Methods, apparatus, equipment and storage media for rapid geometric verification of hyperspectral data

Abstract

This invention relates to the field of data verification technology, and particularly to a method, apparatus, device, and storage medium for rapid geometric verification of hyperspectral data. The method includes: acquiring a high-resolution visible light area array image and preprocessing it to obtain a visible light orthorectified image; acquiring a raw hyperspectral image and performing coarse geometric correction to obtain a coarsely geometrically corrected hyperspectral orthorectified image; performing feature extraction and matching processing on the two images obtained above, constructing a large number of pseudo-control points based on the matching results to obtain a set of point pairs including geographic coordinates; calculating the true scan vector of each scanned pixel in the coordinates of a digital land model, and calculating the corresponding geographic coordinates; constructing a cost function based on the set of point pairs and the calculated geographic coordinates; and iteratively solving the hyperspectral geometric verification parameters with the goal of minimizing the cost function to perform geometric correction on the raw hyperspectral image. The method disclosed in this application can significantly improve the efficiency of geometric verification and reduce the verification cost.

Description

Technical Field

[0001] This invention relates to the field of defect detection technology, and in particular to a rapid geometric verification method, apparatus, equipment, and storage medium for hyperspectral data. Background Technology

[0002] Airborne hyperspectral imaging is a remote sensing technology that uses a hyperspectral camera mounted on an aircraft to scan the Earth's surface at high speed. The imaging typically employs a pushbroom (linear array) method. This remote sensing technology acquires wide-view, high-spectral-resolution linear array image data. After geometric and radiometric corrections, the acquired linear array image data can be rapidly obtained to acquire spectral information of large areas of the Earth's surface. Compared to terrestrial hyperspectral technology, airborne hyperspectral imaging technology has a wider coverage area, higher spatial resolution, and faster data acquisition speed, providing more detailed surface spectral information while reducing acquisition costs.

[0003] The preprocessing of aerial hyperspectral image data mainly includes two steps: radiometric correction and geometric correction. Geometric correction requires the POS system to provide precise positioning and orientation parameters for the image, and with the assistance of the digital terrain model, the accurate calculation of the planar position of the hyperspectral image can be achieved. During the equipment integration and installation process, since the IMU's own coordinate system cannot be kept completely parallel to the camera's line of sight, the resulting line of sight deviation angle will seriously affect the geometric correction effect of the hyperspectral image. Therefore, it is necessary to carry out geometric calibration of the installation angle between the hyperspectral camera and the IMU.

[0004] Geometric calibration methods for airborne hyperspectral cameras can generally be categorized into two types. The first method involves setting up multiple overlapping survey lines, using the consistency of object-space geometric information from different camera perspectives as a constraint, and employing a bundle adjustment model to solve for calibration parameters. However, this method requires setting up a calibration field on the ground and has high requirements for the overlap between flight strips, making it difficult to implement in complex terrain areas. The second method involves setting up a large number of ground control points along a single flight strip, using the consistency of their geographical coordinates with the camera's object-space coordinates as a constraint, and employing collinearity equations to solve for calibration parameters. However, this method requires significant manpower and resources, resulting in high costs and risks. To reduce the cost of setting up control points in the field, high-resolution satellite imagery of the same area can be acquired, and corresponding features of the satellite imagery and hyperspectral imagery can be extracted. Pseudo-control points can then be constructed based on the geographical locations of the satellite imagery. However, due to the difficulty in matching features between satellite and aerial imagery from different time phases and the low positioning accuracy of satellite imagery, this method suffers from poor model stability and inaccurate calculation results.

[0005] It is evident that existing technologies still need improvement and enhancement. Summary of the Invention

[0006] In order to overcome the shortcomings of the prior art, the present invention aims to provide a method, apparatus, device and storage medium for rapid geometric verification of hyperspectral data, which can significantly improve the efficiency of geometric verification and reduce the verification cost.

[0007] The first aspect of this invention provides a rapid geometric calibration method for hyperspectral data, comprising: acquiring a high-resolution visible light array image; preprocessing the high-resolution visible light array image to obtain a visible light orthorectified image; acquiring a raw hyperspectral image; performing coarse geometric calibration on the raw hyperspectral image based on POS data to obtain a coarsely geometrically calibrated hyperspectral image; sequentially performing feature extraction and matching processing on the visible light orthorectified image and the coarsely geometrically calibrated hyperspectral image to obtain matching results; constructing a large number of pseudo-control points based on the matching results to obtain a set of point pairs of pixel coordinates and corresponding geographic coordinates of the raw hyperspectral image; calculating the true scan vector of each scanned pixel in the coordinates of a digital land model in the raw hyperspectral image, and calculating the geographic coordinates corresponding to the scanned pixel based on the calculated true scan vector; constructing a cost function based on the set of point pairs and the calculated geographic coordinates corresponding to the scanned pixel; iteratively solving the hyperspectral geometric calibration parameters with the minimum cost function as the optimization objective; and using the solved hyperspectral geometric calibration parameters to perform geometric calibration on the raw hyperspectral image.

[0008] Optionally, in a first embodiment of the first aspect of the present invention, the step of acquiring a high-resolution visible light array image and preprocessing the high-resolution visible light array image to obtain a visible light orthorectified image specifically includes: acquiring a high-resolution visible light array image; performing orthorectification on the acquired high-resolution visible light array image based on artificial control points and collinearity equations to obtain a corrected image; stitching the corrected image based on the address position of the visible light array image to obtain a stitched image; and downsampling the stitched image to the same spatial resolution as the original hyperspectral image to obtain a visible light orthorectified image.

[0009] Optionally, in a second embodiment of the first aspect of the present invention, the step of sequentially performing feature extraction and matching processing on the visible light orthophoto and the hyperspectral coarse geometrical corrected orthophoto to obtain matching results, and constructing a large number of pseudo control points based on the matching results to obtain a set of point pairs of pixel coordinates and corresponding geographic coordinates of the hyperspectral original image, specifically includes: calculating the spatial distance between the artificial control points and the control points on the hyperspectral coarse geometrical corrected orthophoto, which is set as the offset; using the SIFT operator to perform feature extraction and matching on the visible light orthophoto and the hyperspectral coarse geometrical corrected orthophoto to obtain matching point pairs; using coordinate transformation to obtain the geographic coordinates of the matching point pairs, and calculating the distance between the matching point pairs; if the calculated distance is less than the offset, then the corresponding matching point pair is retained to obtain a set of matching point pairs; and constructing a large number of pseudo control points based on the set of matching point pairs to obtain a set of point pairs of pixel coordinates and corresponding geographic coordinates of the hyperspectral original image.

[0010] Optionally, in a third embodiment of the first aspect of the present invention, the calculation of the true scan vector of each scanned pixel in the coordinates of the digital land model of the hyperspectral raw image specifically includes: calculating the first scan vector L'0 of the hyperspectral raw image in the sensor coordinates in the initial state; calculating the second scan vector L0 under normal flight conditions based on the roll angle, pitch angle, heading angle and the calculated first scan vector L'0; and calculating the true scan vector L1 of each scanned pixel based on the flight altitude and the second scan vector L0.

[0011] Optionally, in a fourth embodiment of the first aspect of the present invention, the step of calculating the geographic coordinates corresponding to the scanned pixels based on the calculated real scan vector specifically includes: creating a vertical projection of the real scan vector based on a digital surface model, and defining the starting point P of the vertical projection according to the flight altitude and the real scan vector L1. S and endpoint P G From the starting point P S Starting from the beginning, traverse pixel by pixel along the projection line of the vertical projection, calculating the pixel distance between the current pixel and the target pixel; if the pixel distance is less than one pixel, obtain the geographic coordinates P corresponding to the scanned pixel. pix .

[0012] Optionally, in a fifth embodiment of the first aspect of the present invention, the step of constructing a cost function based on the set of point pairs and the calculated geographic coordinates corresponding to the scanned pixels specifically includes: obtaining the geographic coordinates P in the set of point pairs. match According to geographic coordinates P match and geographic coordinates P pix Construct a cost function, wherein the cost function is:

[0013]

[0014] Where, δω, δκ represents the installation deflection angles for roll, pitch, and yaw, respectively.

[0015] Optionally, in a sixth embodiment of the first aspect of the present invention, the step of iteratively solving for hyperspectral geometric calibration parameters with the goal of minimizing the cost function, and then using the solved hyperspectral geometric calibration parameters to perform geometric correction on the original hyperspectral image, specifically includes: using the cost function... The minimum is the optimization objective. The Gauss-Newton optimization method is used to iteratively solve the hyperspectral geometric calibration parameters. When the iteration exit condition is met, the iteration is exited and the current value is returned as the final calibration parameter. The final calibration parameter is used to perform geometric correction on the original hyperspectral image.

[0016] A second aspect of the present invention provides a rapid geometric calibration device for hyperspectral data, comprising: a processing module for acquiring a high-resolution visible light array image and preprocessing the high-resolution visible light array image to obtain a visible light orthorectified image; a calibration module for acquiring a hyperspectral raw image and performing coarse geometric calibration on the hyperspectral raw image based on POS data to obtain a coarsely geometrically calibrated hyperspectral orthorectified image; a matching module for sequentially performing feature extraction and matching processing on the visible light orthorectified image and the coarsely geometrically calibrated hyperspectral orthorectified image to obtain a matching result, and constructing a large number of pseudo-control points based on the matching result to obtain a set of point pairs of pixel coordinates and corresponding geographic coordinates of the hyperspectral raw image; a calculation module for calculating the true scan vector of each scanned pixel in the coordinates of a digital land model of the hyperspectral raw image, and calculating the geographic coordinates corresponding to the scanned pixel based on the calculated true scan vector; a construction module for constructing a cost function based on the set of point pairs and the calculated geographic coordinates corresponding to the scanned pixel; and an iteration module for iteratively solving the hyperspectral geometric calibration parameters with the minimum cost function as the optimization objective, and using the solved hyperspectral geometric calibration parameters to perform geometric calibration on the hyperspectral raw image.

[0017] A third aspect of the present invention provides a rapid geometric verification device for hyperspectral data, the rapid geometric verification device for hyperspectral data comprising: a memory and at least one processor, the memory storing instructions; at least one processor calling the instructions in the memory to cause the rapid geometric verification device for hyperspectral data to perform the various steps of the rapid geometric verification method for hyperspectral data described in any of the preceding claims.

[0018] A fourth aspect of the present invention provides a computer-readable storage medium storing instructions that, when executed by a processor, implement the steps of the rapid geometric verification method for hyperspectral data described in any of the preceding claims.

[0019] In the technical solution of this invention, a hyperspectral coarse geometrically calibrated hyperspectral image is obtained by acquiring the original hyperspectral image and performing geometric coarse correction. Feature extraction and matching processing are performed on the two images obtained above. Based on the matching results, a large number of pseudo-control points are constructed to obtain a set of point pairs including geographic coordinates. The true scan vector of each scanned pixel in the coordinates of the digital land model is calculated, and the corresponding geographic coordinates are calculated. A cost function is constructed based on the set of point pairs and the calculated geographic coordinates. The hyperspectral geometric calibration parameters are iteratively solved with the minimum cost function as the optimization objective to perform geometric correction on the original hyperspectral image. The method disclosed in this application can complete the correction without deploying a large number of manual control points and has no special requirements for the calibration flight scheme, significantly reducing the difficulty and complexity of the geometric calibration task, improving geometric calibration efficiency, and reducing geometric calibration costs. Furthermore, the accuracy of the obtained geometric calibration parameters is high, which can effectively improve the geometric correction effect. Attached Figure Description

[0020] Figure 1 This is a first flowchart of a rapid geometric verification method for hyperspectral data provided in an embodiment of the present invention;

[0021] Figure 2 This is a second flowchart of the rapid geometric verification method for hyperspectral data provided in an embodiment of the present invention;

[0022] Figure 3 This is a third flowchart of the rapid geometric verification method for hyperspectral data provided in the embodiments of the present invention;

[0023] Figure 4 This is a fourth flowchart of the rapid geometric verification method for hyperspectral data provided in the embodiments of the present invention;

[0024] Figure 5 The fifth flowchart of the rapid geometric verification method for hyperspectral data provided in the embodiments of the present invention;

[0025] Figure 6 This is a schematic diagram of a rapid geometric verification device for hyperspectral data provided in an embodiment of the present invention;

[0026] Figure 7 This is a schematic diagram of the structure of the rapid geometric verification device for hyperspectral data provided in an embodiment of the present invention;

[0027] Figure 8 This is a schematic diagram of the first scan vector, the second scan vector, and the scan line vector involved in the present invention;

[0028] Figure 9 This is a schematic diagram of the vertical projection of the scan line vector involved in the present invention;

[0029] Figure 10 For the purposes of this invention, the geographical coordinates P are defined.pix A schematic diagram of the calculation method. Detailed Implementation

[0030] This invention provides a method, apparatus, device, and storage medium for rapid geometric calibration of hyperspectral data. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0031] For ease of understanding, the specific process of the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 1 One embodiment of the rapid geometric verification method for hyperspectral data in this invention includes:

[0032] 101. Acquire high-resolution visible light array images, preprocess the high-resolution visible light array images to obtain visible light orthophoto images;

[0033] In this embodiment, the high-resolution visible light array image is acquired and fed back by a hyperspectral camera mounted on an aircraft.

[0034] 102. Acquire the original hyperspectral image, and perform geometric coarse correction on the original hyperspectral image based on POS data to obtain the coarse geometrically corrected hyperspectral orthophoto image.

[0035] In this embodiment, the POS data is provided by the POS system.

[0036] 103. Perform feature extraction and matching processing on the visible light orthophoto and the hyperspectral coarse geometric correction orthophoto in sequence to obtain the matching results. Based on the matching results, construct a large number of pseudo control points to obtain the set of point pairs of pixel coordinates and corresponding geographic coordinates of the original hyperspectral image.

[0037] In this embodiment, feature extraction and matching processes are sequentially performed on visible light orthophotos and hyperspectral coarse geometrical orthophotos to generate multiple pseudo control points, thereby enabling the calculation of pixel coordinates and geographic coordinates of the original hyperspectral image.

[0038] 104. Calculate the true scan vector of each scanned pixel in the coordinates of the digital land model based on the original hyperspectral image, and calculate the geographic coordinates corresponding to the scanned pixel based on the calculated true scan vector;

[0039] 105. Construct a cost function based on the set of point pairs and the calculated geographic coordinates corresponding to the scanned pixels;

[0040] 106. With the goal of minimizing the cost function, iteratively solve for the hyperspectral geometric calibration parameters, and use the solved hyperspectral geometric calibration parameters to perform geometric correction on the original hyperspectral image.

[0041] This application discloses a rapid geometric calibration method for hyperspectral data. It involves acquiring a raw hyperspectral image and performing coarse geometric correction to obtain a coarsely geometrically calibrated hyperspectral image. Feature extraction and matching processing are performed on the two images obtained. Based on the matching results, a large number of pseudo-control points are constructed to obtain a set of point pairs including geographic coordinates. The true scan vector of each scanned pixel in the coordinate system of a digital land model is calculated, and the corresponding geographic coordinates are also calculated. A cost function is constructed based on the set of point pairs and the calculated geographic coordinates. The hyperspectral geometric calibration parameters are iteratively solved with the minimum cost function as the optimization objective to perform geometric calibration on the raw hyperspectral image. The method disclosed in this application can complete the calibration without manually setting up a large number of ground control points, and it has no special requirements for survey line layout and calibration flight schemes. It is applicable to areas with complex terrain, improving applicability, significantly reducing the difficulty and complexity of geometric calibration tasks, improving geometric calibration efficiency, and reducing geometric calibration costs. Furthermore, the accuracy of the obtained geometric calibration parameters is high, which can effectively improve the geometric calibration effect.

[0042] Please see Figure 2 The second embodiment of the rapid geometric verification method for hyperspectral data in this invention includes:

[0043] 201. Acquire high-resolution visible light array images, and perform orthorectification on the acquired high-resolution visible light array images based on artificial control points and collinearity equations to obtain corrected images;

[0044] In this embodiment, the collinearity equation is used to perform orthorectification on the acquired high-resolution visible light array image. The collinearity equation is a mathematical relationship that expresses the three points (object point, image point, and projection center, which is usually the lens center for the image) lying on a straight line. It is one of the most basic formulas in photogrammetry.

[0045] 202. Based on the address location of the visible light array image, the corrected image is stitched together to obtain the stitched image;

[0046] 203. Downsample the mosaicked image to the same spatial resolution as the original hyperspectral image to obtain a visible light orthophoto image;

[0047] In this embodiment, the stitched image is a high-resolution area array image, and the spatial resolution of the visible light orthophoto image is consistent with the spatial resolution of the original hyperspectral image.

[0048] Please see Figure 3 The third embodiment of the rapid geometric verification method for hyperspectral data in this invention includes:

[0049] 301. Calculate the spatial distance between the artificial control point and the control point on the hyperspectral coarse geometric correction image, and set it as the offset;

[0050] In this embodiment, the artificial control point is a ground control point laid manually. The actual geographical location of the artificial control point is known, and the geographical location of the artificial control point in the hyperspectral coarse geometric correction image is known. The distance between two control points is calculated as the offset to determine whether it is necessary to retain the matching point pair.

[0051] 302. Use the SIFT operator to extract and match features from visible light orthophotos and hyperspectral coarse geometrically corrected orthophotos to obtain matching point pairs;

[0052] In this embodiment, the SIFT operator is used to extract and match features from visible light orthophotos and hyperspectral coarse geometrically corrected orthophotos. SIFT stands for Scale Invariant Feature Transform. The steps of SIFT feature detection include: extreme value detection in scale space, feature point localization, feature orientation assignment, and feature point description. SIFT features are invariant to rotation, scale scaling, brightness changes, etc., and are a very stable local feature with characteristics such as good uniqueness, high quantity, high speed, and scalability.

[0053] 303. Use coordinate transformation to obtain the geographic coordinates of the matching point pairs, and calculate the distance between the matching point pairs. If the calculated distance is less than the offset, retain the corresponding matching point pairs to obtain the set of matching point pairs.

[0054] In this embodiment, by filtering the matching point pairs, the accuracy and effectiveness of the subsequently constructed pseudo-control points can be improved, thereby improving the accuracy of the output geometric calibration parameters.

[0055] 304. Construct a large number of pseudo-control points based on the matching point pair set to obtain the set of point pairs of pixel coordinates and corresponding geographic coordinates of the hyperspectral original image;

[0056] In this embodiment, the point pair set includes multiple sets of point pairs, and each set of point pairs includes the pixel coordinates of the hyperspectral original image and the geographic coordinates corresponding to the pixel coordinates.

[0057] Please see Figure 4 and Figures 8 to 10 The fourth embodiment of the rapid geometric verification method for hyperspectral data in this invention includes:

[0058] 401. Calculate the first scan vector L'0 of the initial hyperspectral raw image in the sensor coordinates;

[0059] In this embodiment, we first assume that the initial state is when all flight attitude angles of the sensor are 0 at a certain moment. We then calculate the first scan vector L'0 of the initial state hyperspectral raw image in the sensor coordinates, as follows:

[0060]

[0061]

[0062] L' 0,z =-1;

[0063] In the formula, maxP is the maximum number of pixels in the scan line of the original hyperspectral image, and c is the pixel coordinate of the original hyperspectral image. It is the horizontal field of view. It is the vertical field of view; in the pushbroom sensor coordinate system, the horizontal axis L' 0,x The variation depends on the angle between the scanning direction and the nadir direction, and can be calculated from the horizontal field of view. The vertical axis L' 0,y The variation depends on the angle between the scanning direction and the nadir direction, and can be calculated from the vertical field of view, Z-axis L' 0,z It is -1.

[0064] 402. Based on the roll angle, pitch angle, yaw angle, and the calculated first scan vector L'0, calculate the second scan vector L0 under normal flight conditions;

[0065] In this embodiment, under normal flight conditions, each flight attitude angle is different. At this time, the scan vector needs to take into account the roll angle, pitch angle, and yaw angle. Based on the attitude angle data provided by the POS system and the initial values ​​of the hyperspectral calibration parameters, the transformation relationship between the scan vector of the digital land model coordinate system and the sensor coordinate system can be constructed. The specific formula of the second scan vector L0 is as follows:

[0066] L0 = RPHL'0;

[0067] In the formula, R, P, and H represent the roll angle, pitch angle, and yaw angle, respectively, and the formula is expanded as follows:

[0068]

[0069] In the formula, δω, δκ represents the installation deflection angles for roll, pitch, and yaw, respectively.

[0070] 403. Calculate the true scan vector L1 for each scan pixel based on the flight altitude and the second scan vector L0;

[0071] In this embodiment, the flight altitude is the flight altitude of the aircraft, denoted as h. The vector magnitude is converted into an actual value using the aircraft's flight altitude h, thereby calculating the true scan vector L1 of each scan pixel in the sensor scan row. The formula is as follows:

[0072]

[0073] 404. Create a vertical projection of the real scan vector based on the digital terrain model, and define the starting point P of the vertical projection according to the flight altitude and the real scan vector L1. S and endpoint P G ;

[0074] 405. Starting from point P S Starting from the beginning, traverse pixel by pixel along the projection line of the vertical projection and calculate the pixel distance between the current pixel and the target pixel;

[0075] 406. If the pixel distance is less than one pixel, obtain the geographic coordinates P corresponding to the scanned pixel. pix ;

[0076] In this embodiment, the geographic coordinates corresponding to hyperspectral scanned pixels are calculated based on ray tracing and a digital terrain model. The main idea of ​​ray tracing is to emit several rays from the camera to the pixels on the screen, find the intersection point of the nearest object with the ray, and if the surface at that point is a scattering surface, calculate the color produced by the light source directly illuminating that point; if the surface at that point is a specular or refracting surface, start from that intersection point and trace the ray in the direction of reflection or refraction as a new incident ray, continuing to find intersections with objects in the scene, and so on recursively until the ray escapes the scene or reaches a set maximum recursion depth. Finally, the color values ​​of each ejection point are added together to obtain the color value of the initial pixel; the essential algorithm of ray tracing is the ray intersection problem.

[0077] Please see Figure 5 The fifth embodiment of the rapid geometric verification method for hyperspectral data in this invention includes:

[0078] 501. Obtain the geographic coordinates P in the set of point pairs. match ;

[0079] 502. Based on geographic coordinates P match and geographic coordinates P pix Construct a cost function, wherein the cost function is:

[0080]

[0081] Where, δω, δκ represents the installation deflection angles for roll, pitch, and yaw, respectively.

[0082] 503. Using the cost function The minimum is the optimization objective, and the hyperspectral geometric calibration parameters are solved iteratively using the Gauss-Newton optimization method.

[0083] In this embodiment, the Gauss-Newton optimization method is used to iteratively solve the hyperspectral geometric calibration parameters. The Gauss-Newton method of numerical optimization refers to a relationship between the dependent and independent variables that is not linear, such as a square relationship, a logarithmic relationship, an exponential relationship, a trigonometric function relationship, etc. For such equations, it is often difficult to obtain an exact solution by finding the optimal point of an n-variable real function f in the entire n-dimensional vector space Rn, and an approximate solution is often required. The methods for solving this optimization problem are mostly iterative algorithms of successive one-dimensional search. The basic idea is to select a favorable search direction at an approximate point, perform a one-dimensional search along this direction, obtain a new approximate point, and repeat this iteratively until the predetermined accuracy requirement is met.

[0084] 504. When the iteration exit condition is met, exit the iteration and return the current value as the final calibration parameter;

[0085] In this embodiment, the iteration exit condition can be whether the iteration step size is less than a threshold or whether the number of iterations exceeds the upper limit. When the iteration step size is less than the threshold or the number of iterations exceeds the upper limit, the iteration exits and the current value is returned as the final calibration parameter.

[0086] 505. Perform geometric correction on the original hyperspectral image using the final calibration parameters.

[0087] The rapid geometric calibration method for hyperspectral data in the embodiments of the present invention has been described above. The rapid geometric calibration device for hyperspectral data in the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 6 One embodiment of the rapid geometric calibration device for hyperspectral data in this invention includes:

[0088] Processing module 601 is used to acquire high-resolution visible light array images, preprocess the high-resolution visible light array images, and obtain visible light orthophoto images.

[0089] The correction module 602 is used to acquire the original hyperspectral image and perform geometric coarse correction on the original hyperspectral image based on POS data to obtain a coarse geometrically corrected hyperspectral orthophoto image.

[0090] The matching module 603 is used to perform feature extraction and matching processing on the visible light orthophoto and the hyperspectral coarse geometric correction orthophoto in sequence to obtain the matching results. Based on the matching results, a large number of pseudo control points are constructed to obtain a set of point pairs of pixel coordinates and corresponding geographic coordinates of the original hyperspectral image.

[0091] The calculation module 604 is used to calculate the true scan vector of each scanned pixel in the coordinates of the digital land model of the original hyperspectral image, and to calculate the geographic coordinates corresponding to the scanned pixel based on the calculated true scan vector.

[0092] Module 605 is used to construct a cost function based on the set of point pairs and the calculated geographic coordinates corresponding to the scanned pixels;

[0093] The iteration module 606 is used to iteratively solve the hyperspectral geometric calibration parameters with the goal of minimizing the cost function, and then use the solved hyperspectral geometric calibration parameters to perform geometric correction on the original hyperspectral image.

[0094] above Figure 6 The rapid geometric calibration device for hyperspectral data in this embodiment of the invention is described in detail from the perspective of modular functional entities. The rapid geometric calibration device for hyperspectral data in this embodiment of the invention is described in detail below from the perspective of hardware processing.

[0095] Figure 7 This is a schematic diagram of a rapid geometrical calibration device for hyperspectral data provided in an embodiment of the present invention. The rapid geometrical calibration device 700 for hyperspectral data can vary considerably due to different configurations or performance characteristics. It may include one or more central processing units (CPUs) 710 (e.g., one or more processors) and a memory 720, and one or more storage media 730 (e.g., one or more mass storage devices) storing application programs 733 or data 732. The memory 720 and storage media 730 may be temporary or persistent storage. The program stored in the storage media 730 may include one or more modules (not shown in the diagram), each module including a series of instruction operations on the rapid geometrical calibration device 700 for hyperspectral data. Furthermore, the processor 710 may be configured to communicate with the storage media 730 and execute the series of instruction operations in the storage media 730 on the rapid geometrical calibration device 700 for hyperspectral data to implement the steps of the rapid geometrical calibration method for hyperspectral data provided in the above-described method embodiments.

[0096] The hyperspectral data rapid geometric calibration device 700 may also include one or more power supplies 740, one or more wired or wireless network interfaces 750, one or more input / output interfaces 760, and / or one or more operating systems 731, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. Those skilled in the art will understand that the structure of the hyperspectral data rapid geometric calibration device shown in this application does not constitute a limitation on hyperspectral data-based rapid geometric calibration devices, and may include more or fewer components than illustrated, or combine certain components, or have different component arrangements.

[0097] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when executed on a computer, cause the computer to perform the steps of a rapid geometric verification method for hyperspectral data.

[0098] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the system, device, or unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0099] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0100] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A rapid geometric verification method for hyperspectral data, characterized in that, include: Acquire high-resolution visible light array images, preprocess the high-resolution visible light array images to obtain visible light orthophoto images; Acquire the original hyperspectral image, and perform coarse geometric correction on the original hyperspectral image based on POS data to obtain a coarse geometrically corrected hyperspectral orthophoto image. Feature extraction and matching processes are performed sequentially on visible light orthophotos and hyperspectral coarse geometrically corrected orthophotos to obtain matching results. Based on the matching results, a large number of pseudo control points are constructed to obtain a set of point pairs of pixel coordinates and corresponding geographic coordinates of the original hyperspectral image. Calculate the true scan vector of each scanned pixel in the coordinates of the digital land model based on the original hyperspectral image, and calculate the geographic coordinates corresponding to the scanned pixel based on the calculated true scan vector; A cost function is constructed based on the set of point pairs and the calculated geographic coordinates corresponding to the scanned pixels. With the goal of minimizing the cost function, the hyperspectral geometric calibration parameters are iteratively solved, and the solved hyperspectral geometric calibration parameters are used to perform geometric correction on the original hyperspectral image.

2. The rapid geometric verification method for hyperspectral data according to claim 1, characterized in that, The process of acquiring a high-resolution visible light area array image and preprocessing the high-resolution visible light area array image to obtain a visible light orthophoto image specifically includes: High-resolution visible light array images are acquired, and orthorectification is performed on the acquired high-resolution visible light array images based on artificial control points and collinearity equations to obtain corrected images. The corrected image is stitched together based on the address location of the visible light array image to obtain the stitched image; The mosaicked image is downsampled to the same spatial resolution as the original hyperspectral image to obtain a visible light orthophoto.

3. The rapid geometric verification method for hyperspectral data according to claim 1, characterized in that, The visible light orthorectified image and the hyperspectral coarse geometrically corrected orthorectified image are sequentially subjected to feature extraction and matching processing to obtain matching results. Based on the matching results, a large number of pseudo control points are constructed to obtain a set of point pairs of pixel coordinates and corresponding geographic coordinates of the original hyperspectral image. Specifically, this includes: Calculate the spatial distance between the artificial control point and the control point on the hyperspectral coarse geometrically corrected orthophoto image, and set it as the offset; The SIFT operator is used to extract and match features from visible light orthorectified images and hyperspectral coarse geometrically corrected orthorectified images to obtain matching point pairs. The geographic coordinates of the matching point pairs are obtained by coordinate transformation, and the distance between the matching point pairs is calculated. If the calculated distance is less than the offset, the corresponding matching point pairs are retained to obtain the set of matching point pairs. A large number of pseudo-control points are constructed based on the matching point pair set to obtain a set of point pairs of pixel coordinates and corresponding geographic coordinates of the hyperspectral raw image.

4. The rapid geometric verification method for hyperspectral data according to claim 3, characterized in that, The calculation of the true scan vector of each scanned pixel in the hyperspectral original image under the coordinates of the digital land model specifically includes: Calculate the first scan vector L'0 of the initial hyperspectral raw image in sensor coordinates; Based on the roll angle, pitch angle, yaw angle, and the calculated first scan vector L'0, calculate the second scan vector L0 under normal flight conditions; The true scan vector L1 of each scan pixel is calculated based on the flight altitude and the second scan vector L0.

5. The rapid geometric verification method for hyperspectral data according to claim 4, characterized in that, The calculation of geographic coordinates corresponding to scanned pixels based on the calculated real scan vectors specifically includes: A vertical projection of the real scan vector is created based on the digital terrain model. The starting point P of the vertical projection is defined according to the flight altitude and the real scan vector L1. S and endpoint P G ; From the starting point P S Starting from the beginning, traverse pixel by pixel along the projection line of the vertical projection and calculate the pixel distance between the current pixel and the target pixel; If the pixel distance is less than one pixel, then obtain the geographic coordinates P corresponding to the scanned pixel. pix .

6. The rapid geometric verification method for hyperspectral data according to claim 5, characterized in that, The construction of the cost function based on the set of point pairs and the calculated geographic coordinates corresponding to the scanned pixels specifically includes: Get the geographic coordinates P in the set of point pairs match ; Based on geographic coordinates P match and geographic coordinates P pix Construct a cost function, wherein the cost function is: Where, δω, δκ represents the installation deflection angles for roll, pitch, and yaw, respectively.

7. The rapid geometric verification method for hyperspectral data according to claim 6, characterized in that, The process involves iteratively solving for hyperspectral geometric calibration parameters with the goal of minimizing the cost function, and then using these parameters to perform geometric correction on the original hyperspectral image. Specifically, this includes: With cost function The minimum is the optimization objective, and the hyperspectral geometric calibration parameters are solved iteratively using the Gauss-Newton optimization method. When the iteration exit condition is met, the iteration exits and the current value is returned as the final calibration parameter. Geometric correction was performed on the original hyperspectral image using the final calibration parameters.

8. A rapid geometric calibration device for hyperspectral data, characterized in that, include: The processing module is used to acquire high-resolution visible light array images, preprocess the high-resolution visible light array images, and obtain visible light orthophoto images. The correction module is used to acquire the original hyperspectral image and perform geometric coarse correction on the original hyperspectral image based on POS data to obtain a coarse geometrically corrected hyperspectral orthophoto image. The matching module is used to perform feature extraction and matching processing on visible light orthophotos and hyperspectral coarse geometric correction orthophotos in sequence to obtain matching results. Based on the matching results, a large number of pseudo control points are constructed to obtain a set of point pairs of pixel coordinates and corresponding geographic coordinates of the original hyperspectral image. The calculation module is used to calculate the true scan vector of each scanned pixel in the coordinates of the digital land model of the original hyperspectral image, and to calculate the geographic coordinates corresponding to the scanned pixel based on the calculated true scan vector. The module is used to construct a cost function based on the set of point pairs and the calculated geographic coordinates corresponding to the scanned pixels; The iterative module is used to iteratively solve for the hyperspectral geometric calibration parameters with the goal of minimizing the cost function, and then use the solved hyperspectral geometric calibration parameters to perform geometric correction on the original hyperspectral image.

9. A rapid geometric calibration device for hyperspectral data, characterized in that, The hyperspectral data rapid geometric verification device includes: a memory and at least one processor, wherein the memory stores instructions; At least one of the processors invokes the instructions in the memory to cause the hyperspectral data rapid geometric verification device to perform the steps of the hyperspectral data rapid geometric verification method as described in any one of claims 1-7.

10. A computer-readable storage medium storing instructions thereon, characterized in that, When the instructions are executed by the processor, they implement the steps of the rapid geometric verification method for hyperspectral data as described in any one of claims 1-7.

Images