Method and device for calibrating light curved surface of line structure light sensor with self-adaptive curvature

Abstract

The invention discloses a method and a device for calibrating a light curved surface of a line structured light vision sensor with self-adaptive curvature, and relates to the technical field of line structured light sensors, and the method comprises the steps: collecting target images, laser stripe images and temperature readings of a plurality of positions through a camera, a laser and a temperature sensor; then calculating laser optical center coordinates through a geometric projection method; performing triangulation on the laser stripe point cloud by using a camera internal reference matrix obtained by Zhang's calibration, and calculating curvature distribution of the laser stripe point cloud in combination with a mixed curvature model; generating a curvature gradient weighting coefficient according to the curvature gradient modulus length, and iteratively solving an optimal pose and curvature parameter through a closed-loop optimization unit; and finally, thermal drift compensation optimization is triggered according to the real-time temperature, an optimized light curved surface equation is corrected and output, and accurate calibration of the linear structure light sensor is realized. Through hybrid model compensation, closed-loop optimization and thermal compensation output, active compensation of inherent optical aberration of the laser in the calibration process is realized, and an accurate and efficient calibration solution is provided.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of line structured light sensor, and particularly relates to a line structured light vision sensor light curved surface calibration method and device with self-adaptive curvature. BACKGROUND

[0002] Although the line structured light vision sensor has broken through the error limitation of traditional space transformation in the image-light plane coordinate direct mapping level at present, and realizes high calibration accuracy, it is limited by the following fundamental bottlenecks: systematic error in high curvature area, super-difference deviation of curved surface edge and tooth top area caused by inherent curvature of laser, which cannot meet the demand of precision measurement; temperature drift instability, instrument temperature rise caused by long time operation, which causes certain thermal deformation and further amplifies the error; edge distortion artifact, wave ripples in the reconstructed curved surface in the curvature mutation area, which causes distortion of key feature measurement. At the same time, considering the physical dependence of the precision translation mechanism and the complexity of the double-mode image acquisition process, the existing technology faces the triple challenges of precision, stability and efficiency.

[0003] In the prior art, a line structured light sensor calibration method and system with patent number CN116182703B discloses a line structured light sensor calibration method, which comprises the following steps: S1. obtaining the camera intrinsic parameters and distortion coefficients of the sensor; S2. collecting the calibration board image under the irradiation of the external light source and the calibration board image with projected laser stripe, and correcting the distortion; S3. obtaining the movement vector of each translation of the calibration board; S4. solving the planar coordinates of the feature points in the local coordinate system by using the homography matrix; S5. obtaining the three-dimensional coordinates of the feature points in the world coordinate system; S6. reprojecting the three-dimensional coordinates of the feature points in the world coordinate system into planar coordinates by using the principal component analysis method; S7. calculating the homography matrix from the light plane to the image plane. This method realizes coordinate conversion through homography matrix re-projection, although it improves the basic accuracy, but it strictly depends on the precision displacement mechanism, greatly increases the hardware cost, and still has a large error without breaking through the high curvature error bottleneck.

[0004] The patent number CN115451813B discloses a line structure light calibration method, device, electronic equipment and storage medium, which discloses a line structure light calibration method, device, electronic equipment and storage medium, wherein the method comprises: after the line structure light emitter projects at each calibration position, controlling the camera to collect the target calibration board to obtain the line structure light calibration graph of each calibration position; the number of line structure light lines on the target calibration board is less than the number of multi-line line structure light projected by the line structure light emitter; based on the line structure light calibration graph of each calibration position, the target line structure light three-dimensional point is obtained; based on the target line structure light three-dimensional point and the preset reference line structure light surface parameter, the target rotation translation matrix is obtained, and based on the target rotation translation matrix and the reference line structure light surface parameter, the current line structure light surface parameter is calculated. Although this method gets rid of the dependence on large displacement platform and global calibration board, it ignores temperature drift compensation and edge distortion problem, and does not fundamentally reduce the curvature error. SUMMARY

[0005] The first object of the present application is to provide a light surface calibration method for a line structure light sensor with adaptive curvature, aiming to solve the problem of lack of systematic compensation mechanism for laser beam curvature error in existing methods, resulting in a significant decrease in calibration accuracy.

[0006] To solve the above technical problems, a line structure light sensor light surface calibration method with adaptive curvature optimization is provided, comprising the following steps:

[0007] S1, data acquisition, collecting data on different positions of the target by a camera, a laser and a temperature sensor, and synchronously recording real-time temperature T now , obtaining target image and laser stripe image;

[0008] S2, laser light center positioning, performing speckle cross-correlation analysis on the laser stripe image, obtaining the laser projection center coordinates O v =(u v ,v v ) by geometric projection method, and taking it as the reference of the laser coordinate system;

[0009] S3, hybrid curvature modeling, calculating the camera intrinsic matrix K based on the target image by Zhang calibration method, extracting the laser stripe center pixel coordinates [u,v] T from the laser stripe image by Steger algorithm, and performing triangulation to obtain the three-dimensional point cloud coordinates P c of the laser stripe, and then calculating the local curvature distribution Kappa i of the three-dimensional point cloud coordinates P c by central difference method;

[0010] S4. Closed-loop optimization unit, constructing a Gaussian-aberration hybrid curvature model based on the local curvature distribution κ. i Calculate gradient magnitude Generate curvature-sensitive weights w i The initial pose of the laser installation is denoted as [R0, t0]. The optimal pose [R0, t0] is obtained by iteratively solving the Levenberg-Marquardt algorithm. * ,t * With curvature parameters

[0011] S5. Thermal Compensation and Output: Based on the temperature reading, thermal compensation is performed on the curvature parameter, and the thermally compensated curvature distribution is adopted. The equation of the optical surface is reconstructed by controlling the B-spline surface, and the final output is the calibration result and error analysis.

[0012] In one embodiment, the Steger algorithm is applied to the laser stripe images acquired at different locations to obtain the center lines of the laser stripes, and the beginning and end points of each laser stripe center line are taken. Connecting the beginning and end points of the center lines of each laser stripe constructs a straight line L for the laser stripe. i Its normal pixel Then the laser stripe straight line L i There exists any point p on the line that satisfies the equation of the line. It can be seen that the straight lines L of each laser stripe i intersection point p v That is, the projection point of the optical center of the laser onto the image plane, which is obtained by solving the overdetermined equations:

[0013]

[0014] Finally, the coordinates of the laser's optical center in the image plane coordinate system can be obtained, denoted as O. v =(u v ,v v ).

[0015] In one embodiment, the Zhang calibration method is used to solve for the camera intrinsic parameter matrix K, the distortion coefficients dist, and the transformation matrix from the target image coordinate system to the camera coordinate system on the target image:

[0016]

[0017] in Let be a rotation matrix. As the translation matrix, the Steger algorithm is applied to the laser stripe image to obtain the pixel coordinates p of the center point of the laser stripe. stripe, utilize the camera intrinsic matrix K and the distortion coefficients dist to correct the laser stripe center point coordinates p stripe , and then through triangulation, convert the laser stripe center point coordinates p stripe into three-dimensional coordinates P in the camera coordinate system c , and then sort the three-dimensional coordinates P c according to the laser stripe straight line L i in order, and use the central difference method to calculate the local curvature κ i :

[0018]

[0019] where P' i and P" i are the first derivative and second derivative of the three-dimensional coordinates P c , respectively, and a hybrid curvature model is constructed.

[0020] In one embodiment, the hybrid curvature model parameters {R, t, a, b, λ1, λ2, ∑} are initialized, and a Gaussian-aberration hybrid curvature model is constructed:

[0021]

[0022] where R and t are the pose under the curvature, a is the Gaussian distribution amplitude, b is the beam divergence factor, λ1 and λ2 are Zernike aberration coefficients, the parameters are obtained from the factory usage manual of the laser, in addition, ∑ is the spatial covariance matrix, the initial value is obtained from the factory manual, and in the optimization, the spatial distribution of the three-dimensional coordinates P c is dynamically updated, is the Zernike basis function, where x and y represent normalized two-dimensional coordinates on the laser projection plane, and the curvature-sensitive weight is calculated:

[0023]

[0024] where β is an empirical value, set to 0.5, and the initial coordinate [R0, t0] of mechanical installation is taken as the starting point, and a weighted projection error objective function is constructed:

[0025]

[0026] where π(·) is the camera projection function, C(u i ; κ model ) is an adaptive B-spline curve based on the current all curvatures, and then the Levenberg-Marquardt algorithm is used for iterative solution:

[0027] (J T WJ+μI)Δθ=JT Wr;

[0028] where J is the Jacobian matrix, J T is the transpose matrix of the Jacobian matrix, W is the weight matrix, μ is the adaptive damping factor, set to 0.001, I is the unit matrix, Δθ is the parameter increment vector, r is the residual vector, the hybrid curvature model parameters {R, t, a, b, λ1, λ2, ∑} are updated after each iteration, and κ model is recalculated until convergence, and the optimal hybrid curvature model parameters {R * , t * , a * , b * , λ1 * , λ2 * , ∑ *} are finally output.

[0029] In one of the embodiments, the Zernike aberration coefficient dynamic optimization includes distortion compensation constraints:

[0030]

[0031] where is the factory initial value, is the maximum curvature gradient, is the average curvature gradient of the edge region of the surface.

[0032] In one of the embodiments, the real-time temperature T now and the calibration reference temperature The temperature change ΔT = T now - T calib is calculated, and a thermal drift compensation model is applied to update the curvature parameters:

[0033]

[0034] where α is the material thermal expansion coefficient, obtained from the factory usage manual, and the control point distribution density is calculated according to the local curvature κ i :

[0035]

[0036] where is the average curvature, γ is the curvature sensitivity coefficient, set to 0.5, N base is the basic control point number, set to 20, and then the non-uniform node vector is generated based on the updated :

[0037]

[0038] The nodes are then fitted to a B-spline surface using the weighted least squares method.

[0039]

[0040] Where N k,p Using p-th order B-spline basis functions, the final output is the optical surface equation S(u,v) to complete the calibration task.

[0041] In one embodiment, the triangulation employs a binocular vision reconstruction model:

[0042]

[0043] Where f is the camera focal length and d is the distance between the laser plane and the camera optical center, which is calculated from the feature points of the calibration plate in the target image.

[0044] The second objective of this invention is to provide an adaptive curvature line structured light sensor optical surface calibration device, which aims to solve the problem that existing methods do not construct a collaborative curvature optimization model of the laser-camera-temperature sensor, thus failing to efficiently calculate error compensation.

[0045] To solve the above technical problems, an adaptive curvature optimized line structured light sensor optical surface calibration device is provided. The above adaptive curvature optimized line structured light sensor optical surface calibration method includes: a data acquisition module: acquiring data at different positions of the target through a camera, laser and temperature sensor, synchronously recording temperature sensor readings, and acquiring target image and laser stripe image;

[0046] Laser optical center positioning module: Performs speckle cross-correlation analysis on the laser stripe image, and obtains the coordinates O of the laser projection optical center by calculating the autocorrelation peak. v =[u v ,v v [and use it as the reference for the coordinate system of the laser;]

[0047] Hybrid curvature modeling module: Based on the target image, the camera intrinsic parameter matrix K is calculated using the Zhang calibration method. The Steger algorithm is then used to extract the center pixel coordinates [u,v] of the laser stripe image. T The three-dimensional point cloud coordinates P of the laser stripe were obtained by triangulation. c Then, the coordinates P of the three-dimensional point cloud were... c Calculate the curvature distribution κ using the central difference method. i ;

[0048] Closed-loop optimization unit module: Constructs a Gaussian-aberration hybrid curvature model based on the curvature distribution κ. i Calculate gradient magnitude Generating curvature sensitive weight w i The initial mounting pose of the laser is denoted as [R0, t0], and the optimal pose [R * ,t * ] and the curvature parameter

[0049] Thermal compensation and output module: according to the temperature reading, the curvature parameter is thermally compensated, and the compensated curvature distribution is adopted The B-spline surface reconstruction light surface equation is controlled, and finally the calibration result and error analysis are output.

[0050] In one of the embodiments, the device includes a processor and a memory, and the memory stores computer readable instructions, which, when executed by the processor, realize the module functions of any of the devices.

[0051] In one of the embodiments, the device includes a computer readable storage medium, which stores a computer program, and the computer program, when executed by a processor, realizes the module functions of any of the devices.

[0052] The implementation of the embodiments of the present application has the following beneficial effects:

[0053] 1. The light surface calibration method of the adaptive curvature line structured light sensor in the embodiment, first, the camera, the laser and the temperature sensor cooperate to collect multi-position target images, laser stripe images and real-time temperature; then the laser light center coordinates are determined by using the geometric projection method, and the camera intrinsic parameters are obtained based on Zhang's calibration; then the laser stripe is three-dimensionally reconstructed and the local curvature distribution is calculated; then the Gaussian-aberration hybrid curvature model is constructed, the curvature gradient module length is used to generate the sensitive weight, the closed-loop optimization unit based on the Levenberg-Marquardt algorithm is driven to iteratively solve the optimal pose and the curvature parameter; finally, according to the real-time temperature change, the thermal compensation mechanism is triggered, the curvature model is dynamically corrected, and the B-spline is used to reconstruct the high-precision light surface equation with adaptive control point density. Compared with the prior art, the present application combines the core technical advantages of physical feature driving, hybrid model compensation and closed-loop optimization unit, and innovatively introduces the real-time thermal compensation mechanism, which systematically solves the three technical bottlenecks of systematic error in high curvature area, temperature drift instability and edge distortion artifacts faced by traditional methods. Therefore, the calibration accuracy and stability of the line structured light sensor in the complex curved surface and temperature change environment are significantly improved, and the dependence on external precise displacement mechanism is avoided, realizing an efficient and robust calibration process.

[0054] 2、The light surface calibration device of the self-adaptive curvature line structured light sensor in the second embodiment of the present application, through the cooperative work of the data acquisition module, the laser light center positioning module, the mixed curvature modeling module, the closed loop optimization unit module and the thermal compensation and output module, a hardware-software cooperative architecture integrating image acquisition, feature extraction, three-dimensional reconstruction, mixed modeling, closed loop optimization and thermal compensation is constructed. Among them, the laser light center positioning module accurately determines the light center reference through speckle cross-correlation and geometric projection; the mixed curvature modeling module fuses camera calibration, laser stripe extraction and three-dimensional curvature calculation; the closed loop optimization unit module performs efficient iterative optimization based on the physical heuristic mixed model and the curvature sensitive weight; the thermal compensation and output module realizes real-time parameter correction and adaptive surface reconstruction driven by temperature. Compared with the existing device, the present scheme solves the problems of insufficient calibration accuracy, low efficiency and sensitivity to environmental temperature caused by the lack of systematic error compensation mechanism in the existing device through modular cooperative design, especially the multi-source information fusion of laser-camera-temperature sensor and the closed loop optimization unit based on physical model. The device can efficiently and automatically complete the active compensation of the inherent optical aberration of the laser and the running environment, and provides a high-precision and high-stability calibration solution for the line structured light sensor. BRIEF DESCRIPTION OF DRAWINGS

[0055] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiment or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0056] Figure 1 The flowchart of the self-adaptive curvature line structured light sensor light surface calibration method according to the first embodiment of the present application is shown in the figure.

[0057] Figure 2 The device installation schematic diagram of the self-adaptive curvature line structured light sensor light surface calibration method according to the second embodiment of the present application is shown in the figure.

[0058] Figure 3 The closed loop optimization mechanism principle diagram of the self-adaptive curvature line structured light sensor light surface calibration method according to the first embodiment of the present application is shown in the figure.

[0059] 1, line laser; 2, monocular camera; 3, image plane; 4, laser projection surface; 5, plane target; 6, temperature sensor. DETAILED DESCRIPTION

[0060] In the following, certain example embodiments are simply described. As those skilled in the art will recognize, the described embodiments can be modified in various different ways without departing from the spirit or scope of embodiments of the present application. Therefore, the drawings and descriptions are to be regarded as illustrative in nature and not restrictive. The disclosure below provides many different embodiments or examples for implementing different structures of embodiments of the present application. For the purpose of simplicity, the description in the following will be made with reference to certain examples. Of course, they are merely examples and are not intended to limit the present application. Furthermore, the present application can repeat reference numerals and / or reference letters in different examples and this repetition is for the purpose of simplicity and clarity and does not in itself dictate a relationship between the various implementations and / or configurations discussed. Embodiments of the present application will be described in detail in the following with reference to the attached drawings.

[0061] Embodiment one

[0062] Reference is made to Figure 1 , 3 Embodiment one of the present application provides a self-adaptive curvature line structured light vision sensor light surface calibration method. The embodiment includes system assembly and data acquisition steps. First, a line laser 1 and a monocular camera 2 are rigidly fixed to establish a mapping relationship between the camera coordinate system and the image plane 3. A plane target 5 is placed in the projection range of the line laser 1, and a temperature sensor 6 is placed beside the laser to monitor the real-time temperature. The plane target 5 is moved to three different spatial poses, and the monocular camera 2 is triggered synchronously to collect: 1) target image without laser for camera calibration; 2) laser stripe image with laser stripe; 3) temperature sensor 6 records the temperature reading T now at each position. Laser light center positioning step, speckle cross-correlation analysis is performed on the laser stripe image, and over-determined equations are solved by geometric projection method:

[0063]

[0064] The coordinates of the laser light center in the image plane 3 are output as O v =(u v ,v v ), which is the reference of the laser coordinate system.

[0065] Mixed curvature modeling step, first, Zhang calibration is performed based on the target image to obtain the camera intrinsic matrix K and the distortion coefficient dist. Then, Steger algorithm is applied to extract the laser stripe center pixel coordinates [u, v] T and correct the distortion. Then, the two-dimensional points are converted to three-dimensional point cloud Pc in the camera coordinate system by triangulation. Finally, the local curvature distribution is calculated by central difference method:

[0066]

[0067] Loop optimization unit step, construct a Gaussian-aberration hybrid model to describe laser intrinsic optical aberration:

[0068]

[0069] Then based on the curvature gradient modulus Generate curvature sensitive weight: Take the mechanical installation pose [R0, t0] as the initial value, and iteratively optimize the objective function by Levenberg-Marquardt algorithm:

[0070]

[0071] Output the optimal pose [R * ,t * ] and curvature parameters

[0072] Thermal compensation and surface reconstruction step, first calculate the temperature drift Apply thermal expansion coefficient α to update the curvature model, and generate non-uniform node vector according to the curvature distribution:

[0073]

[0074] Fit B-spline surface using weighted least squares:

[0075]

[0076] Finally output the optical surface equation S(u, v).

[0077] The embodiment one of the present application has the following advantages: firstly, the line laser 1, monocular camera 2 and temperature sensor 6 are used to cooperatively collect multi-position target images, laser stripe images and real-time temperatures; then, the laser light center coordinates are determined by using the geometric projection method, and the camera internal parameters are obtained based on Zhang calibration; then, the laser stripe is three-dimensionally reconstructed and the local curvature distribution is calculated; further, the Gaussian-aberration hybrid curvature model is constructed, the curvature gradient modulus is used to generate a sensitive weight, the closed-loop optimization unit based on the Levenberg-Marquardt algorithm is driven to iteratively solve the optimal pose and curvature parameters; finally, the heat compensation mechanism is triggered according to the real-time temperature change, the curvature model is dynamically corrected, and the B-spline with adaptive control point density is used to reconstruct the high-precision light surface equation. Compared with the prior art, the present application combines the core technical advantages of the physical feature driving, hybrid model compensation and closed-loop optimization unit, and innovatively introduces the real-time heat compensation mechanism, which systematically solves the three technical bottlenecks of systematic error in high-curvature area, temperature drift instability and edge distortion artifacts in the traditional method. Therefore, the calibration accuracy and stability of the line structured light sensor in the complex curved surface and temperature change environment are significantly improved, and the dependence on the external precise displacement mechanism is avoided, realizing an efficient and robust calibration process.

[0078] Embodiment two

[0079] The line structured light sensor light surface calibration device in the embodiment two is different from the subject matter protected by the line structured light sensor light surface calibration method in the embodiment one, and the specific differences are as follows:

[0080] Please refer to Figure 2 For a line structured light visual sensor light surface calibration method with adaptive curvature of an embodiment, the device comprises the following modules:

[0081] Data acquisition module: integrated line laser 1, monocular camera 2 and temperature sensor 6, supporting three-position synchronous trigger acquisition. The laser projector surface 4 forms a structured laser stripe on the plane target 5, and the temperature sensor 6 monitors the laser temperature rise in real time.

[0082] Laser light center positioning module: equipped with speckle cross-correlation analysis unit and geometric projection solver, the laser light center coordinates O are calculated by overdetermined equations v .

[0083] Hybrid curvature modeling module: 1) Zhang calibration unit: solving the camera internal parameters K and the target-camera transformation matrix 2) Real-time triangulator: converting the laser stripe center points extracted by the Steger algorithm into three-dimensional point cloud P c . 3) Curvature calculation engine: outputting the curvature distribution K based on the central difference method i .

[0084] Closed-loop optimization unit module: 1) Hybrid model constructor: initialize Gaussian-aberration model parameters {a, b, λ1, λ2, ∑}, where ∑ is dynamically updated according to the spatial distribution of point cloud. 2) Curvature-sensitive weight generator: calculate gradient magnitude and output exponential weight 3) Levenberg-Marquardt iterative solver: solve the incremental equation with Jacobian matrix J and weight matrix W, and dynamically update the pose and curvature parameters:

[0085] (J T WJ+μI)Δθ=J T Wr;

[0086] Thermal compensation output module: 1) Temperature drift compensator: correct the curvature model according to ΔT and material thermal expansion coefficient α. 2) Adaptive B-spline reconstructor: assign control point density, generate non-uniform node vector and fit the light surface S(u, v).

[0087] Through the collaborative work of the data acquisition module, the laser light center positioning module, the hybrid curvature modeling module, the closed-loop optimization unit module and the thermal compensation and output module, a hardware-software collaborative architecture integrating image acquisition, feature extraction, three-dimensional reconstruction, hybrid modeling, closed-loop optimization and thermal compensation is constructed. Among them, the laser light center positioning module accurately determines the light center reference through speckle cross-correlation and geometric projection; the hybrid curvature modeling module integrates camera calibration, laser stripe extraction and three-dimensional curvature calculation; the closed-loop optimization unit module performs efficient iterative optimization based on the hybrid model and curvature-sensitive weight; the thermal compensation and output module realizes real-time parameter correction and adaptive surface reconstruction driven by temperature. Compared with existing devices, the present scheme solves the problems of insufficient calibration accuracy, low efficiency and sensitivity to environmental temperature caused by the lack of systematic error compensation mechanism in existing devices through modular collaborative design, especially the multi-source information fusion of laser-camera-temperature sensor and the closed-loop optimization unit based on physical model. The device can efficiently and automatically complete active compensation for the inherent optical aberration of the laser and the operating environment, providing a high-precision and high-stability calibration solution for line structured light sensors.

[0088] In some embodiments, the line structured light sensor light surface calibration device and method further comprises: a processor connected with the system assembly and image acquisition module, the laser stripe feature extraction and camera calibration module, the three-dimensional coordinate solving module, the virtual camera pose estimation module, and the structure parameter calibration and light surface generation module respectively; a memory connected with the processor and storing a computer program executable on the processor; wherein when the processor executes the computer program, the processor controls the system assembly and image acquisition module, the laser stripe feature extraction and camera calibration module, the three-dimensional coordinate solving module, the virtual camera pose estimation module, and the structure parameter calibration and light surface generation module to work, so as to realize the line structured light sensor light surface calibration method of any one of the above.

[0089] The above-described embodiments are used to illustrate the present application and are not intended to limit the present application, so that the change of example values or the replacement of equivalent elements should still belong to the scope of the present application.

Claims

1. A method for calibrating a light surface of a line structured light sensor with adaptive curvature, characterized in that, The method comprises the steps of: S1, data acquisition, through the camera, laser and temperature sensor to collect data on different positions of the target, synchronously record real-time temperature T now , obtain target image and laser stripe image; S2, laser light heart positioning, speckle cross-correlation analysis is performed on the laser stripe image, and the light heart coordinate O of the laser is obtained through geometric projection method v = (u v ,v v ), and it is used as the reference of the laser coordinate system; S3. Hybrid curvature modeling: Based on the target image, the camera intrinsic parameter matrix K is calculated using the Zhang calibration method. The Steger algorithm is then used to extract the center pixel coordinates [u,v] of the laser stripe image. T The three-dimensional point cloud coordinates P of the laser stripe were obtained by triangulation. c Then, the coordinates P of the three-dimensional point cloud were... c Calculate the local curvature distribution κ using the central difference method. i ; S4, a closed-loop optimization unit, constructs a Gaussian-aberration hybrid curvature model based on the local curvature distribution κ i Computing the gradient norm Generating curvature-sensitive weights w i Let the initial pose of the laser installation be denoted as [R0, t0], and the optimal pose [R * * ] and curvature parameters ​ S5, thermal compensation and output, according to the temperature reading, thermal compensation is made to the curvature parameter, and the curvature distribution after thermal compensation is adopted The light surface equation is reconstructed by controlling the B-spline surface, and finally the calibration result and error analysis are output.

2. The method of claim 1, wherein, The laser optical center positioning comprises the steps of: Steger algorithm is applied to the laser stripe images collected at the different positions respectively to obtain laser stripe center lines, and the first and last end points of the laser stripe center lines are taken The first and last end points of the laser stripe center lines are connected to construct a laser stripe straight line L i , and a normal vector pixel Then the laser stripe straight line L i has an arbitrary point p on the straight line equation It can be known that the intersection point p of the laser stripe straight lines L i v is the projection point of the laser light source optical center in the image plane, and the over-determined equation is solved:​ The coordinates of the laser optical center in the image plane coordinate system can be finally obtained, denoted as O v = (u v ,v v ).

3. The method of claim 1, wherein, The mixed curvature modeling comprises the steps of: Solving the camera intrinsic parameter matrix K, distortion coefficient dist and the transformation matrix from the target image coordinate system to the camera coordinate system using Zhang's calibration method on the target image: in Let be a rotation matrix. As the translation matrix, the Steger algorithm is applied to the laser stripe image to obtain the pixel coordinates p of the center point of the laser stripe. stripe Using the camera intrinsic parameter matrix K and the distortion coefficient dist, the coordinates p of the center point of the laser stripe are determined. stripe Distortion correction is performed, and then the coordinates p of the center point of the laser stripe are determined by triangulation. stripe Converted to three-dimensional coordinates P in the camera coordinate system c Then, the three-dimensional coordinates P c According to the laser stripe straight line L i The order is sorted, and the local curvature k is calculated using the central difference method. i : Where P′ i ,P″ i The three-dimensional coordinates P are respectively c The first and second derivatives are used to construct a hybrid curvature model.

4. The method of claim 1, wherein, The closed-loop optimization unit comprises the steps of: Initializing the mixed curvature model parameters {R, t, a, b, λ1, λ2, ∑} and constructing a Gaussian-aberration mixed curvature model: where R,t is the pose under the curvature, a is the Gaussian distribution amplitude, b is the beam divergence factor, λ1, λ2 are the Zernike aberration coefficients, and Σ is the spatial covariance matrix, is the Zernike basis function, where x, y represent the normalized two-dimensional coordinates on the laser projection plane, and the curvature-sensitive weight is calculated based on the gradient length: Wherein β is an empirical value, and is set to 0.5, taking the initial pose [R0, t0] of mechanical installation as the starting point, a weighted projection error objective function is constructed: where π(·) is the camera projection function, C(u i ; κ model ) is the adaptive B-spline curve based on the current all curvatures, then the Levenberg-Marquardt algorithm is used to solve iteratively: (J T WJ+μI)Δθ=J T Wr; where J is the Jacobian matrix, J T is the transpose matrix of the Jacobian matrix, W is the weight matrix, μ is the adaptive damping factor, set to 0.001, I is the unit matrix, Δθ is the parameter increment vector, r is the residual vector, the hybrid curvature model parameters {R, t, a, b, λ1, λ2, ∑} are updated after each iteration, and κ model is recalculated, until convergence, and the optimal hybrid curvature model parameters {R * , t * , a * , b * , λ1 * , λ2 * , ∑ *} are finally output.

5. The method of claim 4, wherein, The Zernike aberration coefficient dynamic optimization comprises a distortion compensation constraint: wherein is a factory initial value, is a maximum curvature gradient, is a mean curvature gradient of the edge region of the surface.

6. The self-adapting curvature of wire structured light sensor light curve surface calibration method according to claim 1, wherein, The thermal compensation and output comprises the steps of: Loading the real-time temperature T now and the calibration reference temperature where T i is the synchronous reading of the temperature sensor at the i-th calibration position, n is the total number of calibration positions. This value represents the environmental temperature state of the calibration process and is used as a thermal compensation reference to calculate the temperature variation ΔT = T now - T calib , the thermal drift compensation model is applied to update the curvature parameters: where a is the thermal expansion coefficient of the material, obtained from the factory usage manual, according to the local curvature K i Compute control point distribution density: wherein is the average curvature, γ is the curvature sensitivity factor, set to 0.5, N base is the base number of control points, set to 20, and then based on the updated generating the non-uniform node vector: And using the weighted least squares method to fit the nodes into a B-spline surface: where N k,p are pth order B-spline basis functions, and the final output light surface equation S(u, v) completes the calibration task.

7. The method of claim 1, wherein, The triangulation adopts a binocular vision reconstruction model: Wherein f is the camera focal length, and d is the distance between the laser plane and the camera optical center, which is calculated by the calibration board feature points in the target image.

8. A calibration device for the optical surface of a line structured light sensor with adaptive curvature, characterized in that, Comprise: Data acquisition module: through the camera, laser and temperature sensor, the data of the target at different positions are collected, the temperature sensor readings are recorded synchronously, and the target image and the laser stripe image are obtained; A laser optical center positioning module: performing speckle cross-correlation analysis on the laser stripe image, obtaining the laser projection optical center coordinate O by calculating the autocorrelation peak value v = [u v , v v ], and taking it as the laser coordinate system reference; Hybrid curvature modeling module: Based on the target image, the camera intrinsic parameter matrix K is calculated using the Zhang calibration method. The Steger algorithm is then used to extract the center pixel coordinates [u, v] of the laser stripe image. T The three-dimensional point cloud coordinates P of the laser stripe were obtained by triangulation. c Then, the coordinates P of the three-dimensional point cloud were... c Calculate the curvature distribution κ using the central difference method. i ; Closed loop optimization unit module: construct a Gaussian-aberration hybrid curvature model, based on the curvature distribution K i Compute gradient magnitude Generate curvature-sensitive weight w i , the initial pose of the laser installation is recorded as [R0, t0], and the optimal pose [R * , * ] and the curvature parameter K are obtained by iterative solution through the Levenberg-Marquardt algorithm Thermal compensation and output module: according to the temperature reading, the curvature parameter is thermally compensated, and the compensated curvature distribution is adopted The light surface equation is reconstructed by controlling the B-spline surface, and finally the calibration result and error analysis are output.

Images