Visual pose estimation methods, systems, laser devices and electronic equipment

Abstract

This invention relates to a visual pose estimation method, system, laser device, and electronic device, comprising: acquiring two different features of a target object; acquiring a connected coordinate system of a transmitting unit and a receiving unit, wherein the transmitting unit emits laser beams to the receiving unit and is located on different features; acquiring an image of laser spots formed by multiple laser beams, and transforming the quantized representation of the laser spots in the image coordinate system into a representation equation in the connected coordinate system through homography transformation; establishing a relative pose calculation model between the connected coordinate systems, proposing an optimization objective, and transforming the optimization objective function into a relational expression of a column vector consisting of a corresponding matrix and the unknown quantity through the Kronecker product; transforming a nonlinear system of equations into a linear system of equations, and calculating the solution of the linear system of equations; calculating the gradient matrix of the objective function with respect to the desired spinor using mathematical tools such as Lie groups and spinors, and obtaining the optimal solution through Newton-Raphson iteration using the solution of the linear system of equations as the initial value.

Description

Technical Field

[0001] This invention relates to the fields of computer vision and industrial measurement technology, and in particular to a visual pose estimation method, system, laser device, and electronic device. Background Technology

[0002] In industrial applications, pose measurement is a common requirement, such as in component assembly and robot navigation, where accurate relative pose measurement is essential. Among these methods, vision-based pose estimation has seen significant advancements due to its low hardware cost and simple structure. The foundation of pose estimation lies in establishing a mapping relationship between real-world coordinates and image projection coordinates, thereby obtaining constraint equations and conditions, which are then further calculated.

[0003] Currently, based on the type of sensor used, vision-based pose estimation can be categorized into monocular, binocular, and RGBD pose estimation. Monocular vision pose estimation has the simplest structure, but due to the loss of depth information during the 3D-to-2D mapping process, the estimated results often differ from the true values ​​by a certain scale. Binocular pose estimation obtains scene depth through baseline matching of feature points in the left and right images, while RGBD pose estimation further utilizes color information to improve the accuracy of depth information reconstruction, thereby achieving six-degree-of-freedom pose estimation.

[0004] However, traditional vision-based pose estimation methods have many limitations. First, monocular vision cannot establish depth scenes and obtain depth information. The measurement accuracy of binocular vision is greatly affected by the environment, such as changes in ambient light, which is the biggest problem limiting the widespread application of computer vision in industrial fields. Second, there is no widely applicable method for extracting and describing image feature points. Generally, specific feature point extraction and description methods are developed for specific scenes, and this process consumes significant computational resources, requiring hardware acceleration to ensure computational efficiency.

[0005] Therefore, traditional visual pose estimation methods consume significant computational resources and have low computational efficiency. Summary of the Invention

[0006] Therefore, it is necessary to provide a visual pose estimation method, system, laser device, and electronic device that can reduce computing resource consumption and have high computational efficiency to address the above-mentioned technical problems.

[0007] This invention provides a visual pose estimation method, the method comprising:

[0008] Obtain a first feature and a second feature of the object to be tested, wherein the first feature and the second feature are two different features of the object to be tested;

[0009] Obtain the connected coordinate system corresponding to the transmitting unit and the receiving unit respectively. The transmitting unit is used to emit a laser beam to the receiving unit, and the receiving unit is used to receive the laser beam. The transmitting unit and the receiving unit are located on the first feature and the second feature respectively.

[0010] The image acquisition device acquires images of laser spots formed by multiple laser beams, and the quantitative representation of the laser spots in the image coordinate system is transformed into a representation equation in the connected coordinate system through homography transformation.

[0011] A mapping model for the relative pose transformation between the laser beam direction and the laser spot position to the corresponding coordinate systems of the transmitting and receiving units is established to obtain a nonlinear equation set. All variables and coefficient matrices are then extracted using the Kronecker product.

[0012] The nonlinear equation system is transformed into a linear equation system between the corresponding coefficient matrix and the column vector containing all unknown elements, and the solution of the linear equation system is obtained by calculation as the relative pose estimate.

[0013] The relative pose between two connected coordinate systems is characterized using mathematical tools such as spinors and Lie groups. The value of the nonlinear equation system is used as the objective function. The gradient matrix of the relative pose spinor of the objective function is calculated and obtained. The estimated relative pose value is then substituted into the equation through Newton-Raphson iteration to obtain the optimal solution for the pose estimation of the object under test.

[0014] In one embodiment, the connected coordinate system includes a first connected coordinate system and a second connected coordinate system, wherein the first connected coordinate system and the second connected coordinate system are respectively the connected coordinate systems of the transmitting unit and the receiving unit;

[0015] The step of acquiring an image of laser spots formed by multiple laser beams through an image acquisition device, and transforming the quantized representation of the laser spots in the image coordinate system into a representation equation in the connected coordinate system through homography transformation, includes:

[0016] Based on the distribution of the laser spots on the receiving unit, an image of the laser spots is acquired by an image acquisition device;

[0017] An image coordinate system is constructed based on the laser spot image, and the first coordinates of the laser spot are obtained by machine vision. The first coordinates are the coordinates of the laser spot in the image coordinate system.

[0018] In one embodiment, the characterization equation includes a first characterization equation and a second characterization equation, wherein the first characterization equation and the second characterization equation are respectively the characterization equations corresponding to the first connected coordinate system and the second connected coordinate system.

[0019] The step of acquiring an image of laser spots formed by multiple laser beams through an image acquisition device, and transforming the quantization representation of the laser spots in the image coordinate system into a representation equation in the connected coordinate system through homography transformation, further includes:

[0020] The first coordinates are converted into the first characterization equation through homography transformation, and the first coordinate transformation matrix corresponding to the first characterization equation is obtained.

[0021] The second coordinate transformation is used to obtain the second characterization equation of the laser spot in the second connected coordinate system, and the second coordinate transformation matrix corresponding to the second characterization equation is obtained. The second coordinate transformation matrix is ​​the pose transformation matrix between the two connected coordinate systems.

[0022] In one embodiment, the image acquisition device acquires an image of laser spots formed by multiple laser beams, and transforms the quantization representation of the laser spots in the image coordinate system into a representation equation in the connected volume coordinate system through homography transformation, followed by:

[0023] Elements in the first and second characterization equations are obtained through calibration, and these elements are used to construct a system of nonlinear equations.

[0024] Based on the correspondence between the first and second characterization equations, a relative pose calculation model between the first and second connected coordinate systems is established to obtain the optimization objective function for the pose estimation of the object under test.

[0025] In one embodiment, the mapping model establishing the relative pose transformation between the laser beam direction and laser spot position to the corresponding coordinate systems of the transmitting and receiving units is used to obtain a nonlinear equation set. All variables and coefficient matrices are then extracted using the Kronecker product, including:

[0026] Obtain the constraint sub-equations introduced by the laser beam, and eliminate the linearly dependent equations through transformation to obtain the constraint target with full rank coefficients;

[0027] Based on the stated constraints, the optimization objective function is transformed into a relational expression between matrices and column vectors using the Kronecker product.

[0028] In one embodiment, the method further includes:

[0029] Based on the connected coordinate system, the relative pose of the first feature and the second feature is obtained, and the relative pose is described as a third coordinate transformation matrix;

[0030] The third coordinate transformation matrix is ​​characterized by Lie groups and spinors to obtain the fourth coordinate transformation matrix corresponding to the spinor axis of the transformation from the first feature to the second feature.

[0031] In one embodiment, the step of transforming the nonlinear equation system into a linear equation system between a corresponding coefficient matrix and a column vector containing all unknown elements, and obtaining the solution of the linear equation system as a relative pose estimate, includes:

[0032] The general solution of the linear equation system is obtained by null space decomposition.

[0033] The particular solution of the linear equation system is calculated and obtained by combining the orthogonality constraint of the fourth coordinate transformation matrix.

[0034] The present invention also provides a visual pose estimation system, the system comprising:

[0035] The first acquisition module is used to acquire a first feature and a second feature of the object to be tested, wherein the first feature and the second feature are two different features of the object to be tested;

[0036] The second acquisition module is used to acquire the connected coordinate system corresponding to the transmitting unit and the receiving unit respectively. The transmitting unit is used to emit a laser beam to the receiving unit, and the receiving unit is used to receive the laser beam. The transmitting unit and the receiving unit are located on the first feature and the second feature respectively.

[0037] The quantization characterization module is used to acquire images of laser spots formed by multiple laser beams through an image acquisition device, and to transform the quantization characterization of the laser spots in the image coordinate system into a characterization equation in the connected coordinate system through homography transformation.

[0038] The function processing module is used to establish a mapping model of the relative pose transformation between the laser beam direction and the laser spot position to the corresponding coordinate systems of the transmitting and receiving units, respectively, to obtain a nonlinear equation system, and extract all variables and coefficient matrices through the Kronecker product.

[0039] The computational processing module is used to transform the nonlinear equation system into a linear equation system between the corresponding coefficient matrix and the column vector containing all unknown elements, and to obtain the solution of the linear equation system as a relative pose estimate.

[0040] The optimal solution acquisition module is used to characterize the relative pose between two connected coordinate systems using mathematical tools such as spinors and Lie groups. It takes the value of the nonlinear equation system as the objective function, calculates and obtains the gradient matrix of the relative pose spinor of the objective function, and substitutes the estimated relative pose value into it through Newton-Raphson iteration to obtain the optimal solution for the pose estimation of the object under test.

[0041] The present invention also provides a laser device for visual pose estimation, used to implement any of the above-described visual pose estimation methods, comprising:

[0042] The emitting unit includes a laser emitter and an attitude adjustment device. The laser emitter is fixedly installed on the free end of the attitude adjustment device so that the angle of the laser beam emitted by the laser emitter can be adjusted by the attitude adjustment device.

[0043] The receiving unit includes a semi-transparent screen, a monocular camera, and a base. The semi-transparent screen is fixedly installed in front of the base, and the monocular camera is fixedly installed behind the base, located behind the semi-transparent screen. The transmitting unit and the monocular camera are respectively located on both sides of the semi-transparent screen, so that the monocular camera can capture the laser spot formed on the semi-transparent screen by the laser beam of the transmitting unit.

[0044] The present invention also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the visual pose estimation method as described above.

[0045] The aforementioned visual pose estimation method, system, laser device, and electronic equipment acquire two different features of the target object and set up a transmitting unit and a receiving unit on each feature, respectively, so that the receiving unit receives the laser beam emitted from the transmitting unit. Subsequently, a connected coordinate system corresponding to the transmitting and receiving units is constructed and acquired. An image of laser spots formed by multiple laser beams on the receiving unit is acquired using an image acquisition device, and the quantized representation of the laser spots in the corresponding image coordinate system is transformed into the corresponding representation equation in the connected coordinate system through homography transformation. An optimization objective function is established between the connected coordinate systems corresponding to the transmitting and receiving units, and this objective function is transformed into a relationship between corresponding matrices and column vectors using the Kronecker product. Then, the corresponding nonlinear equations are transformed into a linear equation system using known elements in the matrices and column vectors, and the solution to this linear equation system is obtained through calculation. Finally, the gradient matrix of the desired spinor is calculated and obtained using a spinor mathematical tool, and the solution to the linear equation system is substituted into the solution through Newton-Raphson iteration to obtain the optimal solution for the two target features of the target object pose estimation. This method, aided by the laser beam of the transmitting unit, improves the stability of the measurement and expands the measurement range to a certain extent. The design of arranging the transmitting and receiving units on both sides of the docking feature allows the server to obtain the pose estimation result of the object under test by transforming the coordinates of the laser spot between different coordinate systems, which reduces the consumption of computing resources and improves the computing efficiency. Attached Figure Description

[0046] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0047] Figure 1 This is one of the flowcharts of the visual pose estimation method provided by the present invention;

[0048] Figure 2 This is the second schematic diagram of the visual pose estimation method provided by the present invention;

[0049] Figure 3 This is the third schematic diagram of the visual pose estimation method provided by the present invention;

[0050] Figure 4 This is the fourth schematic diagram of the visual pose estimation method provided by the present invention;

[0051] Figure 5 This is the fifth schematic diagram of the visual pose estimation method provided by the present invention;

[0052] Figure 6 This is the sixth schematic diagram of the visual pose estimation method provided by the present invention;

[0053] Figure 7 This is the seventh schematic diagram of the visual pose estimation method provided by the present invention;

[0054] Figure 8 The measurement model and measurement principle diagram of the visual pose estimation method provided in the specific embodiments of the present invention;

[0055] Figure 9 A schematic diagram of the transmitting and receiving units of the visual pose estimation method provided in a specific embodiment of the present invention;

[0056] Figure 10 This is a schematic diagram illustrating the image homography transformation principle of the receiving unit in a specific embodiment of the visual pose estimation method provided by the present invention.

[0057] Figure 11 This is a schematic diagram of the unit network layout of the visual pose estimation method in a specific embodiment of the present invention;

[0058] Figure 12 A schematic diagram of the visual pose estimation system provided by the present invention;

[0059] Figure 13 An internal structural diagram of the computer device provided by the present invention. Attached image description:

[0061] 100. Transmitting unit; 110. Laser emitter; 111. Laser beam; 120. Attitude adjustment device; 200. Receiving unit; 210. Semi-transparent screen; 220. Base; 230. Monocular camera; 300. Fixed end of ship section; 400. Docking end of ship section. Detailed Implementation

[0062] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0063] The following is combined Figures 1-13 The present invention describes a visual pose estimation method, system, laser device, and electronic device.

[0064] like Figure 1 As shown, in one embodiment, a visual pose estimation method includes the following steps:

[0065] Step S110: Obtain the first feature and the second feature of the object to be tested. The first feature and the second feature are two different features of the object to be tested.

[0066] Specifically, the server obtains two different test features of the object under test.

[0067] Step S120: Obtain the connected coordinate system corresponding to the transmitting unit and the receiving unit respectively. The transmitting unit is used to transmit a laser beam to the receiving unit, and the receiving unit is used to receive the laser beam. The transmitting unit and the receiving unit are located on the first feature and the second feature respectively.

[0068] Specifically, a transmitting unit and a receiving unit are respectively arranged on two different features of the object under test. The transmitting unit is used to emit laser beams to the receiving unit, and the receiving unit is used to receive the laser beams emitted from the transmitting unit. The server obtains the corresponding connected coordinate system of the transmitting unit and the receiving unit based on their positions.

[0069] Step S130: Acquire an image of laser spots formed by multiple laser beams using an image acquisition device, and transform the quantitative representation of the laser spots in the image coordinate system into a representation equation in the connected coordinate system using homography transformation.

[0070] Specifically, the server acquires an image of laser spots formed by multiple laser beams on the semi-transparent screen of the receiving unit through an image acquisition device, and transforms the quantitative representation of the laser spots in the image coordinate system into a representation equation in the connected coordinate system through homography transformation.

[0071] Step S140: Establish a mapping model for the relative pose transformation between the laser beam direction and the laser spot position to the corresponding coordinate systems of the transmitting and receiving units, respectively, to obtain a nonlinear equation set, and extract all variables and coefficient matrices through the Kronecker product.

[0072] Specifically, based on the equation representing the connected coordinate system in step S130, the server establishes a mapping model of the relative pose transformation between the laser beam direction and the laser spot position to the connected coordinate systems corresponding to the transmitting unit and the receiving unit, respectively, to obtain a nonlinear equation set, and extracts all variables and coefficient matrices through the Kronecker product.

[0073] Step S150: The nonlinear equation system is transformed into a linear equation system between the corresponding coefficient matrix and the column vector containing all unknown elements, and the solution of the linear equation system is obtained by calculation as the relative pose estimate.

[0074] Specifically, the server transforms the corresponding nonlinear equation system into a linear equation system using known elements in the matrix and column vectors, and obtains the solution of the linear equation system through calculation, which serves as the relative pose estimate.

[0075] Step S160: The relative pose between the two connected coordinate systems is characterized by mathematical tools such as spinors and Lie groups. The value of the nonlinear equation system is used as the objective function. The gradient matrix of the relative pose spinor of the objective function is calculated and obtained. The relative pose estimate is substituted into the equation through Newton-Raphson iteration to obtain the optimal solution for the pose estimate of the object under test.

[0076] Specifically, the server uses mathematical tools such as spinors and Lie groups to characterize the relative pose between two connected coordinate systems. It takes the value of the nonlinear equation system as the objective function, calculates and obtains the gradient matrix of the relative pose spinor of the objective function, and substitutes the relative pose estimate into it through Newton-Raphson iteration to obtain the optimal solution for the pose estimate of the object under test.

[0077] The aforementioned visual pose estimation method acquires two distinct features of the target object and sets up transmitting and receiving units on these features, respectively, so that the receiving unit receives the laser beam emitted from the transmitting unit. Subsequently, a connected coordinate system corresponding to the transmitting and receiving units is constructed and acquired. An image of laser spots formed by multiple laser beams on the receiving unit is acquired using an image acquisition device. The quantized representation of the laser spots in the corresponding image coordinate system is transformed into a corresponding representation equation in the connected coordinate system using homography transformation. An optimization objective function is established between the connected coordinate systems corresponding to the transmitting and receiving units, and this objective function is transformed into a relationship between corresponding matrices and column vectors using the Kronecker product. Then, the corresponding nonlinear equations are transformed into a linear equation system using known elements in the matrices and column vectors. The solution to this linear equation system is obtained through calculation. Finally, the gradient matrix of the desired spinor is calculated using a spinor mathematical tool, and the solution to the linear equation system is substituted into the solution using Newton-Raphson iteration to obtain the optimal solution for the pose estimation of the target object's two features. This method, aided by the laser beam of the transmitting unit, improves the stability of the measurement and expands the measurement range to a certain extent. The design of arranging the transmitting and receiving units on both sides of the docking feature allows the server to obtain the pose estimation result of the object under test by transforming the coordinates of the laser spot between different coordinate systems, which reduces the consumption of computing resources and improves the computing efficiency.

[0078] like Figure 2 As shown, in one embodiment, the visual pose estimation method provided by the present invention acquires an image of laser spots formed by multiple laser beams through an image acquisition device, and transforms the quantized representation of the laser spots in the image coordinate system into a representation equation in a connected volume coordinate system through homography transformation, including the following steps:

[0079] Step S132: Based on the distribution of laser spots on the receiving unit, acquire laser spot images through an image acquisition device.

[0080] Specifically, the server acquires images of the laser spots using an image acquisition device based on the distribution of the laser spots on the receiving unit.

[0081] Step S134: Construct an image coordinate system based on the laser spot image, and obtain the first coordinates of the laser spot through machine vision. The first coordinates are the coordinates of the laser spot in the image coordinate system.

[0082] Specifically, the server constructs an image coordinate system based on the laser spot image and obtains the coordinates of the laser spot in the image coordinate system, i.e., the first coordinates, through machine vision.

[0083] like Figure 3As shown, in one embodiment, the visual pose estimation method provided by the present invention acquires an image of laser spots formed by multiple laser beams through an image acquisition device, and transforms the quantized representation of the laser spots in the image coordinate system into a representation equation in a connected volume coordinate system through homography transformation, and further includes the following steps:

[0084] Step S136: Convert the first coordinates into the first characterization equation through homography transformation, and obtain the first coordinate transformation matrix corresponding to the first characterization equation.

[0085] Specifically, the server transforms the first coordinates into the corresponding representation equation, i.e., the first representation equation, through homography transformation, and obtains the coordinate transformation matrix corresponding to the first representation equation, i.e., the first coordinate transformation matrix.

[0086] Step S138: Obtain the second representation equation of the laser spot in the second connected coordinate system through the second coordinate transformation, and obtain the second coordinate transformation matrix corresponding to the second representation equation. The second coordinate transformation matrix is ​​the pose transformation matrix between the two connected coordinate systems.

[0087] Specifically, the server obtains the second representation equation of the laser spot in the second connected coordinate system through the second coordinate transformation, and obtains the second coordinate transformation matrix corresponding to the second representation equation. The second coordinate transformation matrix is ​​the pose transformation matrix between the two connected coordinate systems.

[0088] like Figure 4 As shown, in one embodiment, the visual pose estimation method provided by the present invention acquires an image of laser spots formed by multiple laser beams through an image acquisition device, and transforms the quantized representation of the laser spots in the image coordinate system into a representation equation in a connected volume coordinate system through homography transformation, followed by the following steps:

[0089] Step S410: Obtain the elements in the first and second characterization equations through calibration. The elements are used to construct a nonlinear equation system.

[0090] Specifically, the server obtains elements from the first and second characterization equations through calibration, and constructs the corresponding nonlinear equation system using these elements.

[0091] Step S420: Based on the correspondence between the first and second characterization equations, establish a relative pose calculation model between the first and second connected coordinate systems to obtain the optimization objective function for the pose estimation of the object under test.

[0092] Specifically, based on the correspondence between the first and second characterization equations, the server establishes a relative pose calculation model between the connected coordinate systems corresponding to the transmitting and receiving units, that is, the mapping relationship between the first and second connected coordinate systems, in order to obtain the optimization objective function for estimating the pose of the target object corresponding to the target feature. This optimization objective function can also be called the constraint function.

[0093] like Figure 5 As shown, in one embodiment, the visual pose estimation method provided by the present invention establishes a mapping model of the relative pose transformation between the laser beam direction and the laser spot position to the corresponding coordinate systems of the transmitting unit and the receiving unit, respectively, to obtain a nonlinear equation set, and extracts all variables and coefficient matrices through the Kronecker product, including the following steps:

[0094] Step S142: Obtain the constraint sub-equations introduced by the laser beam, and eliminate the linearly dependent equations through transformation to obtain the constraint target with full rank coefficients.

[0095] Specifically, the server obtains the constraint sub-equations introduced by the laser beam emitted by the transmitting unit, and eliminates the linearly dependent equations by transforming the constraint sub-equations to obtain the constraint target with full rank coefficients.

[0096] Step S144: Based on the constrained objective, the optimization objective function is transformed into a relational expression between matrices and column vectors through the Kronecker product.

[0097] Specifically, based on the constraint objective obtained in step S142, the server transforms the optimization objective function into a relational expression between the corresponding matrix and column vectors through the Kronecker product.

[0098] like Figure 6 As shown, in one embodiment, the visual pose estimation method provided by the present invention further includes the following steps:

[0099] Step S610: Based on the connected coordinate system, obtain the relative pose of the first feature and the second feature, and describe the relative pose as the third coordinate transformation matrix.

[0100] Specifically, the server obtains the relative pose relationship between the first feature and the second feature based on the connected coordinate system corresponding to the transmitting unit and the receiving unit, and describes the relative pose relationship as the corresponding coordinate transformation matrix, namely the third coordinate transformation matrix.

[0101] Step S620: The third coordinate transformation matrix is ​​characterized by Lie groups and spinors to obtain the fourth coordinate transformation matrix corresponding to the spinor axis of the transformation from the first feature to the second feature.

[0102] Specifically, the server characterizes the third coordinate transformation matrix using Lie groups and spinors to obtain the coordinate transformation matrix corresponding to the spinor axis of the transformation from the first feature to the second feature, which is the fourth coordinate transformation matrix.

[0103] like Figure 7 As shown, in one embodiment, the visual pose estimation method provided by the present invention transforms a nonlinear system of equations into a linear system of equations between a corresponding coefficient matrix and a column vector containing all unknown elements, and obtains the solution of the linear system of equations as a relative pose estimate, including the following steps:

[0104] Step S152: Obtain the general solution of the linear equation system by null space decomposition.

[0105] Specifically, the server obtains the general solution of the linear equation system through null space decomposition.

[0106] Step S154: Calculate and obtain a particular solution of the linear equation system by combining the orthogonality constraint of the fourth coordinate transformation matrix.

[0107] Specifically, based on step S152, the server calculates and obtains a particular solution to the linear equation system by combining the orthogonality constraint of the fourth coordinate transformation matrix.

[0108] See Figures 8 to 9 As shown, in a specific embodiment, the present invention provides a visual pose estimation method, including a transmitting unit 100 and a receiving unit 200, respectively fixed on two features to be measured. This method can obtain the relative pose of the transmitting unit 100 and the receiving unit 200, thus obtaining the relative pose between the objects to be measured, and describing it using a coordinate transformation matrix. Further detailed and explicit characterization of the relative pose between the objects to be measured is achieved using Lie groups and spinors. The transmitting unit 100 is a laser transmitting device that controls the angle of the laser beam 111 by adjusting the emission direction of the laser emitter 110 through an attitude adjustment device 120. With the laser emission point fixed, the emission direction of the laser beam 111 contains only two degrees of freedom. The receiving unit 200 includes a semi-transparent screen 210 for receiving and imaging laser spots. A monocular camera 230 is fixed at a certain distance behind the semi-transparent screen 210, ensuring that the entire semi-transparent screen 210 is evenly distributed within the image frame of the monocular camera 230.

[0109] When measuring the object under test, the transmitting unit 100 and the receiving unit 200 are fixed on the feature to be measured, and their relative positions are adjusted so that the laser emitted by the transmitting unit 100 can fall on the semi-transparent screen 210 of the receiving unit 200, establishing a connected coordinate system for each of the transmitting unit 100 and the receiving unit 200. A series of laser beams 111 (at least four) with specific directions are acquired by controlling the input of the transmitting unit 100. The direction of the laser beams can be quantified in the connected coordinate system of the transmitting unit. When the laser beams 111 intersect the semi-transparent screen 210, they appear as laser spots on the semi-transparent screen 210. The coordinates of these laser spots in the image coordinate system of the monocular camera 230 can be quantified. Homography transformation can convert these spots into a representation in the connected coordinate system of the receiving unit. The coordinates of the laser spots in the connected coordinate system of the transmitting unit can also be represented according to the direction of the laser beams 111, but a depth-direction scaling factor is lacking. The influence of the laser length can be eliminated by cross-product of the direction itself. After obtaining the representation of the laser spot in two connected coordinate systems, constraint equations relating the coordinate systems of the transmitting and receiving units can be established based on the correspondence. This constraint equation serves as the objective function for estimating the pose of the target object. Since the rank of the antisymmetric matrix corresponding to the cross product operator is 2, only two constraint equations can be introduced from a single laser beam. Linearly dependent equations can be eliminated through mathematical transformations. N laser beams in different directions can yield 2^n constraint equations, which, when combined, form a full-rank constraint objective.

[0110] The Kronecker product transforms the objective function into a matrix and column vector multiplication form, where all matrix elements are known and the column vector elements are elements of the coordinate transformation matrix to be determined, thus converting the nonlinear equations into a linear equations system. For the case where four or more laser beams 111 intersect the semi-transparent screen 210 in different directions, the rank of the coefficient matrix is ​​8. The general solution of this linear equations system is obtained through null space decomposition. Further, by incorporating the orthogonality constraint of the coordinate transformation matrix, a particular solution to the linear equations system can be obtained. This particular solution deviates to some extent from the true value and can be used as an initial value for the optimization problem. Subsequently, using the mathematical tool of spinors, the gradient matrix of the original objective function with respect to the spinor to be determined can be calculated. Substituting the initial values ​​obtained through the Kronecker product using the Newton-Raphson iteration yields the optimal solution for estimating the pose of the object under test.

[0111] In this embodiment, during the detection process, the transmitting unit 100 and the receiving unit 200 are respectively fixed on two features to be measured, and the connected coordinate systems of the two features are represented as {S} and {T}, respectively. The base coordinate systems of the transmitting unit 100 and the receiving unit 200 are aligned with the connected coordinate system of the object under test. The relative pose of the transmitting unit 100 and the receiving unit 200 is the relative pose between the objects under test {S} and {T}, which can be represented by a coordinate transformation matrix. The description is as follows, where R is the direction cosine matrix between the two objects' body coordinate systems, and p represents the displacement deviation between the two objects. This relative pose is characterized in detail using Lie groups and spinors, as shown below:

[0112]

[0113] Where, ξ=[ω T v T ] T ∈se(3) is the spinor axis of the transformation from {S} to {T}.

[0114] In this embodiment, the transmitting unit 100 has an attitude adjustment device 120 that can control the emission angle. A laser emitter 110 is fixed on its moving platform. With the emission point fixed, the emission direction of the laser beam 111 has only two degrees of freedom. The receiving unit 200 includes a semi-transparent screen 210 for receiving and imaging the laser spot. A monocular camera 230 is fixed on the side of the semi-transparent screen 210 away from the transmitting unit 100. The semi-transparent screen 210 and the monocular camera 230 are respectively fixed to both ends of the base 220 to ensure that the semi-transparent screen 210 is fully and evenly distributed in the camera's imaging field. The relative positions are adjusted so that the laser beam 111 emitted by the transmitting unit 100 can fall on the semi-transparent screen 210 of the receiving unit 200, thereby establishing the connected coordinate systems {S} and {T} of the transmitting unit 100 and the receiving unit 200, respectively.

[0115] The input to the control emission unit 100 is controlled, and a series of laser beams 111 (at least 4) with specific directions are acquired. The laser spots are quantitatively characterized in the coordinate system {S} of the emission unit as follows:

[0116] r i =l i u i +r0∈R 3 .

[0117] Among them, l i Let u be the length of the laser beam. i Let be the direction of the laser beam in the coordinate system {S} of the transmitting unit.

[0118] The laser beam 111 intersects with the receiving unit 200 and appears as a laser spot on the semi-transparent screen 210. Its coordinates ρ in the image coordinate system of the monocular camera 230 are... i =[X i Y i ] T This can be obtained through machine vision. Combined with... Figure 10 As shown, homography transformation can convert the coordinates of the laser spot in the image coordinate system into a representation of those coordinates in the body coordinate system {T} of the receiving unit. That is, homography transformation is:

[0119]

[0120] Here, H is the homography transformation matrix, which maps the coordinates in the image coordinate system to coordinates in {T}. Generally, to simplify calculations, H can be set to... 33 =1.

[0121] Each element in H can be obtained through prior calibration. Given n pairs of points with known accurate coordinates, substituting them into the above equation and organizing the coefficients in H into a separate column vector, we can obtain a system of linear equations, namely Ch = 0, where...

[0122]

[0123] h = [H] 11 H 12 H 13 H 21 H 22 H 23 H 31 H 32 H 33 ] T ∈R 9 .

[0124] Given at least four sets of associated coordinates, the above system of linear equations can be easily solved, thus obtaining the homography transformation matrix of the receiving unit. The laser spot position obtained through the receiving unit can be represented in {S} by coordinate transformation, as follows:

[0125] r i =R t ρ i +p t ∈R 3 .

[0126] Two about r i The constraint equations can be obtained by combining the equations:

[0127] r i =R t ρ t +p t=l i u i +r0∈R 3 .

[0128] Through the direction of the laser beam u i The coordinates of the laser spot in the coordinate system {S} of the emitting unit can also be characterized, but a scale factor l in the depth direction is lacking. i The influence of laser length can be eliminated by the cross product direction itself, and the constraint equations can be rearranged as follows:

[0129] u i ×(R t ρ i +p t -r0)=0 3×1 i = 1, 2, ..., n.

[0130] in,

[0131] Due to the antisymmetric matrix corresponding to the cross product operator (×) The rank of the laser beam is 2, so a laser beam u i Only two constraint equations can be introduced. Through mathematical transformation, the linearly dependent equations can be eliminated, i.e.

[0132]

[0133] in, And V i It's U i The zero space.

[0134] By left multiplication n laser beams in different directions can yield 2n constraint equations, which can be combined to obtain a constraint target with full rank coefficients:

[0135]

[0136] The objective function is transformed into a matrix and column vector multiplication form using the Kronecker product, i.e.,

[0137]

[0138] in,

[0139] All elements of matrix A are known, and column vector g v The elements in the matrix are the coordinate transformation matrix g to be determined. sThe elements in the matrix are used to transform the nonlinear equations into a linear system. For the case where four or more laser beams in different directions intersect the semi-transparent screen, the rank of the coefficient matrix A is 8. The general solution of this linear system can be obtained through null space decomposition, which is...

[0140] g v =λ1g1+λ2g2+λ3g3+λ4g4.

[0141] Further, consider the orthogonality constraint of the coordinate transformation matrix:

[0142]

[0143]

[0144]

[0145] in, Represents g s The first, second, and third columns.

[0146] Therefore, a particular solution ξ0 of the linear equation system can be obtained. This solution deviates to some extent from the true value and can be substituted as the initial value for the optimization problem. Subsequently, using the mathematical tools of spinors, the gradient matrix of the original objective function with respect to the desired spinor can be calculated, as follows:

[0147]

[0148] Each item in this block matrix can be represented as:

[0149]

[0150] Where r0 is the coordinate of the center of the ball joint, which is a constant. Further, using a spinor derivative tool, the above equation can be expanded as follows:

[0151]

[0152] Among them, A ξ The calculation of matrices is well-defined in spinor algebra and can be easily obtained.

[0153] Finally, by substituting the initial value ξ0 obtained through the Kronecker product using the Newton-Raphson iteration, the optimal solution to the pose estimation problem of the object under test can be obtained:

[0154]

[0155] In this embodiment, the transmitting unit is a circular laser emitter, and its emission direction can be controlled by an attitude adjustment device. The receiving unit consists of a semi-transparent screen and a monocular camera, both fixed on the same frame. The signals and control of the transmitting and receiving units can be easily controlled by a PC, making the measurement device simple in structure, easy to operate, and flexibly arranged. Therefore, this visual pose estimation method can be used alone to measure the relative pose of two points, or it can be networked for multi-point collaborative measurement. Although this method uses a monocular camera, it can achieve full-degree-of-freedom measurement, providing a vision-based relative pose estimation method. With the laser assistance of the transmitting unit, the six-degree-of-freedom relative pose relationship between two objects can be measured. The laser beam assistance improves the stability of the measurement and expands the measurement range, while reducing the measurement cost, which is conducive to the promotion and application of computers in industries and other fields. In addition, this method is not limited to single-station measurement; see [link to relevant documentation]. Figure 11 As shown, in the assembly and segment joining stages of shipbuilding, there are numerous and varied docking features such as T-shaped profiles on the docking surfaces. During docking and joining, these features move relative to each other from a considerable distance until they are fully aligned, making it difficult for traditional measurement methods to track and measure the entire process. The visual pose estimation method provided by this invention can simultaneously deploy multiple sets of transmitting and receiving units on both sides of the responding docking features to form a measurement network. This network can track and measure the entire process for each docking end 400 of the ship section and summarize the data for the central control system, guiding the attitude adjustment of the fixed end 300 of the ship section.

[0156] The visual pose estimation system provided by the present invention is described below. The visual pose estimation system described below can be referred to in correspondence with the visual pose estimation method described above.

[0157] like Figure 12 As shown, in one embodiment, a visual pose estimation system includes a first acquisition module 1210, a second acquisition module 1220, a quantization representation module 1230, a function processing module 1240, a calculation processing module 1250, and an optimal solution acquisition module 1260.

[0158] The first acquisition module 1210 is used to acquire the first feature and the second feature of the object to be tested, wherein the first feature and the second feature are two different features of the object to be tested.

[0159] The second acquisition module 1220 is used to acquire the connected coordinate systems corresponding to the transmitting unit and the receiving unit respectively. The transmitting unit is used to transmit a laser beam to the receiving unit, and the receiving unit is used to receive the laser beam. The transmitting unit and the receiving unit are located on the first feature and the second feature respectively.

[0160] The quantization characterization module 1230 is used to acquire an image of laser spots formed by multiple laser beams through an image acquisition device, and to transform the quantization characterization of the laser spots in the image coordinate system into a characterization equation in the connected body coordinate system through homography transformation.

[0161] The function processing module 1240 is used to establish a mapping model of the relative pose transformation between the laser beam direction and the laser spot position to the corresponding coordinate systems of the transmitting unit and the receiving unit, respectively, to obtain a nonlinear equation system, and extract all variables and coefficient matrices through the Kronecker product.

[0162] The computational processing module 1250 is used to convert the nonlinear equation system into a linear equation system between the corresponding coefficient matrix and the column vector containing all unknown elements, and to obtain the solution of the linear equation system as the relative pose estimate.

[0163] The optimal solution acquisition module 1260 is used to characterize the relative pose between two connected coordinate systems using mathematical tools of spinors and Lie groups. It takes the value of the nonlinear equation system as the objective function, calculates and obtains the gradient matrix of the relative pose spinor of the objective function, and substitutes the relative pose estimate into it through Newton-Raphson iteration to obtain the optimal solution for the pose estimation of the object under test.

[0164] In this embodiment, the quantization representation module of the visual pose estimation system provided by the present invention is specifically used for:

[0165] Based on the distribution of laser spots on the receiving unit, an image acquisition device acquires images of the laser spots.

[0166] An image coordinate system is constructed based on the laser spot image, and the first coordinate of the laser spot is obtained through machine vision. The first coordinate is the coordinate of the laser spot in the image coordinate system.

[0167] In this embodiment, the quantization representation module of the visual pose estimation system provided by the present invention is further used for:

[0168] The first coordinates are transformed into the first characterization equation through homography transformation, and the first coordinate transformation matrix corresponding to the first characterization equation is obtained.

[0169] The second coordinate transformation is used to obtain the second characterization equation of the laser spot in the second connected coordinate system, and the second coordinate transformation matrix corresponding to the second characterization equation is obtained. The second coordinate transformation matrix is ​​the pose transformation matrix between the two connected coordinate systems.

[0170] In this embodiment, the visual pose estimation system provided by the present invention further includes a function establishment module, used for:

[0171] Elements in the first and second characterization equations are obtained through calibration, and these elements are used to construct a system of nonlinear equations.

[0172] Based on the correspondence between the first and second characterization equations, a relative pose calculation model between the first and second connected coordinate systems is established to obtain the optimization objective function for the pose estimation of the object under test.

[0173] In this embodiment, the visual pose estimation system provided by the present invention has a function processing module specifically used for:

[0174] Obtain the constraint sub-equations introduced by the laser beam, and eliminate the linearly dependent equations through transformation to obtain the constraint target with full rank coefficients.

[0175] Based on the constrained objective, the optimization objective function is transformed into a relationship between matrices and column vectors through the Kronecker product.

[0176] In this embodiment, the visual pose estimation system provided by the present invention further includes a matrix representation module, used for:

[0177] Based on the connected coordinate system, the relative pose of the first feature and the second feature is obtained, and the relative pose is described as the third coordinate transformation matrix.

[0178] The third coordinate transformation matrix is ​​characterized by Lie groups and spinors to obtain the fourth coordinate transformation matrix corresponding to the spinor axis of the transformation from the first eigenvalue to the second eigenvalue.

[0179] In this embodiment, the visual pose estimation system provided by the present invention has a calculation and processing module specifically used for:

[0180] The general solution of a system of linear equations can be obtained by null space decomposition.

[0181] By combining the orthogonality constraint of the fourth coordinate transformation matrix, a particular solution of the linear equation system is calculated and obtained.

[0182] In one embodiment, the present invention also provides a laser device for visual pose estimation, comprising:

[0183] The transmitting unit includes a laser emitter and an attitude adjustment device. The laser emitter is fixedly installed on the free end of the attitude adjustment device so that the angle of the laser beam emitted by the laser emitter can be adjusted by the attitude adjustment device.

[0184] The receiving unit includes a semi-transparent screen, a monocular camera, and a base. The semi-transparent screen is fixedly installed in front of the base, and the monocular camera is fixedly installed behind the base and located behind the semi-transparent screen. The transmitting unit and the monocular camera are respectively located on both sides of the semi-transparent screen, so that the monocular camera can capture the laser spot image formed on the semi-transparent screen by the laser beam emitted by the transmitting unit from one side of the semi-transparent screen.

[0185] Figure 13 This example illustrates a schematic diagram of the physical structure of an electronic device, which can be a smart terminal. Its internal structure diagram can be as follows: Figure 13 As shown. The electronic device includes a processor, memory, and a network interface connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The network interface is used to communicate with external terminals via a network connection. When the computer program is executed by the processor, it implements a visual pose estimation method, which includes:

[0186] Obtain the first feature and the second feature of the object to be tested. The first feature and the second feature are two different features of the object to be tested.

[0187] Obtain the connected coordinate system corresponding to the transmitting unit and the receiving unit respectively. The transmitting unit is used to transmit a laser beam to the receiving unit, and the receiving unit is used to receive the laser beam. The transmitting unit and the receiving unit are located on the first feature and the second feature respectively.

[0188] The image acquisition device acquires an image of laser spots formed by multiple laser beams, and the quantitative representation of the laser spots in the image coordinate system is transformed into a representation equation in the connected body coordinate system through homography transformation.

[0189] A mapping model for the relative pose transformation between the laser beam direction and the laser spot position to the corresponding coordinate systems of the transmitting and receiving units is established to obtain a nonlinear equation set. All variables and coefficient matrices are then extracted using the Kronecker product.

[0190] The nonlinear equation system is transformed into a linear equation system between the corresponding coefficient matrix and the column vector containing all unknown elements, and the solution of the linear equation system is obtained by calculation as the relative pose estimate.

[0191] The relative pose between two connected coordinate systems is characterized using mathematical tools such as spinors and Lie groups. The value of the nonlinear equation system is used as the objective function. The gradient matrix of the relative pose spinor of the objective function is calculated and obtained. The relative pose estimate is substituted into the equation through Newton-Raphson iteration to obtain the optimal solution for the pose estimate of the object under test.

[0192] Those skilled in the art will understand that Figure 13 The structure shown is merely a block diagram of a portion of the structure related to the present invention and does not constitute a limitation on the electronic device to which the present invention is applied. A specific electronic device may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0193] On the other hand, the present invention also provides a computer storage medium storing a computer program, which, when executed by a processor, implements a visual pose estimation method, the method comprising:

[0194] Obtain the first feature and the second feature of the object to be tested. The first feature and the second feature are two different features of the object to be tested.

[0195] Obtain the connected coordinate system corresponding to the transmitting unit and the receiving unit respectively. The transmitting unit is used to transmit a laser beam to the receiving unit, and the receiving unit is used to receive the laser beam. The transmitting unit and the receiving unit are located on the first feature and the second feature respectively.

[0196] The image acquisition device acquires an image of laser spots formed by multiple laser beams, and the quantitative representation of the laser spots in the image coordinate system is transformed into a representation equation in the connected body coordinate system through homography transformation.

[0197] A mapping model for the relative pose transformation between the laser beam direction and the laser spot position to the corresponding coordinate systems of the transmitting and receiving units is established to obtain a nonlinear equation set. All variables and coefficient matrices are then extracted using the Kronecker product.

[0198] The nonlinear equation system is transformed into a linear equation system between the corresponding coefficient matrix and the column vector containing all unknown elements, and the solution of the linear equation system is obtained by calculation as the relative pose estimate.

[0199] The relative pose between two connected coordinate systems is characterized using mathematical tools such as spinors and Lie groups. The value of the nonlinear equation system is used as the objective function. The gradient matrix of the relative pose spinor of the objective function is calculated and obtained. The relative pose estimate is substituted into the equation through Newton-Raphson iteration to obtain the optimal solution for the pose estimate of the object under test.

[0200] In another aspect, a computer program product or computer program is provided, comprising computer instructions stored in a computer-readable storage medium. A processor of an electronic device reads the computer instructions from the computer-readable storage medium, and when the processor executes the computer instructions, it implements a visual pose estimation method, the method comprising:

[0201] Obtain the first feature and the second feature of the object to be tested. The first feature and the second feature are two different features of the object to be tested.

[0202] Obtain the connected coordinate system corresponding to the transmitting unit and the receiving unit respectively. The transmitting unit is used to transmit a laser beam to the receiving unit, and the receiving unit is used to receive the laser beam. The transmitting unit and the receiving unit are located on the first feature and the second feature respectively.

[0203] The image acquisition device acquires an image of laser spots formed by multiple laser beams, and the quantitative representation of the laser spots in the image coordinate system is transformed into a representation equation in the connected body coordinate system through homography transformation.

[0204] A mapping model for the relative pose transformation between the laser beam direction and the laser spot position to the corresponding coordinate systems of the transmitting and receiving units is established to obtain a nonlinear equation set. All variables and coefficient matrices are then extracted using the Kronecker product.

[0205] The nonlinear equation system is transformed into a linear equation system between the corresponding coefficient matrix and the column vector containing all unknown elements, and the solution of the linear equation system is obtained by calculation as the relative pose estimate.

[0206] The relative pose between two connected coordinate systems is characterized using mathematical tools such as spinors and Lie groups. The value of the nonlinear equation system is used as the objective function. The gradient matrix of the relative pose spinor of the objective function is calculated and obtained. The relative pose estimate is substituted into the equation through Newton-Raphson iteration to obtain the optimal solution for the pose estimate of the object under test.

[0207] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. This computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided by this invention can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory.

[0208] By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0209] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0210] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A method of visual pose estimation, the method comprising: The method comprises: obtaining first and second features of a to-be-measured object, the first and second features being different two to-be-measured features of the to-be-measured object; obtaining a conjoined coordinate system corresponding to a transmitting unit and a receiving unit respectively, the transmitting unit being configured to transmit a laser beam to the receiving unit, the receiving unit being configured to receive the laser beam, and the transmitting unit and the receiving unit being located on the first and second features respectively; obtaining, by an image acquisition device, a plurality of laser spot images formed by the laser beams, and converting, by a homographic transformation, a quantitative representation of the laser spots in an image coordinate system into a representation equation in the conjoined coordinate system; establishing a mapping model of a relative pose transformation between a laser beam direction and a laser spot position and the conjoined coordinate system corresponding to the transmitting unit and the receiving unit respectively, to obtain a nonlinear equation set, and extracting all variables and coefficient matrices by a Kronecker product; converting the nonlinear equation set into a linear equation set between a corresponding coefficient matrix and a column vector containing all unknown elements, and obtaining a solution of the linear equation set by calculation as a relative pose estimation value; representing the relative pose between the two conjoined coordinate systems by a mathematical tool of a screw and a Lie group, taking a value of the nonlinear equation set as an objective function, calculating and obtaining a gradient matrix of a relative pose screw of the objective function, and substituting the relative pose estimation value into the Newton-Raphson iteration to obtain an optimal solution of the pose estimation of the to-be-measured object.

2. The visual pose estimation method of claim 1, wherein, The conjoined coordinate system comprises a first conjoined coordinate system and a second conjoined coordinate system, the first and second conjoined coordinate systems being conjoined coordinate systems of the transmitting unit and the receiving unit respectively; The method comprises: obtaining, by an image acquisition device, a plurality of laser spot images formed by the laser beams, and converting, by a homographic transformation, a quantitative representation of the laser spots in an image coordinate system into a representation equation in the conjoined coordinate system; based on the distribution of the laser spots on the receiving unit, obtaining a laser spot image by an image acquisition device; 3. The visual pose estimation method of claim 2, wherein, based on the laser spot image, constructing an image coordinate system, and obtaining a first coordinate of the laser spot by machine vision, the first coordinate being a coordinate of the laser spot in the image coordinate system. The representation equation comprises a first representation equation and a second representation equation, the first and second representation equations being representation equations corresponding to the first and second conjoined coordinate systems respectively; The method further comprises: converting the first coordinate into the first representation equation by the homographic transformation, and obtaining a first coordinate transformation matrix corresponding to the first representation equation; obtaining a second representation equation of the laser spot in the second conjoined coordinate system by a second coordinate transformation, and obtaining a second coordinate transformation matrix corresponding to the second representation equation, the second coordinate transformation matrix being a pose transformation matrix between the two conjoined coordinate systems.

4. The visual pose estimation method of claim 3, wherein, The laser spot images formed by the plurality of laser beams are acquired by an image acquisition device, and a quantitative representation of the laser spot in an image coordinate system is converted into a representation equation in the connected coordinate system by a homographic transformation, and then includes: Elements in the first representation equation and the second representation equation are acquired by calibration, and the elements are used to construct a nonlinear equation set; Based on the corresponding relationship between the first representation equation and the second representation equation, a relative pose calculation model between the first connected coordinate system and the second connected coordinate system is established to obtain an optimization objective function of the pose estimation of the object to be measured.

5. The visual pose estimation method of claim 4, wherein, A mapping model of the relative pose transformation between the laser beam direction and the laser spot position to the connected coordinate system corresponding to the transmitting unit and the receiving unit is established to obtain a nonlinear equation set, and all variables and coefficient matrices are extracted by Kronecker product, including: Constraint sub-equations introduced by the laser beam are acquired, and linearly related equations are eliminated by transformation to obtain a constraint target with full rank of coefficients; Based on the constraint target, the optimization objective function is converted into a relationship between a matrix and a column vector by Kronecker product.

6. The visual pose estimation method of claim 1, wherein, The method further includes: Based on the connected coordinate system, the relative pose of the first feature and the second feature is acquired, and the relative pose is described as a third coordinate transformation matrix; The third coordinate transformation matrix is represented by Lie group and spinor to obtain a fourth coordinate transformation matrix corresponding to the spinor axis of the transformation from the first feature to the second feature.

7. The visual pose estimation method of claim 6, wherein, The nonlinear equation set is converted into a linear equation set between a corresponding coefficient matrix and a column vector containing all unknown elements, and the solution of the linear equation set is obtained by calculation as the relative pose estimation value, including: The general solution of the linear equation set is obtained by the method of null space decomposition; The particular solution of the linear equation set is calculated and obtained in combination with the orthogonal characteristic constraint of the fourth coordinate transformation matrix.

8. A visual pose estimation system, characterized by The system includes: A first acquisition module is configured to acquire a first feature and a second feature of an object to be measured, the first feature and the second feature being different two features of the object to be measured; A second acquisition module is configured to acquire a connected coordinate system corresponding to a transmitting unit and a receiving unit, respectively, the transmitting unit being configured to emit a laser beam to the receiving unit, the receiving unit being configured to receive the laser beam, and the transmitting unit and the receiving unit being located on the first feature and the second feature, respectively; A quantitative representation module is configured to acquire laser spot images formed by a plurality of laser beams by an image acquisition device, and to convert a quantitative representation of the laser spot in an image coordinate system into a representation equation in the connected coordinate system by a homographic transformation; A function processing module is configured to establish a mapping model of the relative pose transformation between the laser beam direction and the laser spot position to the connected coordinate system corresponding to the transmitting unit and the receiving unit, to obtain a nonlinear equation set, and to extract all variables and coefficient matrices by Kronecker product. The computing processing module is configured to convert the nonlinear equation set into a linear equation set between a corresponding coefficient matrix and a column vector containing all unknown elements, and obtain a solution of the linear equation set as a relative pose estimation value through calculation; The optimal solution obtaining module is configured to represent a relative pose between two connected coordinate systems through a mathematical tool of a screw and a Lie group, take a value of the nonlinear equation set as a target function, calculate and obtain a gradient matrix of a relative pose screw of the target function, and substitute the relative pose estimation value into the Newton-Raphson iteration to obtain an optimal solution of the pose estimation of the to-be-measured object.

9. A laser device for visual pose estimation for implementing the method of visual pose estimation according to any one of claims 1 to 7, characterized in that The method comprises the steps of: The transmitting unit comprises a laser transmitter and a posture adjusting device, the laser transmitter is fixedly installed at a free end of the posture adjusting device, and the angle of the laser beam emitted by the laser transmitter is adjusted through the posture adjusting device; The receiving unit comprises a semi-transparent screen, a monocular camera and a base, the semi-transparent screen is fixedly installed in front of the base, the monocular camera is fixedly installed at the back of the base behind the semi-transparent screen, and the transmitting unit and the monocular camera are respectively located on two sides of the semi-transparent screen, so that the monocular camera captures a laser spot formed by the laser beam of the transmitting unit on the semi-transparent screen.

10. An electronic device comprising a memory and a processor, the memory storing a computer program, characterized in that, The processor executes the computer program to implement the steps of the method in any one of claims 1 to 7.

Images