Unmanned aerial vehicle positioning method and system based on manifold optimization

Abstract

The invention discloses an unmanned aerial vehicle positioning method and system based on manifold optimization, and the method comprises the steps: building a pose calculation model based on manifold optimization based on a principle that the projection center of a camera carried by an unmanned aerial vehicle, the coordinates of an ith control point and the coordinates of a corresponding image point are collinear, and enabling the camera carried by the unmanned aerial vehicle to shoot a picture, an industrial personal computer carried by the unmanned aerial vehicle automatically selects a plurality of control points in the picture, coordinate information of the control points is substituted into the resolving model, the position and the posture of the unmanned aerial vehicle in the picture shooting process are obtained through resolving of the resolving model, and the resolving model comprises two contents including a rotation matrix confirmation method and a spatial position confirmation method. And determining two parameters of the rotation matrix and the spatial position by adopting an alternative minimization mode until a set termination criterion is met, and obtaining the rotation matrix and the spatial position which are the attitude and the position of the unmanned aerial vehicle respectively. According to the method, the positioning precision and robustness are improved, and the method has relatively high engineering applicability.

Description

Technical Field

[0001] The present invention belongs to the technical field of computer vision and unmanned system navigation and positioning, and specifically relates to a UAV positioning method and system based on manifold optimization. Background Art

[0002] Drones are typically equipped with cameras. Camera positioning is achieved by calculating the camera's exterior orientation elements. This is a step in the analysis of a single image and is fundamental to photogrammetry. Its goal is to determine the camera's exterior orientation elements, namely its position and attitude, using ground control points in the captured space. Spatial resection is based on the image-based collinearity equation. This equation uses a number of global control points with known coordinates in the captured space of a camera with known intrinsic parameters. The collinearity constraint equation is then established by combining the image point coordinates with the control point coordinates. The least-squares principle is then used to iteratively solve the collinearity constraint equation. The camera's extrinsic parameters in its current pose are then determined, effectively localizing the camera. Spatial resection requires minimal control point placement, has low reception costs, and can quickly determine attitude information within a confined space. Its simplicity, convenience, and flexibility have led to its widespread application in photogrammetry. Currently, the solution for camera exterior orientation elements is often analytically processed using collinearity or coplanarity equations, with the rotation matrices involved typically parameterized using Euler angles or quaternions. Traditional solution methods often linearize the collinear equations using a first-order Taylor expansion, then employ iterative optimization to solve them. However, linearized models can easily introduce large errors during the approximation process, limiting the accuracy and robustness of the solution. To overcome the inherent limitations of linearization methods, some researchers have proposed exterior orientation element solution methods based on nonlinear equations. However, these methods typically fail to explicitly consider the orthogonality constraints of the rotation matrix during the optimization process, making it difficult to guarantee a high-precision feasible solution in the rotation matrix space. In other words, current UAV positioning solution methods suffer from shortcomings such as complexity, low accuracy, and poor stability. Summary of the Invention

[0003] In response to the above-mentioned problems existing in the prior art, the purpose of the present invention is to provide a method and system for UAV positioning based on manifold optimization, so as to solve the problem that the first-order Taylor expansion of the strongly nonlinear collinear equation in the traditional method introduces a larger model error. The invention does not use the three attitude angles and positions as the parameters to be determined, and does not require the first-order Taylor expansion of the collinear equation. Instead, it directly establishes a joint optimization model with the rotation matrix and position as unknown parameters, and directly decomposes the objective function into two sub-problems for alternating optimization and solution without losing the model performance. Taking into account the orthogonal constraints of the rotation matrix and combining the characteristics of the model on the manifold structure, the manifold optimization theory is combined with the posture solution problem to design a model and solution algorithm that conforms to the geometric characteristics of the rotation matrix. The method can effectively converge in various scenarios and has the advantages of accurate calculation results, stability and reliability.

[0004] In order to achieve the above object, the technical solution adopted by the present invention is: A method for positioning a drone based on manifold optimization is proposed. Based on the principle that the projection center of the drone's camera, the coordinates of the i-th control point, and the coordinates of its corresponding image point are collinear, a pose calculation model based on manifold optimization is established. The drone's camera takes a picture, and the drone's industrial computer automatically selects multiple control points in the picture and substitutes the coordinate information of the control points into the calculation model. The position and pose of the drone when taking the picture are obtained by solving the calculation model. The calculation model includes two contents: a method for confirming the rotation matrix and a method for confirming the spatial position. The two parameters, the rotation matrix and the spatial position, are confirmed by an alternating minimization method until the set termination criterion is met. The obtained rotation matrix and spatial position are the pose and position of the drone, respectively.

[0005] As a further improvement of the above technical solution: The objective function of the established pose solution model contains two variables to be optimized: the spatial position and the rotation matrix. The spatial position represents the three-dimensional coordinates of the camera projection center in the ground coordinate system, and the rotation matrix represents the three-dimensional rotation relationship between the image space coordinate system and the auxiliary coordinate system. By alternately solving the spatial position and rotation matrix, a high-precision pose estimation is achieved. In other words, the rotation matrix represents the matrix of the rotation relationship between the image space coordinate system and the image space auxiliary coordinate system. Because the image space auxiliary coordinate system and the ground coordinate system are parallel, the rotation matrix can also be expressed as the matrix of the rotation relationship between the image space coordinate system and the ground coordinate system.

[0006] The method for alternating minimization of spatial position and rotation matrix is ​​as follows: fix the spatial position of one of the variables to the current value as the first variable, optimize the objective function with respect to the rotation matrix of the other variable, and obtain the optimal value of the variable as the second variable; then fix the second variable to the current optimal value, optimize the first variable, and iterate the above steps until the preset termination criteria are met.

[0007] The rotation matrix confirmation method adopts the Riemann conjugate gradient algorithm based on manifold optimization to achieve direct solution of the rotation matrix, including: using Cayley transformation to map variables in Euclidean space to the manifold space where the rotation matrix is ​​located; calculating the gradient of the objective function on the manifold through the gradient information in the Euclidean space; combining the conjugate direction update strategy with the vector transmission operator to determine the search direction on the manifold; selecting a suitable step size according to the line search criterion; and using the contraction mapping to project the update result back to the manifold space to obtain a new rotation matrix estimate, thereby completing an iterative update.

[0008] Determine whether the gradient value on the manifold space is less than the set value. If so, solve the spatial position sub-problem. At this time, the estimated value of the rotation matrix is ; If not, continue iterating until the judgment condition is met.

[0009] The method to confirm the spatial position is: let the spatial position be , in order to avoid obtaining The exact solution causes the algorithm to jump out of the loop early, and a proximity term is added to the objective function when calculating the spatial position, because it contains and , it cannot be solved directly, so it is converted to the classical least squares expression, and the least squares solution method is used to obtain the final closed-form solution of the spatial position .

[0010] Determine whether the estimated update values ​​of the rotation matrix and spatial position meet the termination conditions of alternating minimization. If so, terminate the iteration and obtain the final solution of the rotation matrix and spatial position. If not, continue to iterate the rotation matrix and spatial position.

[0011] A manifold optimization-based unmanned aerial vehicle (UAV) positioning system is used to implement the above-mentioned positioning method. From a hardware perspective, the positioning system includes a UAV, a camera, an industrial computer, and a wireless communication module. The camera, industrial computer, and wireless communication module are all mounted on the UAV. The solution model is implanted into the industrial computer in the form of a program. The UAV, camera, and wireless communication module are all electrically connected to the industrial computer. The camera transmits the captured image to the industrial computer. The industrial computer solves the image based on the built-in program to obtain the position and posture of the UAV when the image was taken. The wireless communication module can wirelessly transmit the information of the industrial computer to the ground control center.

[0012] From a functional perspective, the drone positioning system includes an image acquisition module, a data storage module, an image processing module, a positioning solution module and a transmission module. The image acquisition module is used to collect image information, the data storage module is used to store data, the image processing module processes the collected images to obtain control points and their coordinate information, the positioning solution module performs positioning solution based on the solution model and the input image information to obtain the position and posture of the drone, and the transmission module is used for data transmission.

[0013] The beneficial effects of the present invention are: by solving the optimal solution of the rotation matrix and spatial position, a more accurate position and attitude of the UAV is obtained, which solves the problem of introducing a larger model error in the first-order Taylor expansion of the strongly nonlinear collinear equation in the traditional method. The present invention does not use the three attitude angles and position as the parameters to be determined, and does not require the first-order Taylor expansion of the collinear equation. Instead, the present invention directly establishes a joint optimization model with the rotation matrix and position as unknown parameters, and directly decomposes the objective function into two sub-problems for alternating optimization and solution without losing model performance. Taking into account the orthogonal constraints of the rotation matrix and combining the characteristics of the model on the manifold structure, the manifold optimization theory is combined with the posture solution problem to design a model and solution algorithm that conforms to the geometric characteristics of the rotation matrix. The calculation process of this method is relatively simple, the calculation results are accurate, and the algorithm always maintains convergence. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 It is the exterior orientation element positioning diagram of the oblique photogrammetry of the present invention; Figure 2 It is a coordinate system conversion relationship diagram adopted by the present invention; Figure 3 This is a flowchart of the pose solving algorithm based on manifold alternating minimization; Figure 4 4 is a distribution diagram of control points in a photograph used in an embodiment of the present invention. DETAILED DESCRIPTION

[0015] The following describes the specific embodiments of the present invention in detail with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only used to illustrate and explain the present invention and are not intended to limit the present invention.

[0016] For ease of description, spatially relative terms such as "above," "above," "on the upper surface of," and "upper" may be used herein to describe the spatial positional relationship of a device or feature to other devices or features as shown in the figures. It should be understood that spatially relative terms are intended to encompass different orientations of the device in use or operation in addition to the orientation depicted in the figures. For example, if the device in the figures is inverted, a device described as "above" or "on top of" other devices or structures would then be positioned as "below" or "below" the other devices or structures. Thus, the exemplary term "above" can include both the "above" and "below" orientations. The device may also be positioned in other different ways (rotated 60 degrees or in other orientations), and the spatially relative descriptions used herein should be interpreted accordingly.

[0017] A method and system for positioning an unmanned aerial vehicle (UAV) based on manifold optimization, wherein the UAV positioning system is used to implement the positioning method.

[0018] From a hardware perspective, the positioning system includes a drone, a camera, an industrial computer, and a wireless communication module. The camera, industrial computer, and wireless communication module are all mounted on the drone. The industrial computer is equivalent to a computer, and the solution model (described below) is embedded in the computer as a program. The drone, camera, and wireless communication module are all electrically connected to the industrial computer. The camera transmits captured images to the industrial computer, which uses a built-in program to solve the image and determine the drone's position and attitude at the time the image was captured. The wireless communication module wirelessly transmits information from the industrial computer to a control center on the ground.

[0019] From a functional perspective, the drone positioning system includes an image acquisition module, a data storage module, an image processing module, a positioning solution module, and a transmission module. The image acquisition module is used to capture image information, a function implemented by the camera and the industrial computer program. The data storage module is used to store data. The image processing module analyzes and processes the captured images. The positioning solution module performs positioning solutions based on the solution model and the input image information to determine the drone's position and attitude. The functions of the data storage module, image processing module, and positioning solution module are all implemented by the program embedded in the industrial computer. The transmission module is used for data transmission, a function implemented by the wireless communication module and the industrial computer program.

[0020] The UAV positioning method is: Based on the principle that the camera projection center, the coordinates of the i-th control point, and the corresponding image point coordinates are collinear, a least-squares exterior orientation element solution model is established. The drone's onboard camera captures an image, and a program embedded in an industrial computer automatically selects multiple control points from the image. The coordinates of each control point are then substituted into the solution model. The solution model then calculates the drone's position and attitude at the time the image was taken.

[0021] The method and principle for selecting control point coordinates are as follows: The industrial computer contains publicly available commercial map information, which contains the geographic coordinates of each point on the map. The industrial computer pre-sets the rules for selecting control points in the image (for example, dividing the captured image into multiple grids, with the center point of each grid serving as the control point). After the camera captures the image, the image processing module automatically selects multiple control points in the image based on the pre-set rules. The image processing module then compares the control points on the image with the commercial map, locates each control point on the commercial map, and obtains the geographic coordinates of each control point. To reduce the computational complexity during the comparison, the current city, region, and other settings are pre-set when the drone is launched. This narrows the comparison scope and improves comparison efficiency and accuracy.

[0022] The solution model is described as follows:

[0023] Figure 1 It is a specific diagram for solving the exterior orientation elements. The exterior orientation elements are solved based on the conversion relationship between the photographic center, control point coordinates and image point coordinates. The specific coordinate conversion process is as follows: Figure 2 As shown, assume that the camera projection center S is marked as , No. The coordinates of the control points are , the corresponding image point coordinates are In general, due to the oblique photography of drones, the image space coordinate system and the control point coordinate system are not parallel. With the help of the image space auxiliary coordinate system, the image point coordinates are a i And the coordinates of the photography center S, the coordinates of the control point On a straight line, the image space auxiliary coordinates of the image point coordinates are .

[0024] according to Figure 1 The spatial geometric relationship between the ground control point, the corresponding image point and the projection center is collinear. The rotation relationship between the image space coordinate system and the image space auxiliary coordinate system is considered. Considering the ground fluctuation, the scale of each image point coordinate and the control point coordinate is not fixed. The scale coefficient of each control point is , , is the focal length of the camera. Therefore, the least squares-based exterior orientation element solution model is expressed as: (1)

[0025] In formula (1) Indicates the The coordinates of the visual base stations in the object space coordinate system, represents its image point coordinates, Indicates the coordinates of the camera's projection center in the ground coordinate system, that is, the spatial position; Indicates the The estimated coordinate value of the projection center of the secondary cycle camera in the ground coordinate system, represents the rotation matrix, T represents the transpose of the matrix, and the scale factor determines the mapping relationship between the image point coordinates and the actual ground coordinates, thereby achieving accurate conversion of image registration and measurement. In the above formula, st represents the constraint on minimizing the pose function. Represents the 3×3 identity matrix (diagonal elements are 1 and the rest are 0).

[0026] The solution model uses the Stiefel-Manifold-based Alternating-minimization Pose-estimation (SMAP) algorithm.

[0027] Considering the rotation matrix With orthogonal constraints, it just satisfies the geometric characteristics of manifold space. Starting from the manifold optimization thinking, we can transform the optimization problem in Euclidean space into an optimization problem in manifold space. From the manifold optimization framework, we can solve the difficulties brought by linearization of strong nonlinear equations and orthogonal constraints, and ensure the reliability of the rotation matrix solution and its feasibility. Without losing the model performance, we can directly decompose the objective function (1) into two sub-problems and perform alternating minimization (AM) to solve the exterior orientation elements (position and attitude).

[0028] The principle of the alternating minimization algorithm is to first fix a variable to a constant value, such as The variable is After that, another rotation matrix variable in the optimization objective function get ; then keep Estimated value of Unchanged, yes Optimize to get , this process constitutes a cycle in the alternating minimization process. By continuously looping until a certain termination criterion is met, the objective function reaches the optimal value and the iteration is stopped. This is the optimal solution to the original problem. Therefore, by directly optimizing the original objective function through the alternating minimization algorithm, the original problem can be cleverly decomposed into two sub-problems to be solved separately: Subproblem (rotation matrix subproblem) and Subproblem (spatial position subproblem). Among them, and Respectively and The final value of . represents the number of iterations, .

[0029] In this embodiment, first, the rotation matrix is ​​solved based on manifold optimization .

[0030] For the attitude rotation matrix subproblem To solve the problem, a Riemann conjugate gradient algorithm based on manifold optimization is designed to directly solve the rotation matrix. The rotation matrix represents the matrix of the rotation relationship between the image space coordinate system and the image space auxiliary coordinate system. The rotation matrix is ​​obtained based on the properties of Cayley transformation, contraction mapping and vector transmission mapping in manifold optimization. The gradient of the manifold space is calculated from the gradient of the Euclidean space, that is, the Riemann gradient. The search direction is determined by combining the conjugate parameter and the vector migration operator. A suitable step size is found through the line search criterion, and the contraction operator is used to map it to the manifold to obtain a new iteration point.

[0031] Determining the gradient on the manifold Is it true? If it is true, proceed to the next step. If it is not true, solve the rotation matrix. Continue iterating until the gradient judgment condition is met. Pick .

[0032] Then, find the estimated three-dimensional coordinates of the camera.

[0033] Solving the three-dimensional coordinates of the camera is the spatial position subproblem - solving the camera posture and position, that is, solving the three-dimensional coordinates of the camera No. sub-estimated objective function , in order to avoid obtaining The exact solution of causes the algorithm to jump out of the loop early, so in the calculation Add a proximity term to the objective function when solving the subproblem ; in, Is the adjacent term factor, which is the set value; because the objective function contains and It cannot be calculated directly and converted into an expression like the classic least squares Then, take its derivative and set the derivative to 0 to get the closed-form solution. The closed-form solution of in, .

[0034] Finally, judge Is it true? If it is true, terminate the iteration and get the final solution If it does not hold, jump to solving the rotation matrix Continue iterating.

[0035] The following is an example for explanation. Specifically, the calculation of the exterior orientation elements of images in high and low altitude and large inclination angle scenes is taken as an example to illustrate the specific implementation process of the present invention.

[0036] The photo is of great height , Xiaohang Gaowei , the focal length of the camera is , the inner orientation elements of the photography center are ; Nine control points are selected, and the coordinates of the control points, the corresponding image exterior orientation elements, and the image point coordinates are shown in Tables 1 to 4 respectively.

[0037] Table 1: Ground coordinates of 9 control points (unit: meters)

[0038] Table 2: Image point coordinates corresponding to the images in the high altitude simulation scene (unit: mm)

[0039] Table 3: Image point coordinates corresponding to the images in the low altitude simulation scene (unit: mm)

[0040] Table 4: Exterior orientation elements of images in various scenes

[0041] In Table 4, Represents the three attitude angles that make up the rotation matrix, namely pitch angle, roll angle and yaw angle.

[0042] a1: Substitute the corresponding coordinates of the 9 points into the model and set the initial rotation matrix is the identity matrix, is the average value of each control point, the search direction is the negative value of the gradient, and the search step size is , .

[0043] a2: Use the alternating minimization algorithm to decompose the model into two simple sub-problems: Subproblem (rotation matrix subproblem) and Subproblem (spatial location subproblem).

[0044] a3: Solving the rotation matrix subproblem: Calculating the rotation matrix using the Riemann conjugate gradient algorithm based on manifold optimization , and get the optimal solution .

[0045] a4: Determine the gradient on the manifold Is it satisfied? If not, continue iterating at a3; if satisfied, jump to a5.

[0046] a5: Solution of the spatial position subproblem: Fix the new solution in a3 , calculate the new camera's 3D coordinates .

[0047] a6: Judgment Is it true? If it is true, terminate the iteration and get If not, jump to a3 and continue iteration.

[0048] Conclusion: The comparison of the results using the least squares method (Algorithm 1), the Lebervinger-Marquardt method (Algorithm 2), and the method of the present invention are shown in Tables 5 and 6.

[0049] Table 5: Comparison of solution results for low altitude and large inclination angle scenarios

[0050] Table 6: Comparison of solution results for the large-angle scenario

[0051] Because the traditional algorithm is calculated directly based on the Euler angle, the rotation matrix calculated by the algorithm proposed in this paper is It also needs to be converted to Euler angles. This is done by using the conversion relationship between the rotation matrix and the three attitude angles.

[0052] It can be seen from Tables 5 and 6 that at different altitudes, Algorithm 1 has a large iterative solution error in extremely large tilt angle scenarios, while Algorithm 2 and the method of the present invention can effectively converge at extremely large tilt angles. The method of the present invention is more accurate than Algorithm 2. The main reason is that Algorithm 1 and Algorithm 2 use Euler angles to represent the rotation matrix and solve collinear equations containing sine and cosine, which will cause errors when the collinear equations are linearly expanded. The method provided in this application is based on the principle of collinear equations, directly expressed in terms of rotation matrices, and a camera external orientation element solution model based on the alternating minimization algorithm is established, which converts the original problem into two simple sub-problems: the rotation matrix sub-problem and the spatial position sub-problem. For the rotation matrix sub-problem, the rotation matrix direct solution algorithm based on manifold optimization designed in this application can achieve the following results: Then, by fixing one solution in a subproblem to solve the solution in another subproblem, alternately solve until a certain termination criterion is met and the objective function reaches the optimal value, then stop the iteration, and the solution is calculated. This is the optimal solution to the original problem, let The present invention ensures that the solution to the problem to be solved always maintains convergence and constraint feasibility, and has the advantages of a relatively simple calculation process, accurate calculation results, and the algorithm always maintains convergence.

[0053] Finally, it is necessary to explain here that the above examples are only used to further illustrate the technical solution of the present invention in detail and cannot be understood as limiting the scope of protection of the present invention. Some non-essential improvements and adjustments made by technicians in this field based on the above content of the present invention all fall within the scope of protection of the present invention.

Claims

1. A UAV positioning method based on manifold optimization, characterized in that: Based on the principle that the projection center of the camera carried by the drone, the coordinates of the i-th control point and its corresponding image point coordinates are collinear, a posture solution model based on manifold optimization is established. The camera carried by the drone takes a picture, and the industrial computer carried by the drone automatically selects multiple control points in the picture and substitutes the coordinate information of the control points into the solution model. The position and posture of the drone when taking the picture are obtained by solving the solution model. The solution model includes two contents: a method for confirming the rotation matrix and a method for confirming the spatial position. The two parameters of the rotation matrix and the spatial position are confirmed by alternating minimization until the set termination criteria are met. The obtained rotation matrix and spatial position are the posture and position of the drone, respectively.

2. The positioning method according to claim 1, wherein: The objective function of the established pose solution model contains two variables to be optimized: spatial position and rotation matrix. The spatial position represents the three-dimensional coordinates of the camera projection center in the ground coordinate system, and the rotation matrix represents the three-dimensional rotation relationship between the image space coordinate system and the auxiliary coordinate system. By alternately solving the spatial position and the rotation matrix, high-precision estimation of the pose can be achieved.

3. The positioning method according to claim 2, wherein: The method for alternating minimization of the rotation matrix and spatial position is as follows: fix the spatial position of one of the variables to the current value as the first variable, optimize the objective function with respect to the rotation matrix of the other variable, and obtain the optimal value of the variable as the second variable; then fix the second variable to the current optimal value, optimize the first variable, and iterate the above steps until the preset termination criteria are met.

4. The positioning method according to claim 3, wherein: The rotation matrix confirmation method adopts the Riemann conjugate gradient algorithm based on manifold optimization to directly solve the rotation matrix, including: using Cayley transformation to map variables in Euclidean space to the manifold space where the rotation matrix is ​​located; calculating the gradient of the objective function on the manifold through the gradient information in the Euclidean space; combining the conjugate direction update strategy with the vector transmission operator to determine the search direction on the manifold; selecting an appropriate step size based on the line search criterion; and using contraction mapping to project the updated result back to the manifold space to obtain a new rotation matrix estimate, thereby completing an iterative update.

5. The positioning method according to claim 4, wherein: Determine whether the gradient value on the manifold space is less than the set value. If so, solve the spatial position. At this time, the estimated value of the rotation matrix is ; If not, continue iterating until the judgment condition is met.

6. The positioning method according to claim 5, wherein: The method to confirm the spatial position is: let the spatial position be , in order to avoid obtaining The exact solution causes the algorithm to jump out of the loop early, and a proximity term is added to the objective function when calculating the spatial position, because it contains and , it cannot be solved directly, so it is converted to the classical least squares expression, and the least squares solution method is used to obtain the final closed-form solution of the spatial position .

7. The positioning method according to claim 6, wherein: Determine whether the estimated update values ​​of the rotation matrix and spatial position meet the termination conditions of alternating minimization. If so, terminate the iteration and obtain the final solution of the rotation matrix and spatial position. If not, continue to iterate the rotation matrix and spatial position.

8. A UAV positioning system based on manifold optimization, used to implement the positioning method according to any one of claims 1 to 7, characterized in that: From a hardware perspective, the positioning system includes a drone, a camera, an industrial computer, and a wireless communication module. The camera, industrial computer, and wireless communication module are all mounted on the drone. The solution model is implanted into the industrial computer in the form of a program. The drone, camera, and wireless communication module are all electrically connected to the industrial computer. The camera transmits the captured image to the industrial computer. The industrial computer solves the image based on the built-in program to obtain the position and posture of the drone when the image was taken. The wireless communication module can wirelessly transmit the information of the industrial computer to the ground control center.

9. The positioning system according to claim 8, wherein: From a functional perspective, the drone positioning system includes an image acquisition module, a data storage module, an image processing module, a positioning solution module and a transmission module. The image acquisition module is used to collect image information, the data storage module is used to store data, the image processing module processes the collected images to obtain control points and their coordinate information, the positioning solution module performs positioning solution based on the solution model and the input image information to obtain the position and posture of the drone, and the transmission module is used for data transmission.

Images