determine the image-side position factor of the candidate virtual control point based on the ratio of the distance to the diagonal length; determine the evaluation score of the candidate virtual control point based on the slope factor, the height difference anomaly factor, the line-of-sight angle factor, and the image-side position factor, and select multiple virtual control points for each candidate virtual control point based on the evaluation score.

[0102] In specific application scenarios, to determine the ideal pixel grayscale value, the absolute radiometric correction unit 33 can be used to determine the band pixel reference information of each band of the preset pseudo-invariant calibration field, wherein the band pixel reference information includes reference apparent reflectance, solar zenith angle, and Earth-Sun distance factor; and based on the band pixel reference information, determine the ideal pixel grayscale value of each band in the calibration image window region. ,in, For band Ideal pixel grayscale value For band Reference apparent reflectance, For the solar zenith angle, Earth-Sun distance factor For band The radiation calibration conversion factor.

[0103] In specific application scenarios, in order to determine the current absolute radiometric calibration parameters, the absolute radiometric correction unit 33 can be used to determine the current absolute radiometric correction gain of the band: ,in, For band The current absolute radiation correction gain, For band The average actual pixel grayscale value in the calibration image window area, For band The historical absolute radiometric correction bias; determine the current absolute radiometric correction bias for the band: ,in, For band The current absolute radiometric correction bias; determine the current absolute fused radiometric correction gain for the band: ,in, For band The current absolute fusion radiation correction gain, For the moving average smoothing coefficient, For band Historical absolute radiometric correction gain; determine the current absolute fused radiometric correction bias for the band: ,in, For band The current absolute fusion radiative correction bias, For band The historical absolute radiometric correction bias; based on the current absolute fused radiometric correction gain and the current absolute fused radiometric correction bias, the current absolute radiometric calibration parameters for the corresponding band are determined.

[0104] It should be noted that other corresponding descriptions of the functional modules involved in the on-orbit correction device for spaceborne optical pushbroom images provided in this embodiment of the invention can be found in [reference needed]. Figure 1 The corresponding description of the method shown will not be repeated here.

[0105] Based on the above, Figure 1The method shown in the invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, performs the following steps: acquiring a pushbroom image of a ground-covered area to be calibrated obtained by a spaceborne optical imaging device; identifying a calibration reference data source in the ground-covered area; determining a calibration reference image region in the pushbroom image based on the calibration reference data source; determining the current relative radiometric calibration parameter of the corresponding probe based on the pixel grayscale value detected by each probe in the calibration reference image region; performing relative radiometric calibration on the pushbroom image pixel by pixel based on the current relative radiometric calibration parameter; and locating the preset pseudo-invariant calibration field in the relative radiometric calibration area based on the geographical range of the preset pseudo-invariant calibration field. The projection area in the subsequent pushbroom image is used to obtain the calibration image window area. The ideal pixel gray values ​​of each band in the calibration image window area corresponding to the preset pseudo-invariant calibration field are obtained. Based on the ideal pixel gray values ​​and the actual pixel gray values ​​of each band in the calibration image window area in the pushbroom image, the current absolute radiometric calibration parameters of the corresponding band are determined, and the absolute radiometric correction of the pushbroom image after relative radiometric correction is performed. The position of the key area in the ground coverage area within the absolute radiometric correction pushbroom image is identified as the registration window area. Within the registration window area, the geometric correction of the absolute radiometric correction pushbroom image is performed according to the feature point matching relationship between the key area and the absolute radiometric correction pushbroom image.

[0106] Based on the above, Figure 1 The method shown and as Figure 3 The embodiment of the device shown in the invention also provides a physical structure diagram of a computer device, such as... Figure 5As shown, the computer device includes: a processor 41, a memory 42, and a computer program stored in the memory 42 and executable on the processor. Both the memory 42 and the processor 41 are mounted on a bus 43. When the processor 41 executes the program, it performs the following steps: acquiring a pushbroom image of the ground coverage area to be calibrated obtained by a spaceborne optical imaging device; identifying a calibration reference data source in the ground coverage area; determining a calibration reference image region in the pushbroom image based on the calibration reference data source; determining the current relative radiometric calibration parameter of the corresponding probe based on the pixel grayscale value detected by each probe in the calibration reference image region; performing relative radiometric correction on the pushbroom image element by element based on the current relative radiometric calibration parameter; and locating the target location based on the geographical range of a preset pseudo-invariant calibration field. The projection area of ​​the preset pseudo-invariant calibration field in the pushbroom image after relative radiometric correction is used to obtain the calibration image window area. The ideal pixel gray values ​​of each band in the calibration image window area corresponding to the preset pseudo-invariant calibration field are obtained. Based on the ideal pixel gray values ​​and the actual pixel gray values ​​of each band in the calibration image window area in the pushbroom image, the current absolute radiometric calibration parameters of the corresponding band are determined, and absolute radiometric correction of the pushbroom image after relative radiometric correction is performed. The position of the key area in the ground coverage area in the pushbroom image after absolute radiometric correction is identified as the registration window area. Within the registration window area, the pushbroom image after absolute radiometric correction is geometrically corrected according to the feature point matching relationship between the key area and the pushbroom image after absolute radiometric correction.

[0107] Through the technical solution of this invention, the pushbroom image is corrected by using existing reference image data sources, preset pseudo-invariant calibration fields, key areas, etc. on the satellite. That is, the pushbroom image is corrected using only the imaging data stored on the satellite, without relying on calibration hardware equipment. This saves image correction resources and improves the correction efficiency and accuracy of the pushbroom image. At the same time, this invention can directly correct the pushbroom image at the satellite end, avoiding the time and bandwidth consumption of transmitting the pushbroom image to the ground for correction. Specifically, this invention eliminates image stripes or non-uniformity caused by inconsistent responses of different detectors in spaceborne optical imaging devices by obtaining a calibration reference data source, determining relative radiometric calibration parameters for each detector element based on this data source, and performing correction. This ensures the relative radiometric consistency within the image. By utilizing the deviation between the known ideal pixel gray values ​​and the actual pixel gray values ​​in the image from each band of the pseudo-invariant calibration field, the current absolute radiometric calibration parameters are solved band by band. Utilizing the long-term stable physical property of the pseudo-invariant field's spectral characteristics, a traceable true radiometric reference is provided for the image gray values. Using the projection area of ​​the pseudo-invariant calibration field as the calibration image window, the correction range is precisely constrained within the effective area, ensuring calibration accuracy and improving the precision of absolute radiometric correction. By using the feature point matching relationship between the key area and the pushbroom image to solve the geometric transformation relationship, sub-pixel-level high-precision registration can be achieved, fully utilizing the accuracy advantages of external reference data.

[0108] It is obvious to those skilled in the art that the modules or steps of the present invention described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. Optionally, they can be implemented using computer-executable program code, thereby storing them in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those presented herein, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0109] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for on-orbit correction of spaceborne optical pushbroom images, characterized in that, include: Acquire pushbroom images of the ground coverage area captured by the spaceborne optical imaging device, which are to be corrected. Identify the calibration reference data source in the ground coverage area, determine the calibration reference image area in the pushbroom image based on the calibration reference data source, determine the current relative radiometric calibration parameter of the corresponding probe based on the pixel gray value detected by each probe in the calibration reference image area, and perform relative radiometric correction on the pushbroom image element by element based on the current relative radiometric calibration parameter. Based on the geographical range of the preset pseudo-invariant calibration field, the projection area of ​​the preset pseudo-invariant calibration field in the pushbroom image after relative radiometric correction is located to obtain the calibration image window area. The ideal pixel gray values ​​of each band in the calibration image window area corresponding to the preset pseudo-invariant calibration field are obtained. Based on the ideal pixel gray values ​​and the actual pixel gray values ​​of each band in the calibration image window area in the pushbroom image, the current absolute radiometric calibration parameters of the corresponding band are determined, and the absolute radiometric correction of the pushbroom image after relative radiometric correction is performed. The location of a key area within the ground coverage region in the absolute radiometrically corrected pushbroom image is identified as a registration window region. Within the registration window region, geometric correction is performed on the absolute radiometrically corrected pushbroom image based on the feature point matching relationship between the key area and the absolute radiometrically corrected pushbroom image.

2. The on-orbit correction method for spaceborne optical pushbroom images according to claim 1, characterized in that, Identifying the calibration reference data source in the ground coverage area includes: The overlapping area between the ground coverage area and the preset uniform desert area is determined. If the area of ​​the overlapping area is less than the preset area threshold, the calibration reference data source is determined to be a cloud data source. If the area of ​​the overlapping area is greater than or equal to the preset area threshold, the calibration reference data source is determined to be a desert data source.

3. The on-orbit correction method for spaceborne optical pushbroom images according to claim 2, characterized in that, Determining a calibration reference image region in the pushbroom image based on the calibration reference data source includes: When the calibration reference data source is the cloud data source, the historical relative radiometric calibration parameters of each probe element are determined; Based on the historical relative radiometric calibration parameters, the mean pixel gray value and standard deviation of each probe element in the cloud region of the pushbroom image are calculated row by row. Based on the standard deviation of pixel gray value, the intra-row pixel gray value variation amplitude of the corresponding row pixels is determined. Based on the mean pixel gray value, the inter-row pixel gray value variation amplitude of adjacent rows pixels is determined. The cloud region in which the intra-row pixel gray value variation amplitude and the inter-row pixel gray value variation amplitude of a consecutive preset number of rows are both less than the corresponding preset amplitude threshold is taken as the calibration reference image region. When the calibration reference data source is the desert data source, the projection area of ​​the preset uniform desert region in the push-broom image is used as the calibration reference image region.

4. The on-orbit correction method for spaceborne optical pushbroom images according to claim 1, characterized in that, The current relative radiometric calibration parameters of the corresponding detector element are determined based on the pixel grayscale values ​​detected by each detector element in the calibration reference image region, including: Determine the average pixel grayscale value detected by all detectors in the calibration reference image region: ,in, The average pixel grayscale value common to all probes, The total number of detector elements in the calibration reference image area, For the first The average gray value of pixels detected by each probe in the calibration reference image region; The current relative radiometric correction gain for each detector element is determined based on the average pixel grayscale value: ,in, For the first The current relative radiation correction gain of each detector element; The current relative fused radiometric correction gain for each probe is determined based on the current relative radiometric correction gain: ,in, For the first The current relative fusion radiation correction gain of each detector, For smoothing coefficient, For the first Historical relative radiation correction gain of each detector element; The current relative convergent radiometric correction bias for each element is determined based on the current relative radiometric correction bias: ,in, For the first The current relative fusion radiometric correction bias of each probe, For the first The current relative radiometric correction bias of each detector element is set to zero. For the first Historical relative radiation correction bias of each detector element; The current relative radiometric calibration parameters of the corresponding detector element are determined based on the current relative fused radiometric correction gain and the current relative fused radiometric correction bias of each detector element. Specifically, the current relative radiometric calibration parameters for each detector element are determined based on its current relative fused radiometric correction gain and current relative fused radiometric correction bias, including: If the calibration reference image region is a cloud region, the historical relative fused radiometric correction gain determined most recently based on the projection region of a uniform desert is obtained. If the gain difference between the current relative fused radiometric correction gain and the historical relative fused radiometric correction gain exceeds a preset gain threshold, the historical relative radiometric calibration parameter determined most recently based on the projection region of a uniform desert is used as the current relative radiometric calibration parameter of the corresponding probe. If the gain difference does not exceed the preset gain threshold, the current relative fused radiometric correction gain and the current relative fused radiometric correction offset of each probe are used as the current relative radiometric calibration parameter of the corresponding probe.

5. The on-orbit correction method for spaceborne optical pushbroom images according to claim 1, characterized in that, The method further includes: If the key area does not exist in the ground coverage area, multiple virtual control points are selected in the push-broom image after absolute radiometric correction based on digital elevation information. The geometric correction coefficient of the current pixel coordinate is determined based on the pixel coordinate and ground physical coordinate of each virtual control point. The geometric correction of the push-broom image after absolute radiometric correction is performed based on the geometric correction coefficient of the current pixel coordinate.

6. The on-orbit correction method for spaceborne optical pushbroom images according to claim 5, characterized in that, Multiple virtual control points are selected in the pushbroom image after absolute radiometric correction based on digital elevation information, including: The elevation range is determined based on the digital elevation information of the ground coverage area, the number of elevation layers in the ground coverage area is determined based on the elevation range, and the ground coverage area is divided into multiple elevation layers based on the number of elevation layers. The absolute radiometrically corrected pushbroom image is divided into multiple grids. For each grid, an elevation coverage area belonging to an arbitrary elevation layer is determined in the grid, and multiple candidate virtual control points are determined in the elevation coverage area. Select multiple virtual control points from the multiple candidate virtual control points corresponding to each grid.

7. The on-orbit correction method for spaceborne optical pushbroom images according to claim 6, characterized in that, Select multiple virtual control points from the multiple candidate virtual control points corresponding to each grid, including: For each candidate virtual control point, a rectangular window is determined with the ground projection position of the candidate virtual control point as the center, the ground slope angle within the rectangular window is determined, and the slope factor of the candidate virtual control point is determined based on the ground slope angle. Determine the elevation standard deviation of multiple elevation values ​​within the rectangular window, and determine the elevation difference anomaly factor of the candidate virtual control point based on the elevation standard deviation; Determine the physical coordinates of the ground projection position of the candidate virtual control point and its image coordinates in the corresponding grid. Based on the line-of-sight vector pointing from the image coordinates to the physical coordinates, determine the slope normal vector of the ground slope where the ground projection position is located. Determine the vector angle between the line-of-sight vector and the slope normal vector. Based on the vector angle, determine the line-of-sight angle factor of the candidate virtual control point. The distance between the candidate virtual control point and the image center of the pushbroom image after absolute radiometric correction is determined, the diagonal length of the pushbroom image after absolute radiometric correction is determined, and the image-side position factor of the candidate virtual control point is determined based on the ratio of the distance to the diagonal length. The evaluation scores of the candidate virtual control points are determined based on the slope factor, the height difference anomaly factor, the line-of-sight-slope angle factor, and the image position factor. Multiple virtual control points are selected from each candidate virtual control point based on the evaluation scores.

8. The on-orbit correction method for spaceborne optical pushbroom images according to claim 1, characterized in that, Obtaining the ideal pixel grayscale values ​​of each band in the calibration image window region corresponding to the preset pseudo-invariant calibration field includes: Determine the band pixel reference information for each band of the preset pseudo-invariant calibration field, wherein the band pixel reference information includes reference apparent reflectance, solar zenith angle, and Earth-Sun distance factor; Based on the band pixel reference information, determine the ideal pixel grayscale value of each band in the calibration image window region: ,in, For band Ideal pixel grayscale value For band Reference apparent reflectance, For the solar zenith angle, Earth-Sun distance factor For band Radiative calibration conversion coefficient; The current absolute radiometric calibration parameters for the corresponding bands are determined based on the ideal pixel grayscale value and the actual pixel grayscale value of each band in the calibration image window region of the pushbroom image, including: Determine the current absolute radiation correction gain for the band: ,in, For band The current absolute radiation correction gain, For band The average actual pixel grayscale value in the calibration image window area, For band Historical absolute radiation correction bias; Determine the current absolute radiometric correction bias for the band: ,in, For band The current absolute radiation correction bias; Determine the current absolute fused radiative correction gain for the band: ,in, For band The current absolute fusion radiative correction gain, For the moving average smoothing coefficient, For band Historical absolute radiation correction gain; Determine the current absolute fused radiative correction bias for the band: ,in, For band The current absolute fusion radiative correction bias, For band Historical absolute radiation correction bias; The current absolute radiometric calibration parameters for the corresponding band are determined based on the current absolute fused radiometric correction gain and the current absolute fused radiometric correction bias.

9. A spaceborne optical pushbroom image on-orbit correction device, characterized in that, include: The acquisition unit is used to acquire pushbroom images of the ground coverage area to be corrected obtained by the spaceborne optical imaging device. The relative radiometric correction unit is used to identify the calibration reference data source in the ground coverage area, determine the calibration reference image area in the pushbroom image based on the calibration reference data source, determine the current relative radiometric calibration parameter of the corresponding probe based on the pixel gray value detected by each probe in the calibration reference image area, and perform relative radiometric correction on the pushbroom image element by element based on the current relative radiometric calibration parameter. An absolute radiometric correction unit is used to locate the projection area of ​​the preset pseudo-invariant calibration field in the pushbroom image after relative radiometric correction, based on the geographical range of the preset pseudo-invariant calibration field, to obtain the calibration image window area, obtain the ideal pixel gray value of each band in the calibration image window area corresponding to the preset pseudo-invariant calibration field, and determine the current absolute radiometric calibration parameters of the corresponding band based on the ideal pixel gray value and the actual pixel gray value of each band in the calibration image window area in the pushbroom image, and perform absolute radiometric correction on the pushbroom image after relative radiometric correction. The geometric correction unit is used to identify the position of the key area in the ground coverage area within the pushbroom image after absolute radiometric correction as a registration window area. Within the registration window area, geometric correction is performed on the pushbroom image after absolute radiometric correction based on the feature point matching relationship between the key area and the pushbroom image after absolute radiometric correction.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 8.

Images