Error correction method and device, computer equipment, readable storage medium and program product

Abstract

The invention relates to an error correction method and device, computer equipment, a readable storage medium and a program product. The method comprises the following steps: constructing a linear correction model based on position data of a multispectral sensor and multispectral remote sensing image distortion; the linear correction model represents a residual value between a real elevation and image point elevations in different multispectral remote sensing images; linearizing an original error equation corresponding to the linear correction model to obtain projection image points; according to the projection image point, iteratively updating the calibration parameter of the original error equation, and determining a correction error equation according to the correction value of the calibration parameter in each iteration; and correcting the geometric error of the multispectral remote sensing image through the correction error equation. The geometric distortion of the multispectral remote sensing image can be effectively corrected.

Description

Technical Field

[0001] This application relates to the field of image processing technologies, and particularly to an error correction method, apparatus, computer device, computer-readable storage medium, and computer program product. Background Art

[0002] In traditional remote sensing images, ortho-images or approximate ortho-images have inevitable geometric errors (also known as geometric distortions). Among them, the geometric distortions of multi-spectral remote sensing images can generally be divided into two categories: internal distortions caused by differences in sensor performance and external distortions caused by changes in the attitude of the carrier vehicle or changes in the target object.

[0003] However, traditional remote sensing image processing methods lack technical means for efficiently processing the above geometric distortions and are difficult to correct the geometric errors of multi-spectral remote sensing images. Summary of the Invention

[0004] Based on this, it is necessary to provide an error correction method, apparatus, computer device, computer-readable storage medium, and computer program product that can effectively correct the geometric distortions of multi-spectral remote sensing images for the above technical problems.

[0005] In a first aspect, in one embodiment, this application provides an error correction method, and the method includes:

[0006] Based on the position data of the multi-spectral sensor and the distortion of the multi-spectral remote sensing image, a linear correction model is constructed; the linear correction model represents the residual value between the true elevation and the elevation of the image points in different multi-spectral remote sensing images;

[0007] The original error equation corresponding to the linear correction model is linearized to obtain the projected image points;

[0008] According to the projected image points, the calibration parameters of the original error equation are iteratively updated, and the correction error equation is determined according to the correction value of the calibration parameters in each round of iteration;

[0009] The geometric errors of the multi-spectral remote sensing image are corrected through the correction error equation.

[0010] In one of the embodiments, the position data of the multi-spectral sensor includes the sensor position and the unknown ground position;

[0011] Based on the position data of the multi-spectral sensor and the distortion of the multi-spectral remote sensing image, constructing a linear correction model includes:

[0012] According to the sensor position, the unknown ground position, and the distortion of the multi-spectral remote sensing image, combined with the rotation matrix and the offset matrix, an original correction model is established; the offset matrix is used to compensate for the exterior orientation error;

[0013] Expand the offset matrix based on the rotation parameters in the body coordinate system of the multispectral sensor;

[0014] Construct a linear correction model according to the expanded offset matrix and in combination with the original correction model.

[0015] In one embodiment, linearize the original error equation corresponding to the linear correction model to obtain the projection image points, including:

[0016] Determine the projection image points according to the distortion correction parameters, image point correction parameters of the linear correction model, and the geographical longitude and latitude of the object point.

[0017] In one embodiment, update the calibration parameters of the original error equation according to the projection image points, and determine the correction error equation according to the correction value of the calibration parameters, including:

[0018] Determine the normal equation of the original error equation according to the projection image points;

[0019] Based on the correction value of the geographical longitude and latitude of the object point, iteratively update the calibration parameters through a preset iterative algorithm, and obtain the correction value of the calibration parameters in each round of iteration;

[0020] When the correction value of the calibration parameters in this round of iteration meets the preset numerical condition, determine the correction error equation according to the current calibration parameters, in combination with the number of multispectral remote sensing images collected by the multispectral sensor and the number of ground points; wherein, the ground points represent the intersection points of different collected multispectral remote sensing images.

[0021] In one embodiment, the preset iterative algorithm includes the iterative correction eigenvalue method.

[0022] In one embodiment, correct the geometric error of the multispectral remote sensing image through the correction error equation, including:

[0023] Eliminate the heterogeneity and heterogeneity in the multispectral remote sensing image data set collected by the multispectral sensor through the correction error equation, so as to assimilate the spatial data set of the multispectral remote sensing image; wherein, the spatial data set is formed after the geometric combination of the multispectral remote sensing images.

[0024] In a second aspect, in one embodiment, the present application provides an error correction device, and the device includes:

[0025] A model construction module, configured to construct a linear correction model based on the position data of the multispectral sensor and the distortion of the multispectral remote sensing image; the linear correction model represents the residual value between the true elevation and the elevation of the image points in different multispectral remote sensing images of the multispectral sensor;

[0026] A projection image point acquisition module, configured to linearize the original error equation corresponding to the linear correction model to obtain the projection image points;

[0027] A calibration error equation acquisition module, configured to update the calibration parameters of the original error equation according to the projected image points, and determine a calibration error equation according to the correction values of the calibration parameters;

[0028] A calibration module, configured to correct the geometric error of the multi-spectral remote sensing image through the calibration error equation.

[0029] In a third aspect, in an embodiment, the present application provides a computer device, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the steps in the method embodiments of the first aspect are implemented.

[0030] In a fourth aspect, in an embodiment, the present application provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps in the method embodiments of the first aspect are implemented.

[0031] In a fifth aspect, in an embodiment, the present application provides a computer program product, including a computer program. When the computer program is executed by a processor, the steps in the method embodiments of the first aspect are implemented.

[0032] The above error correction method, device, computer device, computer-readable storage medium, and computer program product can construct a corresponding linear calibration model according to the position data of the multi-spectral sensor and the distortion of the multi-spectral remote sensing image. After linearizing the original error equation corresponding to the linear calibration model, projected image points can be obtained. Then, according to the projected image points, the calibration parameters of the original error equation can be further iteratively updated. When the calibration parameters meet the preset numerical conditions, a calibration error equation can be further determined based on the current calibration parameters and the original error equation. The geometric error of the multi-spectral remote sensing image can be corrected through the calibration error equation. Through the above method, the present application can effectively correct the geometric distortion of the multi-spectral remote sensing image. Description of the Drawings

[0033] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following will briefly introduce the drawings required for the description of the embodiments of the present application or related technologies. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts.

[0034] Figure 1 It is a schematic flowchart of an error correction method in an embodiment;

[0035] Figure 2 It is a schematic flowchart of constructing a linear calibration model in an embodiment;

[0036] Figure 3 Schematic flow chart for determining calibration error equation in an embodiment;

[0037] Figure 4 Multispectral remote sensing image before geometric calibration in an embodiment;

[0038] Figure 5 Multispectral remote sensing image after geometric calibration in an embodiment;

[0039] Figure 6 Structural block diagram of an error correction device in an embodiment;

[0040] Figure 7 Internal structure diagram of a computer device in an embodiment. Specific embodiments

[0041] In order to make the objectives, technical solutions and advantages of the present application clearer and more understandable, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0042] For the convenience of understanding the technical solutions of the present invention, first, the technical feature proper nouns that may appear in the embodiments of the present application are explained:

[0043] Geometric calibration: refers to the work of correcting the geometric distortion of an image. There are two types of geometric distortions: (1) One type is the distortion caused by the non-ideal structure performance of the remote sensor itself or the deviation of its indicators from the nominal values. For example, the calibrated principal distance of the camera is not equal to the actual principal distance, the optical distortion of the objective lens system, the non-linearity of the scanning motion of the scanning type remote sensor, etc. (2) One type is the distortion caused by the position, attitude of the remote sensor and the target object. For example, the change of the height and attitude angle of the remote sensor, atmospheric refraction, earth rotation, earth curvature, map projection, terrain undulation, etc. The principle of geometric calibration is to transform each element of the distorted image to the corresponding position of the selected projected image.

[0044] Spectrum: The pattern formed by the sequential arrangement of the wavelengths (or wave numbers) after a polychromatic light is dispersed by a dispersion system (such as a grating or a prism). For example, after sunlight is dispersed by a triangular prism, a continuous color spectrum is formed in the order of red, orange, yellow, green, cyan, blue, and purple, which corresponds to the visible light with wavelengths of 0.78 - 0.39 micrometers.

[0045] Laser point cloud: A set of points obtained by using a lidar device to scan the ground to obtain the three-dimensional coordinates of the ground reflection points and distributed in a three-dimensional space in a discrete, random and irregular manner, which is the point cloud result after classification processing.

[0046] Visible light data: Visible light data refers to visible light photos and video data obtained by using sensor devices for shooting.

[0047] In one embodiment, as Figure 1 shown, an error correction method is provided. In this embodiment, taking the application of this method to a server as an example, it can be understood that this method can also be applied to a terminal, and can also be applied to a system including a terminal and a server, and is implemented through the interaction between the terminal and the server. In this embodiment, the method includes the following steps S102 to step S108. Among them:

[0048] Step S102, based on the position data of the multispectral sensor and the distortion of the multispectral remote sensing image, construct a linear correction model.

[0049] Among them, the multispectral sensor can be set on carriers such as unmanned aerial vehicles, airplanes, and satellites, and can be used to collect multispectral remote sensing images within the field of view, and can also obtain stereo images of the same track or different tracks, so as to realize the three-dimensional coordinate calculation of ground targets and provide a data basis for spatial geolocation. The linear correction model can be used to characterize the residual difference between the true elevation and the elevation of the image points in different multispectral remote sensing images.

[0050] It can be understood that the uncorrected multispectral remote sensing image has geometric errors (or geometric distortions). Exemplarily, the distortion of the multispectral remote sensing image can include internal distortion caused by differences in the performance of the multispectral sensor, and external distortion caused by the attitude changes of the carrier (such as yaw, pitch, roll) and the target object.

[0051] Optionally, the internal distortion of the multispectral remote sensing image distortion can include scale distortion, center shift distortion, scanning nonlinear distortion, radial distortion, orthogonal distortion distortion, etc.; the external distortion can include projection distortion caused by attitude tilt, distortion caused by inconsistent scales due to changes in flight altitude, distortion caused by the target object itself (such as terrain undulation), etc. It can be understood that the above geometric distortions need to be corrected accordingly. For example, the distortion caused by the terrain needs to be corrected point by point, while the distortion caused by the earth's curvature requires more than 2 high-order equation transformations to be effectively corrected.

[0052] Specifically, the server can construct a corresponding linear correction model according to the position data of the multispectral sensor configured on the transportation vehicle (such as an unmanned aerial vehicle, etc.) and the distortion of the multispectral remote sensing image.

[0053] Step S104, linearize the original error equation corresponding to the linear correction model to obtain the projected image points.

[0054] Among them, the multispectral remote sensing image is a data source for space-based earth observation. Earth space information can be extracted from the image and projected in a fixed spatial reference system. In this way, the geometric distortions existing in the original image can be corrected, and geometric measurements and complex analyses of the image information can also be carried out. The projected image point represents the image-side point after the multispectral remote sensing image is projected in the fixed spatial reference system.

[0055] Specifically, by linearizing the linear correction model, the projected image points corresponding to the multispectral remote sensing image can be obtained.

[0056] Step S106: According to the projected image points, iteratively update the calibration parameters of the original error equation, and determine the correction error equation based on the correction values of the calibration parameters in each iteration.

[0057] Exemplarily, the calibration parameters may include parameters such as distortion correction parameters, image point correction parameters, and geographical longitude and latitude of object points corresponding to the original error equation. It can be understood that when the correction values of the calibration parameters in this round of iteration meet the preset threshold conditions, the correction error equation that can be used to correct geometric distortions can be determined based on the current calibration parameters.

[0058] Specifically, the server iteratively updates the calibration parameters of the original error equation according to the projected image points using a preset iterative algorithm, and determines the correction error equation based on the correction values of the calibration parameters in each iteration. Optionally, when the correction value of the calibration parameter in this round of iteration is less than the preset value, the correction error equation that can be used to correct geometric distortions can be determined based on the current calibration parameters.

[0059] Step S108: Correct the geometric error of the multispectral remote sensing image through the correction error equation.

[0060] Among them, the correction error equation can be used to correct the internal and external distortions existing in the multispectral remote sensing image.

[0061] Specifically, the server can perform image processing on the multispectral remote sensing image collected by the multispectral sensor according to the iteratively obtained correction error equation, so as to correct the geometric error existing in the multispectral remote sensing image.

[0062] The above error correction method can construct a corresponding linear correction model according to the position data of the multispectral sensor and the distortion of the multispectral remote sensing image. After linearizing the original error equation corresponding to the linear correction model, the projected image points can be obtained. Then, according to the projected image points, the calibration parameters of the original error equation can be further iteratively updated. When the calibration parameters meet the preset numerical conditions, the correction error equation can be further determined based on the current calibration parameters and the original error equation. This application effectively corrects geometric errors such as internal and external distortions of the multispectral remote sensing image through the constructed correction error equation.

[0063] In one embodiment, as Figure 2 shown, the position data of the multispectral sensor includes the sensor position and the unknown ground position; based on the position data of the multispectral sensor and the distortion of the multispectral remote sensing image, a linear correction model is constructed, including the following steps S202 to step S206. Among them:

[0064] Step S202, according to the sensor position, the unknown ground position and the distortion of the multispectral remote sensing image, combined with the rotation matrix and the offset matrix, establish an original correction model; the offset matrix is used to compensate for the exterior orientation error.

[0065] Exemplarily, according to the following formula 1, an original correction model can be established according to the sensor position, the unknown ground position and the distortion of the multispectral remote sensing image, combined with the rotation matrix and the offset matrix:

[0066] (Formula 1)

[0067] Among them, represents the sensor position of the multispectral sensor, represents the rotation matrix, represents the direction of the incident beam of the multispectral sensor, is the unknown scale factor, is the unknown ground position in the multispectral remote sensing image, represents the offset matrix for compensating the exterior orientation error of the multispectral remote sensing image, represents the geometric distortion of the multispectral remote sensing image.

[0068] Specifically, the server can construct an original correction model according to the sensor position of the multispectral sensor, the rotation matrix, the beam direction, the unknown scale factor, the unknown ground position, the offset matrix for compensating the exterior orientation error, and the distortion of the multispectral remote sensing image.

[0069] Step S204, based on the rotation parameters in the body coordinate system of the multispectral sensor, expand the offset matrix.

[0070] Exemplarily, according to the following formula 2, expand the offset matrix:

[0071] (Formula 2)

[0072] Among them, , , and are the rotation parameters in the body coordinate system of the multispectral sensor. It should be noted that different multispectral remote sensing images have different values.

[0073] Specifically, the server can further expand the offset matrix according to the rotation parameters in the body coordinate system of the multispectral sensor.

[0074] Step S206: Construct a linear correction model based on the expanded offset matrix and in combination with the original correction model.

[0075] Exemplarily, according to Equation 3 below, based on the original correction model constructed above, the distortion of the multispectral remote sensing image can be obtained:

[0076] (Equation 3)

[0077] where the variables and are distortion correction parameters, which can be used to describe the corresponding geometric distortion; s is the off-track image coordinate; i represents the image point number of the k-th ground point (control point or tie point) on the j-th image. It should be noted that the multispectral remote sensing images collected at different times have the same distortion.

[0078] Furthermore, according to Equation 4 below, based on the above-mentioned distortion of the remote sensing image, the original correction model can be further transformed into the first correction model shown below:

[0079] (Equation 4)

[0080] where , , , and are the correction parameters to be solved (i.e., calibration parameters) of the original correction model, and the variable represents the unknown ground position.

[0081] Optionally, when the unknown ground position is the ground position in the geocentric Cartesian coordinate system, according to Equation 5 below, the unknown ground position can be further transformed:

[0082] (Equation 5)

[0083] where represents the radius of curvature of the prime meridian with respect to latitude, represents the first eccentricity of the Earth, and L is the geodetic longitude corresponding to the unknown ground position.

[0084] Furthermore, according to Equation 6 below, based on the above-transformed unknown ground position, the first correction model can be further transformed into the second correction model shown below:

[0085] (Equation 6)

[0086] Optionally, according to the following formula 7, when the elevation is known, the second correction model can be further converted into a linear correction model:

[0087] (Formula 7)

[0088] It can be understood that the above linear correction model is a model for geometric calibration, which can represent the residual value between the true elevation and the elevation of the image point in different multispectral remote sensing images. Exemplarily, the ellipsoidal elevation in the plain area can be assumed to be a constant, and introducing this constant into the linear correction model can obtain a strict planar structure.

[0089] In one embodiment, linearizing the original error equation corresponding to the linear correction model to obtain the projected image point includes the following steps:

[0090] Determine the projected image point according to the distortion correction parameter, image point correction parameter, and geographic longitude and latitude of the object point of the linear correction model.

[0091] Specifically, the server can further obtain the corresponding projected image point according to the distortion correction parameter, image point correction parameter, and geographic longitude and latitude of the object point of the linear correction model.

[0092] Exemplarily, if the linear correction model is the linear correction model shown in the above formula 7, then the linear correction model is differentiated with respect to the parameters , , , , , B, and L, and the error equation of the object point k of the linear correction model (i.e., the original error equation) is linearized. Through the following formula 8, the projected image point (i.e., the projected image point) j can be obtained:

[0093] (Formula 8)

[0094] Among them, represents the x coordinate of the projected image point j, represents the y coordinate of the projected image point j, represents the x angle error value obtained by linearizing the error equation after differentiating with respect to L, represents the y angle error value obtained by linearizing the error equation after differentiating with respect to L, and L is the residual vector of the image point coordinates.

[0095] Furthermore, according to the following formula 9, the above projected image point can be further simplified:

[0096] (Formula 9)

[0097] Among them, Represents the correction value of the distortion correction parameter, Represents the correction value of the translation matrix of the correction parameter for image point j, Represents the object point The correction value of the geographical longitude and latitude.

[0098] In one embodiment, as Figure 3 shown, according to the projected image point, update the calibration parameters of the original error equation, and determine the correction error equation according to the correction value of the calibration parameter, including the following steps S302 to step S306. Wherein:

[0099] Step S302, according to the projected image point, determine the normal equation of the original error equation.

[0100] Specifically, the server can further determine the normal equation of the original error equation based on the projected image point obtained above.

[0101] Exemplarily, if the projected image point is the projected image point as shown in the above formula 9, then according to the following formula 10, the normal equation corresponding to the original error equation can be determined based on the projected image point:

[0102] (Formula 10)

[0103] Where, A represents the transposed matrix of the affine transformation parameter and the connecting point object coordinate coefficient, D represents the geometric distortion matrix in the least squares solution, X represents the correction value of the longitude and latitude of the image point, K represents the constant vector, and t represents the correction value of the correction parameter of the image point.

[0104] Furthermore, according to the following formula 11, the above normal equation can be simplified:

[0105] (Formula 11)

[0106] Where, N represents and The numerical simplification marking method of the coefficient matrix of the error equation, X represents the correction value of the longitude and latitude of the image point, and K represents the constant vector.

[0107] Step S304, based on the correction value of the geographical longitude and latitude of the object point, through a preset iterative algorithm, iteratively update the calibration parameters, and obtain the correction value of the calibration parameter for each round of iteration.

[0108] In some examples, the calibration parameters may include the calibration parameters to be solved in the original calibration model, such as 、 、 、 and etc.

[0109] Exemplarily, according to the following Formula 12, based on the correction value of the geographical longitude and latitude of the object point, the above normal equation can be further transformed:

[0110] (Formula 12)

[0111] It can be understood that the strong correlation between calibration parameters will lead to an ill-conditioned equation, and the embodiment of the present application can solve this problem by adopting a preset iterative algorithm.

[0112] Specifically, the server can iteratively update the calibration parameters based on the above normal equation through a preset iterative algorithm, and obtain the correction value of the calibration parameters for each round of iteration.

[0113] Step S306, when the correction value of the calibration parameters in this round of iteration meets the preset numerical condition, determine the correction error equation according to the current calibration parameters, in combination with the number of multi-spectral remote sensing images collected by the multi-spectral sensor and the number of ground points.

[0114] Wherein, the ground points represent the intersection points of different multi-spectral remote sensing images collected.

[0115] Specifically, if the correction value of the calibration parameters in this round of iteration is less than the preset value, the final calibration parameters can be determined, and then the correction error equation can be determined according to the number of multi-spectral remote sensing images collected by the multi-spectral sensor and the number of ground points.

[0116] Exemplarily, as shown in the following Formula 13, for different object points and image points, the correction error equation can be expressed as:

[0117] (Formula 13)

[0118] Wherein, m represents the number of multi-spectral remote sensing images, and m > 1; n represents the number of ground points, and the ground points are also the corresponding intersection points between different multi-spectral remote sensing images.

[0119] Furthermore, for convenience, as shown in Formula 14, the above Formula 13 can be further simplified to:

[0120] (Formula 14)

[0121] (Formula 15)

[0122] (Formula 16)

[0123] Wherein, t represents the correction value of the image point correction parameter, X represents the correction value of the object point longitude and latitude, A and D are the coefficient matrices of the correction error equation, and K is a constant vector.

[0124] In one embodiment, the preset iterative algorithm includes the iterative correction eigenvalue method.

[0125] Specifically, the server can iteratively update the calibration parameters of the original error equation through the iterative correction eigenvalue method, and obtain the correction values of the calibration parameters for each round of iteration. If the correction value of the calibration parameter is less than the preset value during the iteration process, the calibration parameter corresponding to the correction error equation can be determined. Then, according to the number of multi-spectral remote sensing images collected by the multi-spectral sensor and the number of ground points, the correction error equation can be finally determined.

[0126] In one embodiment, the geometric error of the multi-spectral remote sensing image is corrected through the correction error equation, including:

[0127] Through the correction error equation, the heterogeneity in the multi-spectral remote sensing image dataset collected by the multi-spectral sensor is eliminated, so as to assimilate the spatial dataset of the multi-spectral remote sensing image.

[0128] Among them, the spatial dataset is formed after the geometric combination of the multi-spectral remote sensing images.

[0129] Specifically, based on the obtained correction error equation, the server can eliminate the heterogeneity in the multi-spectral remote sensing image dataset collected by the multi-spectral sensor, thus solving the geometric consistency problem of the spatial dataset formed after the geometric combination of the multi-spectral remote sensing images, and finally realizing the assimilation of the dataset.

[0130] In some examples, such as Figure 4 and Figure 5 shown, Figure 4 The multi-spectral remote sensing image shown was taken in a suburb of Shanghai at 23:00 at night. After the area was photographed by the multi-spectral remote sensing image based on scale distortion, the global accuracy corresponding to the point cloud classification result reached 92%. The accuracies corresponding to the eight types of data in the multi-spectral remote sensing image dataset were respectively: artificial terrain 91.5%, natural terrain 75.6%, high vegetation 78.3%, low vegetation 71.7%, buildings 94.4%, artificial landscape 56.8%, scanned artificial sculptures 52.9%, and cars 88.4%. Due to the certain scale distortion of the multi-spectral remote sensing image, the geometric distortion correction can be performed on the multi-spectral remote sensing image to be corrected through the correction error equation (or the linear correction model corresponding to the correction error equation) obtained in the above embodiment. The corrected multi-spectral remote sensing image is as shown in Figure 5 shown. It can be understood that the image correction process of Figures 4 to 5 actually undergoes a certain translation correction and scale correction. In this application, the geometric correction of the above scale distortion is realized by constructing a correction error equation.

[0131] In addition, the calibration error equation constructed in this application can also be applied to solve the following problems: ① It can solve the registration problem between the airborne multi-spectral images of the UAV and the lidar point cloud data; ② It can correct the deviation between the captured image and the real image caused by the difference in sensor performance; ③ It can remove the geometric errors on the image and obtain the required orthophoto or approximate orthophoto; ④ It can solve the problems such as sparse point cloud data, limited point cloud information volume, and disordered point cloud data caused by the difference in sensor coordinate systems.

[0132] In a possible implementation, the overall deformation of the multi-spectral remote sensing image can be regarded as the combined result of the translation, scaling, rotation, affine, skew, bending, and higher-order deformations of the image. It can be understood that it is difficult to describe the above deformations with a strict and unified mathematical expression, but the coordinate relationship between the corresponding points of the multi-spectral remote sensing image before and after correction can be described by an appropriate empirical model, such as polynomial transformation. The above empirical model can adopt 2D or 3D polynomial empirical models and 3D rational function models. Among them, the 2D or 3D polynomial empirical model can be applied to the situation where the acquisition system parameters or the strict three-dimensional imaging model are not available. Since it cannot reflect the source of image distortion, the 2D or 3D polynomial empirical model can do without any prior information of the imaging system (platform, sensor, earth, and map projection, etc.).

[0133] Exemplarily, for the 2D polynomial empirical model, the first-order polynomial function (6 coefficients) only allows correcting translation, rotation, scaling, and tilt; the second-order polynomial function (12 coefficients) allows correcting the image distortion corresponding to the previous parameters in addition to the previous ones, including torsion and convexity; the third-order polynomial function (20 coefficients) allows correcting the same distortion as the second-order polynomial function, and it does not necessarily correspond to any physically real image acquisition system.

[0134] It can be understood that for the 2D polynomial empirical model, it cannot correct the image point offset caused by terrain undulation, and is limited to correcting less or smaller image distortions, such as sub-satellite view images, images corrected by the system, and small images in flat areas, etc. The 2D polynomial empirical model can correct local distortions near the control points, so it is extremely sensitive to input errors and requires sufficient and evenly distributed ground control points.

[0135] Furthermore, as shown in formula 17 below, the matrix expression of the two-dimensional polynomial transformation corresponding to the 2D polynomial empirical model is:

[0136] (Formula 17)

[0137] Among them, is the image point coordinate; is the terrain or map coordinate; , is an integer increment; and is a preset integer value, generally ranging from 0 to 3; and are polynomial coefficients.

[0138] Exemplarily, for a 3D polynomial empirical model, as shown in Formula 18 below, the matrix expression of the three-dimensional polynomial transformation corresponding to the empirical model is:

[0139] (Formula 18)

[0140] where is the image point coordinate; is the terrain or map coordinate; and and are integer increments; and and are integer values, generally ranging from 0 to 3; and and are polynomial coefficients.

[0141] It can be understood that the 3D polynomial empirical model has the same problems as other empirical functions, and it cannot correct the image distortion caused by terrain undulation. Moreover, it requires sufficient and uniformly distributed control points and is very sensitive to the introduced errors. In the actual environment, the 3D polynomial empirical model often lacks robustness and consistency.

[0142] Exemplarily, the 3D rational function model can be used to approximately represent a solved three-dimensional physical model, or generally use Ground Control Points (GCPs) to calculate the unknowns of the polynomial function. However, there are also some disadvantages, such as being unable to correct local deformation, being limited by the image size, lacking physical meaning, possible denominator being zero, and strong correlation of polynomial coefficients.

[0143] In some examples, as shown in Formula 19, the matrix expression of the three-dimensional rational function transformation corresponding to the 3D rational function model is:

[0144] (Formula 19)

[0145] where is the image point coordinate; is the terrain or map coordinate; , , are integer increments; , and is an integer value, generally ranging from 0 to 3; , , , are rational polynomial coefficients. For the above problems, rational functions are applicable to geolocating systematically corrected images, rather than raw data without any correction. Because there are still errors after using rational function correction, a small number of GCPs are needed for post-processing or the coefficients of the original rational polynomial function are refined using a linear equation with high-precision GCPs. It should be noted that the 3D rational polynomial coefficients here are generally provided by the image distributor or obtained by deriving from a strict physical model. The method of using the 3D rational function model is also called the terrain-independent method. In addition, if the 3D rational polynomial coefficients are not provided by the image distributor or derived from a strict physical model, 20, 40, or 80 GCPs can also be used to solve the 3D rational polynomial coefficients of the 1st, 2nd, and 3rd orders respectively. Using this method cannot completely eliminate the local deformation between control points. However, the latter method of solving coefficients through GCPs is more flexible for correcting medium- and high-resolution images. Since it depends on terrain undulation, the number and distribution of control points, it is suitable for multi-spectral remote sensing images in flat areas.

[0146] It should be understood that although the steps in the flowcharts involved in the above-described embodiments are shown in sequence according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear description in this article, the execution of these steps has no strict order limit, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-described embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily executed at the same moment, but can be executed at different moments. The execution order of these steps or stages is not necessarily sequential, but can be executed alternately or in turn with at least a part of other steps or steps or stages in other steps.

[0147] Based on the same inventive concept, an embodiment of the present application also provides an error correction device for implementing the above-mentioned error correction method. The solution provided by this device to solve the problem is similar to the solution described in the above method. Therefore, the specific limitations in one or more embodiments of the error correction device provided below can refer to the limitations on the error correction method in the above text, and will not be repeated here.

[0148] In an exemplary embodiment, as Figure 6 shown, an error correction device 600 is provided. The device 600 includes:

[0149] A model construction module 602, configured to construct a linear correction model based on the position data of the multispectral sensor and the distortion of the multispectral remote sensing image; the linear correction model represents the residual value between the true elevation and the elevation of the image points in different multispectral remote sensing images.

[0150] A projected image point acquisition module 604, configured to linearize the original error equation corresponding to the linear correction model to obtain projected image points.

[0151] A calibration error equation acquisition module 606, configured to update the calibration parameters of the original error equation according to the projected image points, and determine a calibration error equation according to the correction value of the calibration parameters.

[0152] A correction module 608, configured to correct the geometric error of the multispectral remote sensing image through the calibration error equation.

[0153] In one embodiment, the position data of the multispectral sensor includes the sensor position and the unknown ground position; the model construction module 602 is further configured to:

[0154] According to the sensor position, the unknown ground position, and the distortion of the multispectral remote sensing image, combined with the rotation matrix and the offset matrix, establish an original correction model; the offset matrix is used to compensate for the exterior orientation error.

[0155] Based on the rotation parameters in the body coordinate system of the multispectral sensor, expand the offset matrix.

[0156] According to the expanded offset matrix, combined with the original correction model, construct a linear correction model.

[0157] In one embodiment, the projected image point acquisition module 604 is further configured to:

[0158] Determine the projected image points according to the distortion correction parameters, the image point correction parameters, and the geographical longitude and latitude of the object point of the linear correction model.

[0159] In one embodiment, the calibration error equation acquisition module 606 is further configured to:

[0160] Determine the normal equation of the original error equation according to the projected image points.

[0161] Based on the correction value of the geographical longitude and latitude of the object point, through a preset iterative algorithm, iteratively update the calibration parameters, and obtain the correction value of the calibration parameters for each round of iteration.

[0162] When the correction value of the calibration parameters in this round of iteration meets the preset numerical conditions, according to the current calibration parameters, combined with the number of multispectral remote sensing images collected by the multispectral sensor and the number of ground points, determine the calibration error equation; wherein, the ground points represent the intersection points of different multispectral remote sensing images collected.

[0163] In one embodiment, the preset iterative algorithm includes the iterative correction eigenvalue method.

[0164] In one embodiment, the correction module 608 is further configured to:

[0165] Eliminate the heterogeneity in the multi-spectral remote sensing image data set collected by the multi-spectral sensor through the correction error equation, so as to assimilate the spatial data set of the multi-spectral remote sensing image; wherein, the spatial data set is formed after geometric combination of the multi-spectral remote sensing image.

[0166] Each module in the above error correction device can be implemented in whole or in part by software, hardware, and their combination. The above modules can be embedded in the processor of the computer device in hardware form or independent of it, or stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to the above modules.

[0167] In an exemplary embodiment, a computer device is provided. The computer device can be a server, and its internal structure diagram can be as Figure 7 shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, abbreviated as I / O), and a communication interface. Among them, the processor, the memory, and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the position data of the multi-spectral sensor and the multi-spectral remote sensing image data, etc. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals through a network connection. The computer program, when executed by the processor, implements an error correction method.

[0168] Those skilled in the art can understand that Figure 7 the structure shown in is only a block diagram of some structures related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0169] In one embodiment, a computer device is further provided, including a memory and a processor. A computer program is stored in the memory, and when the processor executes the computer program, the steps in the above method embodiments are implemented.

[0170] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the foregoing method embodiments are implemented.

[0171] In one embodiment, a computer program product is provided, including a computer program. When the computer program is executed by a processor, the steps in the foregoing method embodiments are implemented.

[0172] It should be noted that the image information and data involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use, and processing of relevant data need to comply with relevant regulations.

[0173] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, database, or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The databases involved in the embodiments provided in the present application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided in the present application can be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, data processing logics based on quantum computing, artificial intelligence (AI) processors, etc., without limitation.

[0174] The technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in the present application.

[0175] The above-described embodiments merely represent several implementation manners of the present application. The description thereof is relatively specific and detailed, but it should not be construed as a limitation on the patent scope of the present application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all fall within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the appended claims.

Claims

1. An error correction method, characterized in that: The method comprises: Based on the position data of the multispectral sensor and the distortion of the multispectral remote sensing image, a linear correction model is constructed; the linear correction model represents the residual value between the true elevation and the elevation of the image points in different multispectral remote sensing images; Linearizing the original error equation corresponding to the linear correction model to obtain a projection image point; Iteratively updating the calibration parameters of the original error equation according to the projected image points, and determining the correction error equation according to the correction values ​​of the calibration parameters in each round of iteration; The geometric error of the multispectral remote sensing image is corrected by the correction error equation.

2. The method according to claim 1, characterized in that The position data of the multispectral sensor includes the sensor position and the unknown ground position; the linear correction model is constructed based on the position data of the multispectral sensor and the multispectral remote sensing image distortion, including: According to the sensor position, the unknown ground position and the multispectral remote sensing image distortion, a primitive correction model is established in combination with a rotation matrix and an offset matrix; the offset matrix is ​​used to compensate for the exterior orientation error; Expanding the offset matrix based on a rotation parameter in a body coordinate system of the multispectral sensor; The linear correction model is constructed according to the expanded offset matrix in combination with the original correction model.

3. The method according to claim 1, characterized in that The linearizing the original error equation corresponding to the linear correction model to obtain the projection image point includes: The projected image point is determined according to the distortion correction parameters of the linear correction model, the image point correction parameters and the geographical longitude and latitude of the object point.

4. The method according to claim 3, characterized in that The iterative updating of the calibration parameters of the original error equation according to the projection image points, and determining the correction error equation according to the correction values ​​of the calibration parameters in each round of iteration, comprises: Determining a normal equation of the original error equation according to the projected image point; Based on the correction values ​​of the geographic longitude and latitude of the object point, the calibration parameters are iteratively updated through a preset iterative algorithm, and the correction values ​​of the calibration parameters in each round of iteration are obtained; When the correction value of the calibration parameter meets the preset numerical condition in this round of iteration, the correction error equation is determined according to the current calibration parameter, combined with the number of multispectral remote sensing images collected by the multispectral sensor and the number of ground points; wherein the ground point represents the intersection of different multispectral remote sensing images collected.

5. The method according to claim 4, characterized in that The preset iterative algorithm includes an iterative correction characteristic value method.

6. The method according to any one of claims 1 to 5, characterized in that: The method of correcting the geometric error of the multispectral remote sensing image by using the correction error equation includes: The correction error equation is used to eliminate heterogeneity in the multispectral remote sensing image data set acquired by the multispectral sensor, so as to assimilate the spatial data set of the multispectral remote sensing image; wherein the spatial data set is formed by geometrically combining the multispectral remote sensing images.

7. An error correction device, characterized in that: The device comprises: A model building module is used to build a linear correction model based on the position data of the multispectral sensor and the multispectral remote sensing image distortion; the linear correction model represents the residual value of the true elevation and the elevation of the image point in different multispectral remote sensing images of the multispectral sensor; A projection image point acquisition module, used to linearize the original error equation corresponding to the linear correction model to obtain projection image points; A correction error equation acquisition module, used to obtain calibration parameters of the original error equation according to the projection image points, and determine the correction error equation according to the correction values ​​of the calibration parameters; The correction module is used to correct the geometric error of the multispectral remote sensing image through the correction error equation.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images