Method, apparatus, and device for determining error range

Abstract

The present application provides a method, apparatus and device for determining an error range. The method includes: obtaining at least four 3D-2D point pair coordinates, where the 3D-2D point pair coordinates are the coordinates corresponding to any point in the world coordinate system and the coordinates corresponding to the point in the pixel coordinate system; determining the uncertainty of the positioning result to be evaluated and the translation vector between the camera coordinate system and the world coordinate system according to at least four 3D-2D point pair coordinates and the n-point perspective positioning model; determining the error range corresponding to each axis of the world coordinate system of the positioning result to be evaluated according to the uncertainty of the translation vector; and outputting the error range. By this method, the process of determining the error range can be simplified, and the efficiency of determining the error range can be improved.

Description

Technical Field

[0001] The embodiments of the present application relate to the technical field of data processing, and in particular, to a method, device, and equipment for determining an error range. Background Art

[0002] Visual positioning technology is one of the very crucial technologies in autonomous driving technology. Among them, the monocular vision positioning algorithm based on the high-precision map technology (Map for Highly Automated Driving, HADmap) is based on the n-point perspective positioning model algorithm for positioning. By using HADmap to extract the 3D point coordinates in front of the vehicle and projecting them onto the image to match with the detected 2D point coordinates, 3D-2D point pair coordinates are formed. Then, the n-point perspective positioning model algorithm is used to calculate the positioning result. However, during the visual positioning process, due to the errors in the 3D-2D point pair coordinates, it is necessary to evaluate the error range corresponding to any positioning result, and then it can be determined whether the positioning result is valid.

[0003] There are two common methods for determining the error range in the prior art. One is to use a high-precision tool to obtain the position of an object as the true value, and then use the n-point perspective vision model to determine the result of visual positioning, and further obtain the error between the visual positioning result and the true value. By moving the object in the actual scene to obtain a large amount of error statistical data, the error range of the visual positioning result is determined according to the obtained large amount of error data. The other is to use a method combining hardware and software to determine the error range of the positioning result. Specifically, the error range of the visual positioning result is determined according to the accuracy of the hardware positioning device.

[0004] However, the prior art needs to rely on hardware devices to determine the error range corresponding to the positioning result of the n-point perspective positioning model, which results in a more complex process for determining the error range of the n-point perspective positioning model and a lower determination efficiency. Summary of the Invention

[0005] The present application provides a method, device, and equipment for determining an error range, which can simplify the process of determining the error range and improve the efficiency of determining the error range.

[0006] In a first aspect, the present application provides a method for determining an error range, including: obtaining at least four 3D-2D point pair coordinates, where the 3D-2D point pair coordinates are the coordinates corresponding to any point in the world coordinate system and the coordinates corresponding to the point in the pixel coordinate system; determining the positioning result to be evaluated and the uncertainty of the translation vector between the camera coordinate system and the world coordinate system according to at least four 3D-2D point pair coordinates and the n-point perspective positioning model; determining the error range corresponding to each axis of the world coordinate system of the positioning result to be evaluated according to the uncertainty of the translation vector; and outputting the error range.

[0007] Optionally, before determining the uncertainty of the translation vector between the camera coordinate system and the world coordinate system according to at least four 3D-2D point pair coordinates and the n-point perspective model, it further includes: obtaining the depth information corresponding to any point in the physical space, the coordinates corresponding to the point in the pixel coordinate system, the camera parameters, the coordinates corresponding to the point in the world coordinate system, the rotation matrix and the translation vector between the camera coordinate system and the world coordinate system; determining the n-point perspective positioning model according to the depth information corresponding to any point in the physical space, the coordinates corresponding to the point in the pixel coordinate system, the camera parameters, the coordinates corresponding to the point in the world coordinate system, the rotation matrix and the translation vector between the camera coordinate system and the world coordinate system.

[0008] By determining the n-point perspective positioning model, the method for determining the error range provided by the present application can be made not limited by the type of the original n-point perspective positioning model, thus expanding the applicable range.

[0009] Optionally, determining the uncertainty of the translation vector between the camera coordinate system and the world coordinate system according to at least four 3D-2D point pair coordinates and the n-point perspective positioning model includes: determining an error transfer model according to at least four 3D-2D point pair coordinates and the n-point perspective positioning model, where the error transfer model is used to transfer the error of the 3D-2D point pair coordinates to the rotation matrix and the translation vector between the camera coordinate system and the world coordinate system; determining the uncertainty corresponding to the error transfer model according to at least four 3D-2D point pair coordinates and the error transfer model; determining the uncertainty of the translation vector according to the uncertainty corresponding to the error transfer model and the error transfer model.

[0010] Through this method, the error of the 3D-2D point pair coordinates can be transferred to the rotation matrix and the translation vector, simplifying the calculation process, and further simplifying the process of determining the error range corresponding to each axis of the positioning result to be evaluated obtained based on the vision algorithm in the world coordinate system, and improving the efficiency of determining the error range corresponding to each axis of the positioning result in the world coordinate system.

[0011] Optionally, determining the error range corresponding to each axis of the positioning result to be evaluated in the world coordinate system according to the uncertainty of the translation vector includes: determining the error range corresponding to each axis of the positioning result to be evaluated in the world coordinate system according to the uncertainty of the translation vector, the rotation matrix and the translation vector.

[0012] Through this method, the uncertainty of the translation vector determined in the camera coordinate system can be converted into the error range corresponding to each axis in the world coordinate system.

[0013] Optionally, according to the uncertainty of the translation vector, the rotation matrix, and the translation vector, determine the error intervals corresponding to the axes of the world coordinate system, including: according to the uncertainty of the translation vector, the rotation matrix, the translation vector, and the 3δ principle, determine the error intervals corresponding to the axes of the world coordinate system for the positioning result to be evaluated.

[0014] Through this method, not only can the uncertainty of the translation vector determined in the camera coordinate system be converted into the error intervals corresponding to the axes in the world coordinate system, but also the range of the error intervals can be limited by the 3δ principle, improving the effectiveness and readability of the error intervals.

[0015] Optionally, the method further includes: obtaining a preset error interval threshold; comparing whether the error interval is within the error interval threshold; if the error interval is within the error interval threshold, determine that the positioning result to be evaluated is valid.

[0016] Through this method, it is possible to determine whether to trust a certain positioning result, that is, to determine whether the positioning result to be evaluated is valid, thereby improving the accuracy of the positioning device and enhancing the user experience.

[0017] Optionally, the method further includes: the n-point perspective model includes: a linear n-point perspective model, a non-linear n-point perspective model, or a 3-point perspective model.

[0018] In a second aspect, the present application provides a device for determining an error interval, including:

[0019] An acquisition module, configured to acquire at least four 3D-2D point pair coordinates, where the 3D-2D point pair coordinates are the coordinates corresponding to any point in the world coordinate system and the coordinates corresponding to the point in the pixel coordinate system.

[0020] A determination module, configured to determine the positioning result to be evaluated and the uncertainty of the translation vector between the camera coordinate system and the world coordinate system according to at least four 3D-2D point pair coordinates and the n-point perspective positioning model.

[0021] The determination module is further configured to determine the error intervals corresponding to the axes of the world coordinate system for the positioning result to be evaluated according to the uncertainty of the translation vector.

[0022] An output module, configured to output the error interval.

[0023] Optionally, the acquisition module is further configured to acquire the depth information corresponding to any point in the physical space, the coordinates corresponding to the point in the pixel coordinate system, the camera parameters, the coordinates corresponding to the point in the world coordinate system, the rotation matrix and the translation vector between the camera coordinate system and the world coordinate system.

[0024] The determining module is further configured to determine an n-point perspective positioning model according to the depth information corresponding to any point in the physical space, the coordinates corresponding to the point in the pixel coordinate system, the camera parameters, the coordinates corresponding to the point in the world coordinate system, the rotation matrix and the translation vector between the camera coordinate system and the world coordinate system.

[0025] Optionally, the determining module is specifically configured to determine an error transfer model according to at least four 3D-2D point pair coordinates and the n-point perspective positioning model, where the error transfer model is used to transfer the error of the 3D-2D point pair coordinates to the rotation matrix and the translation vector between the camera coordinate system and the world coordinate system; determine the uncertainty corresponding to the error transfer model according to at least four 3D-2D point pair coordinates and the error transfer model; and determine the uncertainty of the translation vector according to the uncertainty corresponding to the error transfer model and the error transfer model.

[0026] Optionally, the determining module is specifically configured to determine the error intervals corresponding to the respective axes of the positioning result to be evaluated in the world coordinate system according to the uncertainty of the translation vector, the rotation matrix, and the translation vector.

[0027] Optionally, the determining module is specifically configured to determine the error intervals corresponding to the respective axes of the positioning result to be evaluated in the world coordinate system according to the uncertainty of the translation vector, the rotation matrix, the translation vector, and the 3δ principle.

[0028] Optionally, the obtaining module is further configured to obtain a preset error interval threshold.

[0029] The determining module is further configured to compare whether the error interval is within the error interval threshold; if the error interval is within the error interval threshold, determine that the positioning result to be evaluated is valid.

[0030] Optionally, the n-point perspective model includes: a linear n-point perspective model, a non-linear n-point perspective model, or a 3-point perspective model.

[0031] In a third aspect, the present application provides an electronic device, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein, the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method according to the first aspect or the optional manner of the first aspect.

[0032] In a fourth aspect, the present application provides a computer-readable storage medium, in which computer-executable instructions are stored, and when the computer-executable instructions are executed by a processor, they are used to implement the method according to the first aspect or the optional manner of the first aspect.

[0033] The present application provides a method, apparatus, and device for determining an error range. By obtaining the 3D-2D point pair coordinates formed by the coordinates of at least four points in the world coordinate system and the coordinates of the points in the pixel coordinate system, and then according to at least four 3D-2D point pair coordinates and the n-point perspective positioning model, the uncertainty of the positioning result to be evaluated and the translation vector between the camera coordinate system and the world coordinate system is determined. Furthermore, the error of the 3D-2D point pair coordinates can be transmitted to the translation vector, and then based on the conversion relationship between the camera coordinate system and the world coordinate system, according to the uncertainty of the translation vector, the error ranges corresponding to each axis of the positioning result in the world coordinate system are determined and the error ranges are output, without the need to rely on other hardware positioning devices, the error ranges corresponding to each axis of the positioning result obtained based on the vision algorithm in the world coordinate system can be determined. Therefore, the process of determining the error range can be simplified and the efficiency of determining the error range can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 FIG. 6 is a schematic diagram of an application scenario of the method for determining an error range provided by the present application;

[0035] Figure 2 FIG. 10 is a schematic flowchart of the method for determining an error range provided by the present application;

[0036] Figure 3 FIG. 14 is a schematic diagram of a camera coordinate system and a world coordinate system provided by the present application;

[0037] Figure 4 FIG. 18 is a schematic diagram of an interface of an electronic device provided by the present application;

[0038] Figure 5 FIG. 22 is another schematic flowchart of the method for determining an error range provided by the present application;

[0039] Figure 6 FIG. 26 is a schematic structural diagram of the device for determining an error range provided by the present application;

[0040] Figure 7 FIG. 30 is a schematic structural diagram of an electronic device provided by the present application.

[0041] Through the above-mentioned drawings, specific embodiments of the present disclosure have been shown, and there will be more detailed descriptions hereinafter. These drawings and the textual descriptions are not intended to limit the scope of the concept of the present disclosure in any way, but to illustrate the concept of the present disclosure to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0042] Exemplary embodiments will be described in detail herein, and examples thereof are shown in the accompanying drawings. When the following description refers to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. On the contrary, they are merely examples of devices and methods consistent with some aspects of the present disclosure as detailed in the appended claims.

[0043] The monocular vision positioning algorithm based on the high-precision map technology HADmap is based on the n-point perspective positioning model algorithm for positioning. By using HADmap, the 3D point coordinates in front of the vehicle are extracted and projected onto the image to match with the detected 2D point coordinates, thereby forming 3D-2D point pair coordinates. Then, the n-point perspective positioning model algorithm is used to calculate the positioning result. The monocular n-point perspective model algorithm calculates the positioning result using the 3D-2D point pair coordinates. However, in vision positioning, due to the errors in the 3D-2D point pair coordinates, it is necessary to evaluate the error interval corresponding to each positioning result, and then it can be determined whether the positioning result is valid.

[0044] There are two common methods for determining the error interval in the prior art. One is to use a high-precision tool to obtain the position of the object as the true value, and then use the n-point perspective vision model to determine the result of vision positioning, and then obtain the error between the vision positioning result and the true value. By moving the object in the actual scene to obtain a large amount of error statistical data, the error interval of the vision positioning result is determined according to the obtained large amount of error data. The other is to use a method combining hardware and software to determine the error interval of the positioning result. Specifically, the error interval of the vision positioning result is determined according to the accuracy of the hardware positioning device. For example, in an autonomous driving system, the hardware usually used in the positioning module includes a general Global Positioning System (GPS), high-precision Real-Time Kinematic (RTK), integrated inertial navigation, etc., and the software usually includes a monocular n-point perspective positioning model. The positioning accuracy of each hardware is different. Specifically, the positioning accuracy of general GPS is about within 10m, so the error interval of the positioning result of its corresponding n-point perspective positioning model is [-10m, 10m]; the positioning accuracy of integrated inertial navigation is about within 1m, so the error interval of the positioning result of its corresponding n-point perspective positioning model is [-1m, 1m].

[0045] However, the prior art all needs to rely on hardware devices to determine the error interval of the n-point perspective positioning model positioning result, which results in a low efficiency in determining the error interval of the n-point perspective positioning model positioning result.

[0046] Based on this, the present application proposes a method for determining the error range of the positioning result of the n-point perspective positioning model without relying on other hardware devices except for the electronic device for determining the execution error range. By obtaining the coordinates corresponding to at least four points in the world coordinate system and the coordinates corresponding to the points in the pixel coordinate system, that is, obtaining at least four 3D-2D point pair coordinates; then, according to at least four 3D-2D point pair coordinates and the n-point perspective positioning model, determining the uncertainty of the positioning result to be evaluated and the translation vector between the camera coordinate system and the world coordinate system; and then, according to the uncertainty of the translation vector, determining the error range corresponding to each axis of the world coordinate system of the positioning result to be evaluated, and outputting the error range, which simplifies the process of determining the error range and improves the efficiency of determining the error range.

[0047] Figure 1 FIG. is a schematic diagram of an application scenario of a method for determining an error range provided by the present application. The execution subject of the method for determining the error range is a device for determining the error range. This device can be part or all of an electronic device, and this electronic device can be installed on a movable device such as an automobile. As Figure 1 shown, this electronic device is installed on the automobile 11. During the stop or driving process of the automobile 11, the electronic device can obtain the 3D-2D point pair coordinates corresponding to multiple markers in the front, such as the sign 12 and the vegetation 13, and determine the positioning result of the automobile 11 based on the visual positioning algorithm, that is, the positioning result to be evaluated, through the obtained 3D-2D point pair coordinates and the n-point perspective model. Further, an error transfer model for transferring the error of the 3D-2D point pair coordinates to the rotation matrix and translation vector between the camera coordinate system and the world coordinate system can also be determined through the 3D-2D point pair coordinates and the n-point perspective model, and the error range corresponding to each axis of the world coordinate system of the corresponding positioning result of the automobile 11 can be determined according to this error transfer model.

[0048] The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in some embodiments. The embodiments of the present invention will be described below with reference to the accompanying drawings.

[0049] Figure 2 FIG. is a flowchart of a method for determining an error range provided by the present application. The execution subject of this method is a device for determining the error range. This device can be part or all of an electronic device, as Figure 2 shown, this method includes:

[0050] S201. Obtain at least four 3D-2D point pair coordinates.

[0051] Among them, the 3D-2D point pair coordinates are the coordinates corresponding to any point in the world coordinate system and the coordinates corresponding to the point in the pixel coordinate system.

[0052] An electronic device can obtain 3D-2D point pair coordinates from an entity or a network storage medium that pre-stores 3D-2D point pair coordinates; and can obtain the 3D coordinates corresponding to any point in the world coordinate system through a positioning device such as GPS, then capture a photo including the point through a camera, camera, etc., and obtain the 2D coordinates corresponding to the point in the pixel coordinate system through the photo. The electronic device can also obtain 3D-2D point pair coordinates input by the user through its interaction interface.

[0053] S202. Determine the positioning result to be evaluated and the uncertainty of the translation vector between the camera coordinate system and the world coordinate system according to at least four 3D-2D point pair coordinates and the n-point perspective positioning model.

[0054] Optionally, the n-point perspective model includes: a linear n-point perspective model, a non-linear n-point perspective model, or a 3-point perspective model.

[0055] The following specifically describes the camera coordinate system, the world coordinate system, and the translation vector between the camera coordinate system and the world coordinate system through a schematic diagram.

[0056] Figure 3 A schematic diagram of a camera coordinate system and a world coordinate system provided by this application is as Figure 3 shown. Point O c is the optical center of the camera, also known as the projection center. The X c axis and the Y c axis are parallel to the x-axis and y-axis of the imaging plane coordinate system. The Z c axis is the optical axis of the camera and is perpendicular to the image plane. The intersection point of the optical axis and the image plane is the principal point O of the image. The right-angled coordinate system composed of point O c and the X c axis, the Y c axis, and the Z c axis is the coordinate system of the camera, that is, the camera coordinate system. Among them, O c -O is the focal length of the camera. The right-angled coordinate system composed of point O w and the X w axis, the Y w axis, and the Z w axis is the world coordinate system. The translation vector and the rotation matrix are used to represent the relationship between the camera coordinate system and the world coordinate system. The rotation matrix is the product of three axial rotation matrices. The translation vector represents the translation distances in three axial directions. The uncertainty of the translation vector between the camera coordinate system and the world coordinate system refers to the credibility of the translation vector. The smaller the uncertainty, the higher the reliability of the translation vector; the larger the uncertainty, the lower the reliability of the translation vector.

[0057] Specifically, according to at least four 3D-2D point pair coordinates and the n-point perspective positioning model, the positioning result to be evaluated can be obtained by inputting at least four 3D-2D point pair coordinates into the n-point perspective model through operations. To improve the accuracy of the obtained positioning result to be evaluated, multiple 3D-2D point pair coordinates can be input into the n-point perspective model. For example, 6 or more 3D-2D point pair coordinates can be input into the n-point perspective model, and the positioning result to be evaluated is obtained through operations. There are many n-point perspective models. Taking EPNP as an example, the method for solving the positioning result includes the following steps:

[0058] 1. Obtain at least four 3D-2D point pair coordinates;

[0059] 2. Calculate the homography matrix and obtain the rotation matrix R and translation vector T from the world coordinate system to the camera coordinate system;

[0060] 3. Use the obtained R and T to convert the coordinates of the camera in the camera coordinate system to the world coordinate system, thereby obtaining the positioning result.

[0061] Specifically, according to at least four 3D-2D point pair coordinates and the n-point perspective positioning model, determining the uncertainty of the translation vector between the camera coordinate system and the world coordinate system may include the following steps: First, determine the error propagation model according to at least four 3D-2D point pair coordinates and the n-point perspective positioning model. Second, determine the uncertainty of the translation vector between the camera coordinate system and the world coordinate system according to at least four 3D-2D point pair coordinates and the error propagation model.

[0062] Among them, the specific process of determining the uncertainty of the translation vector is described in detail below.

[0063] S203. According to the uncertainty of the translation vector, determine the error intervals corresponding to each axis of the positioning result to be evaluated in the world coordinate system.

[0064] Among them, the error intervals can be used to evaluate the effectiveness of the positioning result to be evaluated.

[0065] Specifically, the uncertainty of the translation vector is determined based on the camera coordinate system. To determine the error intervals corresponding to each axis of the 3D-2D point pair coordinates to be evaluated in the world coordinate system, the uncertainty of each dimension of the translation vector determined based on the camera coordinate system can be converted into the error intervals corresponding to each axis of the world coordinate system through the rotation matrix and translation vector between the camera coordinate system and the world coordinate system.

[0066] S204. Output the error intervals.

[0067] The electronic device can output the error intervals to other devices; it can also output the error intervals through its display interface. Figure 4An interface schematic diagram of the output error range provided for this application is as follows Figure 4 As shown, the electronic device can output on its display interface "The position of the positioning result to be evaluated in the world coordinate system is [100, 89, 85], and the X w 、Y w 、Z w error ranges are [-6, 6], [-3, 3], [-9, 9] respectively!"

[0068] The method for determining the error range provided by the embodiment of this application obtains the 3D-2D point pair coordinates composed of the coordinates corresponding to at least four points in the world coordinate system and the coordinates corresponding to the points in the pixel coordinate system, and then determines the uncertainty of the positioning result to be evaluated and the translation vector between the camera coordinate system and the world coordinate system according to at least four 3D-2D point pair coordinates and the n-point perspective positioning model. Furthermore, according to the uncertainty of the translation vector, the error ranges corresponding to each axis of the positioning result to be evaluated in the world coordinate system are determined and the error ranges are output. Without the need to rely on other hardware positioning devices, the error ranges corresponding to each axis of the positioning result to be evaluated in the world coordinate system can be determined. Therefore, the process of determining the error range can be simplified and the efficiency of determining the error range can be improved.

[0069] Furthermore, there are many types of n-point perspective models. Different n-point perspective models have different sensitivities to algorithm complexity and errors. By establishing the n-point perspective model, the computational complexity and accuracy of the n-point perspective model can be balanced. In order to further improve the accuracy of determining the error range, on the basis of the above embodiment, the step of constructing the n-point perspective model can also be included. Figure 5 Another process schematic diagram of the method for determining the error range provided for this application. The execution subject of this method is the device for determining the error range. This device can be part or all of the electronic device, as follows Figure 5 As shown, this method includes

[0070] S501. Obtain the depth information corresponding to any point in the physical space, the coordinates corresponding to the point in the pixel coordinate system, the camera parameters, the coordinates corresponding to the point in the world coordinate system, the rotation matrix and the translation vector between the camera coordinate system and the world coordinate system.

[0071] The camera internal parameters can use the camera calibration algorithm, such as Zhang Zhengyou calibration method, to obtain the camera calibration parameters; the coordinates corresponding to the point in the pixel coordinate system can use the corner detection algorithm, such as using deep learning to detect signs, lane line corners, etc.; the coordinates of the point in the world coordinate system can be extracted from the high-precision map and matched with the point in the pixel coordinate system to complete the correspondence of the 3D-2D point pair.

[0072] In this step, the electronic device can obtain corresponding data through an entity or virtual storage medium that pre-stores any one of the depth information corresponding to any point in the physical space, the coordinates corresponding to the point in the pixel coordinate system, the camera parameters, the coordinates corresponding to the point in the world coordinate system, the rotation matrix between the camera coordinate system and the world coordinate system, and the translation vector between the camera coordinate system and the world coordinate system, or from the electronic device; it can also obtain any one of the above data input by the user through a human-computer interaction interface, etc.; it can obtain some of the above data through an entity or virtual storage medium, and then obtain another part of the data input by the user through the human-computer interaction interface.

[0073] S502. Determine the n-point perspective positioning model according to the depth information corresponding to any point in the physical space, the coordinates corresponding to the point in the pixel coordinate system, the camera parameters, the coordinates corresponding to the point in the world coordinate system, the rotation matrix and translation vector between the camera coordinate system and the world coordinate system.

[0074] The n-point perspective positioning model will be described below with reference to the accompanying drawings. Continue to refer to Figure 3 , as Figure 3 shown, O W -X W Y W Z W is the world coordinate system, O c -X c Y c Z c is the camera coordinate system, O-xy is the image plane coordinate system, and O0-uv is the pixel coordinate system. P(X w , Y w , Z w ) is a three-dimensional coordinate point in the world coordinate system, and p(x, y) is the two-dimensional coordinate point corresponding to the projection of P(X w , Y w , Z w ) in the image plane coordinate system. Then the relationship between them is shown in formula (1):

[0075]

[0076] where λ is the depth information, is the camera internal parameter.

[0077] is the rotation matrix and translation vector between the world coordinate system and the camera coordinate system.

[0078] The relationship between the pixel coordinates and the image plane coordinates is shown in formula (2):

[0079]

[0080] According to formulas (1) and (2), the relationship between the three-dimensional coordinates in the world coordinate system and the image plane coordinates is obtained, that is, the n-point perspective positioning model, as shown in formula (3):

[0081]

[0082] S503. Obtain at least four 3D-2D point pair coordinates.

[0083] S503 is similar to S201. For specific descriptions, refer to S201 and will not be elaborated here.

[0084] S504. Determine the positioning result to be evaluated and the uncertainty of the translation vector between the camera coordinate system and the world coordinate system according to at least four 3D-2D point pair coordinates and the n-point perspective positioning model.

[0085] For detailed descriptions of the camera coordinate system, the world coordinate system, and the translation vector between the camera coordinate system and the world coordinate system, refer to S202 and will not be elaborated here.

[0086] Optionally, a possible implementation of determining the uncertainty of the translation vector between the camera coordinate system and the world coordinate system according to at least four 3D-2D point pair coordinates and the n-point perspective positioning model is as follows: First, determine an error propagation model according to at least four 3D-2D point pair coordinates and the n-point perspective positioning model. The error propagation model is used to propagate the error of the 3D-2D point pair coordinates to the rotation matrix and the translation vector between the camera coordinate system and the world coordinate system. Second, determine the uncertainty corresponding to the error propagation model according to at least four 3D-2D point pair coordinates and the error propagation model. Finally, determine the uncertainty of the translation vector according to the uncertainty corresponding to the error propagation model and the error propagation model. Through this method, the error of the 3D-2D point pair coordinates can be propagated to the rotation matrix and the translation vector, simplifying the calculation process, and further simplifying the process of determining the error intervals corresponding to each axis of the positioning result to be evaluated in the world coordinate system, and improving the efficiency of determining the error intervals corresponding to each axis of the positioning result to be evaluated in the world coordinate system.

[0087] Next, a specific example is used to illustrate the above implementation:

[0088] The electronic device obtaining at least four 3D-2D point pair coordinates can be obtained from one or more physical or virtual storage media pre-storing 3D-2D point pair coordinates; it can also be the at least four 3D-2D point pair coordinates input by the user obtained through the human-computer interaction interface; or it can be the at least four 3D-2D point pair coordinates sent by other electronic devices to the electronic device.

[0089] The electronic device substitutes at least four 3D-2D point coordinates it has obtained into the formula (3) corresponding to the n-point perspective positioning model, and after performing arithmetic transformation, obtains formula (4):

[0090]

[0091] And uses the least squares method to obtain A and the error transfer function shown in formula (5):

[0092]

[0093] Since t3 is unknown, only

[0094] L′ = [r′ 11 r′ 12 r′ 13 t′1 r′ 21 r′ 22 r′ 23 t′2 r′ 31 r′ 32 r′ 33 T = L / t3

[0095] As can be seen from formula (4), for every four 3D-2D point pair coordinates, there is A i L = b i , where:

[0096]

[0097]

[0098] Since L is a function of the 3D-2D point pair coordinates, and the 3D-2D point pair coordinates are independent of each other, the covariance matrix of the 3D-2D point pair coordinates can be constructed using the standard deviation, and its uncertainty of L′ can be solved:

[0099]

[0100] Among them, Λ p is the covariance matrix of the 3D-2D point pair coordinates, Λ′ L is the covariance matrix of L′, J is the Jacobian matrix of matrix A, and p = [X, Y, Z, x, y] is the 3D-2D point pair coordinates.

[0101] And for any one 3D-2D point pair coordinate:

[0102]

[0103] And

[0104] ​

[0105]

[0106] Since knowing A i and b i , the following can be obtained:

[0107]

[0108]

[0109]

[0110] Similarly, solve for and Substitute them into Equation (7), and substitute each Equation (7) formed by the 3D-2D point pairs into Equation (6), then the uncertainty of L′ can be obtained.

[0111] However, the elements in L′ obtained according to Equation (4) are not the elements of the rotation matrix R and the translation vector T. To solve for L, a scale factor t3 is still missing, and the true RT is:

[0112]

[0113] Since the rotation matrix is an orthogonal unit matrix, so the modulus of R3 = [t3r′ 31 , t3r′ 32 , t3r′ 33 is 1, from which the following can be solved:

[0114]

[0115] Thus, the uncertainties of the rotation matrix R, and the translation vectors t1 and t2 can be obtained:

[0116] Λ L =(t3) 2 Λ′ L (12)

[0117] where, Λ L is an 11*11 dimensional matrix. From Equations (4) and (6), it can be obtained that Λ t1 =Λ L (4, 4) is the variance of t1, Λ t2 =Λ L (8, 8), is the variance of t2, where, t1 identifies the 4th row of the matrix represented by Λ L , and t2 identifies the 8th row of the matrix represented by Λ L .

[0118] According to Equation (11) and the delta method, it can be known that r′31 , r' 32 , r' 33 The uncertainty of and t3 is r' 31 , r' 32 , r' 33 is a function of, so we can obtain:

[0119] Λ t3 = CΛ R3 C T (13)

[0120] Among them, Λ t3 is the variance of t3, and Λ R3 is the 31 , r' 32 , r' 33 covariance matrix of, r', r'.

[0121] Among them,

[0122] Thus, the error intervals in the x-axis direction, y-axis direction, and z-axis direction in the camera coordinate system can be obtained as

[0123] Among them, the delta method can be: Let gj(1 ≤ j ≤ m) be an n-ary function with a first-order total differential, and x = (x_1, x_2…, x_n)^T be an n-order random vector. If the following conditions are satisfied:

[0124]

[0125] Among them, B ≥ 0 is a k-order square matrix, then,

[0126]

[0127] Among them, C is an m*n-dimensional matrix, and its

[0128] S505. Determine the error intervals corresponding to each axis of the positioning result to be evaluated in the world coordinate system according to the uncertainty of the translation vector.

[0129] Among them, the error interval is used to evaluate the effectiveness of the positioning result to be evaluated.

[0130] Based on the above examples, it can be seen that the uncertainty of the previously solved translation vector is based on the camera coordinate system. To determine the error intervals corresponding to each axis of the positioning result to be evaluated in the world coordinate system, it is necessary to convert to the world coordinate system. Therefore, it is necessary to use the rotation matrix and translation vector from the camera coordinate system to the world coordinate system, and convert the uncertainty of the translation vector solved based on the camera coordinate system to the world coordinate system through coordinate transformation, so as to determine the error intervals corresponding to each axis of the positioning result to be evaluated in the world coordinate system.

[0131] Optionally, a possible implementation method for determining the error intervals corresponding to each axis of the positioning result to be evaluated in the world coordinate system according to the uncertainty of the translation vector is: determine the error intervals corresponding to each axis of the positioning result to be evaluated in the world coordinate system according to the uncertainty of the translation vector, the rotation matrix, and the translation vector.

[0132] The formula for converting the translation vector from the camera coordinate system to the world coordinate system is:

[0133] t w =-R cw *t c (14)

[0134] where t w and t c represent the translation vectors of the world coordinate system and the camera coordinate system respectively, and R cw represents the corresponding rotation matrix after the rotation matrix is converted from the camera coordinate system to the world coordinate system.

[0135] Thus, according to Equation (6), the uncertainty of the translation vector in the world coordinate system can be obtained:

[0136]

[0137] where Λ tc =diag(Λ t1 , Λ t2 , Λ t3 ), is a 3*3 diagonal matrix, and Λ tw =diag(Λ tw1 , Λ tw2 , Λ tw3 ), is also a 3*3 diagonal matrix. Then, the error intervals in the x-axis direction, y-axis direction, and z-axis direction in the world coordinate system can be obtained as

[0138] To further constrain the error intervals corresponding to the axes of the world coordinate system, limit the range of the error intervals, and improve the effectiveness of the error intervals, the 3δ principle can also be used to further constrain the determination of the error intervals corresponding to the axes of the world coordinate system based on the uncertainty of the translation vector, the rotation matrix, and the translation vector. The 3δ principle means that if a value follows a normal distribution, then within ±3δ, 99.73% of the values are included, that is, most of the values are distributed within ±3δ.

[0139] Specifically, another possible implementation of determining the error intervals corresponding to the axes of the world coordinate system based on the uncertainty of the translation vector, the rotation matrix, and the translation vector is: determining the error intervals corresponding to the axes of the world coordinate system based on the uncertainty of the translation vector, the rotation matrix, the translation vector, and the 3δ principle. The 3δ principle is based on the basic ideas of "small probability events" and hypothesis testing, and it is considered that this event is almost impossible to occur in one experiment. Therefore, in practical problems, it is considered that events outside (μ - 3σ, μ + 3σ) will not occur.

[0140] S506. Output the error interval.

[0141] S506 is similar to S204. For specific descriptions, refer to S204 and will not be elaborated here.

[0142] On the basis of determining the error interval, further, the effectiveness of any positioning result of the positioning device can be determined by applying this error interval. Based on this, on the basis of the above steps, optionally, the method further includes:

[0143] S507. Obtain a preset error interval threshold.

[0144] The electronic device can obtain the preset error interval threshold through an entity or virtual storage medium that pre-stores the error interval threshold; or it can obtain the error interval threshold input by the user through a human-computer interaction interface.

[0145] The error interval threshold can be set according to actual needs. For example, when the accuracy requirement for the positioning result is not high, the error interval threshold can be set relatively large; when the accuracy requirement for the positioning result is relatively high, the error interval threshold can be set relatively small.

[0146] S508. Compare whether the error interval is within the error interval threshold.

[0147] S509. If the error interval is within the error interval threshold, determine that the positioning result to be evaluated is valid; if the error interval is not within the error interval threshold, determine that the positioning result to be evaluated is invalid.

[0148] The following uses an example to illustrate S508 and S509:

[0149] For example, the preset error interval threshold is [-8, 8]. Assume that the positioning result to be evaluated obtained by the positioning device is p1 = [X1, Y1, Z1], and the error intervals corresponding to each axis in the world coordinate system are [-3, 3]. Since [-3, 3] is within the range defined by [-8, 8], it can be determined that the positioning result p1 to be evaluated obtained based on the vision algorithm is valid; assume that the positioning result to be evaluated obtained by the positioning device is p2 = [X2, Y2, Z2], and the error intervals corresponding to each axis in the world coordinate system are [-9, 9]. Since [-9, 9] is not within the range defined by [-8, 8], it can be determined that the positioning result p2 to be evaluated obtained based on the vision algorithm is invalid.

[0150] The method for determining the error interval provided by the present application, by obtaining the depth information corresponding to any point in the physical space, the coordinates of the point in the pixel coordinate system, the camera parameters, the coordinates of the point in the world coordinate system, the rotation matrix and translation vector between the camera coordinate system and the world coordinate system, and then determining the n-point perspective positioning model according to the depth information corresponding to any point in the physical space, the coordinates of the point in the pixel coordinate system, the camera parameters, the coordinates of the point in the world coordinate system, the rotation matrix and translation vector between the camera coordinate system and the world coordinate system, can make the method for determining the error interval provided by the present application not restricted by the types of existing n-point perspective positioning models, and expand the applicable range. Further, on the basis of the above embodiments, the present application also obtains a preset error interval threshold, and by comparing whether the error interval is within the error interval threshold, if the error interval is within the error interval threshold, it is determined that the positioning result to be evaluated is valid; if the error interval is not within the error interval threshold, it is determined that the positioning result to be evaluated is invalid, which can judge the effectiveness of the positioning result of the positioning device and improve the accuracy of the positioning device. At the same time, the present application can not only output the error interval, but also output whether the positioning result is valid. Therefore, the method has a wider range of use and better user experience.

[0151] Figure 6 It is a schematic structural diagram of a device for determining the error interval provided by the present application, as Figure 6 shown, the device includes:

[0152] An acquisition module 61, configured to acquire at least four 3D-2D point pair coordinates, where the 3D-2D point pair coordinates are the coordinates of any point in the world coordinate system and the coordinates of the point in the pixel coordinate system.

[0153] A determination module 62, configured to determine the positioning result to be evaluated and the uncertainty of the translation vector between the camera coordinate system and the world coordinate system according to at least four 3D-2D point pair coordinates and the n-point perspective positioning model.

[0154] The determination module 62 is further configured to determine the error intervals corresponding to each axis of the world coordinate system for the positioning result to be evaluated according to the uncertainty of the translation vector.

[0155] The output module 63 is configured to output the error intervals.

[0156] Optionally, the acquisition module 61 is further configured to acquire the depth information corresponding to any point in the physical space, the coordinates corresponding to the point in the pixel coordinate system, the camera parameters, the coordinates corresponding to the point in the world coordinate system, the rotation matrix and the translation vector between the camera coordinate system and the world coordinate system.

[0157] The determination module 62 is further configured to determine the n-point perspective positioning model according to the depth information corresponding to any point in the physical space, the coordinates corresponding to the point in the pixel coordinate system, the camera parameters, the coordinates corresponding to the point in the world coordinate system, the rotation matrix and the translation vector between the camera coordinate system and the world coordinate system.

[0158] Optionally, the determination module 62 is specifically configured to determine an error transfer model according to at least four 3D-2D point pair coordinates and the n-point perspective positioning model, where the error transfer model is used to transfer the error of the 3D-2D point pair coordinates to the rotation matrix and the translation vector between the camera coordinate system and the world coordinate system; determine the uncertainty corresponding to the error transfer model according to at least four 3D-2D point pair coordinates and the error transfer model; and determine the uncertainty of the translation vector according to the uncertainty corresponding to the error transfer model and the error transfer model.

[0159] Optionally, the determination module 62 is specifically configured to determine the error intervals corresponding to each axis of the world coordinate system for the positioning result to be evaluated according to the uncertainty of the translation vector, the rotation matrix and the translation vector.

[0160] Optionally, the determination module 62 is specifically configured to determine the error intervals corresponding to each axis of the world coordinate system for the positioning result to be evaluated according to the uncertainty of the translation vector, the rotation matrix, the translation vector and the 3δ principle.

[0161] Optionally, the acquisition module 61 is further configured to acquire a preset error interval threshold.

[0162] The determination module 62 is further configured to compare whether the error intervals are within the error interval threshold; if the error intervals are within the error interval threshold, determine that the positioning result to be evaluated is valid.

[0163] Optionally, it further includes: the n-point perspective model includes: a linear n-point perspective model, a non-linear n-point perspective model or a 3-point perspective model.

[0164] The device for determining the error intervals can execute the above method for determining the error intervals, and the content and effects can be referred to the method embodiment part, which will not be elaborated here.

[0165] Figure 7 A structural schematic diagram of the electronic device provided for this application is as follows Figure 7 As shown, the underwriting data processing device of this embodiment includes: a processor 71 and a memory 72; the processor 71 is communicatively connected to the memory 72. The memory 72 is used to store computer programs. The processor 71 is used to call the computer programs stored in the memory 72 to implement the methods in the above method embodiments.

[0166] Optionally, the electronic device further includes: a transceiver 73, which is used to communicate with other devices.

[0167] The electronic device can execute the above method for determining the error range, and its content and effects can be referred to the method embodiment part, which will not be elaborated here.

[0168] This application also provides a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement the above method for determining the error range.

[0169] When the computer-executable instructions stored in the computer-readable storage medium are executed by a processor, they can implement the above method for determining the error range, and its content and effects can be referred to the method embodiment part, which will not be elaborated here.

[0170] After considering the specification and practicing the invention disclosed herein, those skilled in the art will readily conceive of other embodiments of the present disclosure. This application is intended to cover any variations, uses, or adaptations of the present disclosure, which follow the general principles of the present disclosure and include known common general knowledge or conventional technical means in the technical field not disclosed in the present disclosure. The specification and embodiments are only regarded as exemplary, and the true scope and spirit of the present disclosure are pointed out by the following claims. It should be understood that the present disclosure is not limited to the exact structures described above and shown in the drawings, and various modifications and changes can be made without departing from its scope. The scope of the present disclosure is only limited by the appended claims.

Claims

1. A method for determining an error range, characterized in that Including: Obtain at least four 3D-2D point pair coordinates, where the 3D-2D point pair coordinates are the coordinates corresponding to any point in the world coordinate system and the coordinates corresponding to the point in the pixel coordinate system; Determine the positioning result to be evaluated and the uncertainty of the translation vector between the camera coordinate system and the world coordinate system according to the at least four 3D-2D point pair coordinates and the n-point perspective positioning model; Based on the uncertainty of the translation vector, the rotation matrix, the translation vector, and 3 principles, determine the error intervals corresponding to each axis of the world coordinate system for the positioning result to be evaluated; Output the error interval; The determining the uncertainty of the translation vector between the camera coordinate system and the world coordinate system according to the at least four 3D-2D point pair coordinates and the n-point perspective positioning model includes: Determine an error propagation model according to the at least four 3D-2D point pair coordinates and the n-point perspective positioning model, where the error propagation model is used to propagate the error of the 3D-2D point pair coordinates to the rotation matrix and the translation vector between the camera coordinate system and the world coordinate system; Based on the covariance matrix of the 3D-2D point pair coordinates, calculate the covariance matrix of the intermediate variables: transfer the covariance matrix to the intermediate variables through the Jacobian matrix, calculate the uncertainty of the intermediate variables, calculate the rotation matrix and the translation vector through the proportionality coefficient and the uncertainty of the intermediate variables, and use the Delta method to calculate the variance of the translation vector to obtain the uncertainty corresponding to the error propagation model; Determine the uncertainty of the translation vector according to the uncertainty corresponding to the error propagation model and the error propagation model.

2. The method according to claim 1, wherein Before determining the uncertainty of the translation vector between the camera coordinate system and the world coordinate system according to the at least four 3D-2D point pair coordinates and the n-point perspective model, it further includes: Obtain the depth information corresponding to any point in the physical space, the coordinates corresponding to the point in the pixel coordinate system, the camera parameters, the coordinates corresponding to the point in the world coordinate system, the rotation matrix and the translation vector between the camera coordinate system and the world coordinate system; Determine the n-point perspective positioning model according to the depth information corresponding to any point in the physical space, the coordinates corresponding to the point in the pixel coordinate system, the camera parameters, the coordinates corresponding to the point in the world coordinate system, the rotation matrix and the translation vector between the camera coordinate system and the world coordinate system.

3. The method according to claim 1, wherein The method further includes: Obtain a preset error interval threshold; Compare whether the error interval is within the error interval threshold; If the error interval is within the error interval threshold, determine that the positioning result to be evaluated is valid.

4. The method according to claim 1, characterized in that The n-point perspective positioning model includes: a linear n-point perspective model, a non-linear n-point perspective model or a 3-point perspective model.

5. A device for determining an error range, characterized in that, Including: An acquisition module, configured to obtain at least four 3D-2D point pair coordinates, where the 3D-2D point pair coordinates are the coordinates corresponding to any point in the world coordinate system and the coordinates corresponding to the point in the pixel coordinate system; A determination module, configured to determine the positioning result to be evaluated and the uncertainty of the translation vector between the camera coordinate system and the world coordinate system according to the at least four 3D-2D point pair coordinates and the n-point perspective positioning model; The determining module is further configured to determine an error interval corresponding to each axis of the world coordinate system of the to-be-evaluated positioning result according to the uncertainty of the translation vector, the rotation matrix, the translation vector, and the 3 principles. An output module, configured to output the error interval; The determining module is specifically configured to determine an error transfer model according to the at least four 3D-2D point pair coordinates and the n-point perspective positioning model, where the error transfer model is used to transfer the error of the 3D-2D point pair coordinates to the rotation matrix and the translation vector between the camera coordinate system and the world coordinate system; calculate the covariance matrix of intermediate variables based on the covariance matrix of the 3D-2D point pair coordinates; transfer the covariance matrix to the intermediate variables through the Jacobian matrix, calculate the uncertainty of the intermediate variables, calculate the rotation matrix and the translation vector through the proportionality coefficient and the uncertainty of the intermediate variables, and calculate the variance of the translation vector by using the Delta method to obtain the uncertainty corresponding to the error transfer model; determine the uncertainty of the translation vector according to the uncertainty corresponding to the error transfer model and the error transfer model.

6. An electronic device, characterized in that, Comprising: At least one processor; And A memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method according to any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that, Computer-executable instructions are stored in the computer-readable storage medium, and when the computer-executable instructions are executed by a processor, they are used to implement the method according to any one of claims 1 to 4.

8. A computer program product comprising a computer program which, when executed by a processor, implements the method according to any one of claims 1-4.

Images