A self-calibration method for the consistency of a deflection system based on light ray constraints

Abstract

The present invention belongs to the technical field of deflection measurement, and specifically relates to a method for self-calibrating the consistency of a deflection measurement system based on light constraint. The present invention simplifies the system parameters to the direction and position of light rays in the screen coordinate system, measures a standard plane mirror three times, models the surface shape error of the standard mirror surface and the defocused virtual image of the screen, and according to the principle of linear propagation of light rays, uses the collinearity constraint between corresponding points of the virtual image to establish a light ray model of the system; then measures the surface shape of the standard part, and establishes a cost function that fuses the measurement error and the light ray collinearity error to correct the system parameters; finally, enhances the sensitivity of the system through a multi-frequency high-sensitivity encoding / decoding method. The present invention effectively shortens the error transmission chain, anneals the rigid calibration parameters of each link, and increases the flexibility of the system parameters.

Description

Technical Field

[0001] The present invention belongs to the technical field of deflection measurement, and particularly relates to a method for self-calibrating the consistency of a deflection measurement system. Background Art

[0002] With the continuous development of modern optical technology, various optical elements are widely used in various fields such as astronomy, medicine, military, aerospace, etc. High-precision three-dimensional measurement of the surface topography of components is an important technical means to detect their optical performance, stability and reliability, and has important research significance and broad application prospects.

[0003] Phase measurement deflectometry (PMD) is an efficient, flexible and robust complex surface measurement technology developed in recent years. This method has the advantages of large measurement dynamic range, high precision, high sensitivity, strong anti-interference ability, low cost, etc. PM is based on the basic principle of ray tracing in geometric optics and generally consists of two parts: a camera and a screen. It displays a series of fringe patterns on the screen, which are captured by the camera after being reflected by the component to be measured, so as to establish the corresponding relationship between the camera and the screen. The normal direction of the measured point is determined by ray tracing, and finally the surface shape of the component to be measured is reconstructed by gradient integration.

[0004] As can be seen from the above, the phase measurement deflectometry is a gradient measurement tool, and its measurement accuracy highly depends on the calibration accuracy of the system. The traditional system calibration is based on the camera calibration theory in machine vision. First, the camera is approximated as a pinhole model, and the internal and external parameters of the camera are calibrated using the Zhang Zhengyou algorithm [Zhang Z. A flexible new technique for camera calibration[J]. IEEE Transactions on pattern analysis and machine intelligence, 2000, 22(11): 1330-1334.] to determine the direction of the image-side light rays. Second, the external parameters of the screen relative to the camera need to be calibrated to determine the direction of the object-side light rays. Since the camera cannot directly image the screen, a plane mirror is usually used to deflect the optical path, and the pose of the screen is solved by establishing appropriate constraint relationships. Among them, Xiao et al. [Xiao Y L, Su X, Chen W. Flexible geometrical calibration for fringe-reflection 3D measurement[J]. Optics Letters, 2012, 37(4): 620-622.] used a markerless plane mirror with multiple poses (greater than or equal to 3 poses) to image the screen, and the poses of the plane mirror and the screen were solved simultaneously with the constraint that the pose of the screen remained unchanged. Xu et al. [Xu X, Zhang X, Niu Z, et al. Self-calibration of in situ monoscopic deflectometric measurement in precision optical manufacturing[J]. Optics Express, 2019, 27(5): 7523-7536.] used a turntable to constrain the poses between the plane mirrors, and the pose of the screen could be determined through two poses.Similarly, Li et al. [Li C, Zhang X, Tu D. Posed relationship calibration with parallel mirror reflection for stereo deflectometry [J]. Optical Engineering, 2018, 57(3): 034103-034103.] constrained the poses between two planar mirrors by machining an integrated parallel mirror, and Niu et al. [Niu Z, Zhang X, Ye J, et al. Flexible one-shot geometric calibration for off-axis deflectometry [J]. Applied Optics, 2020, 59(13): 3819-3824.] used fiducial points to constrain the poses of the planar mirrors, and the screen pose could be solved with only one pose.

[0005] There are a large number of approximations in the above calibration methods: First, in the calibration process of the reflected light, the camera is approximated as a pinhole model, assuming that all light rays pass through the optical center, and then the distortion is used to correct the light ray direction in the two-dimensional image domain. However, due to the existence of pupil aberration, the actual light rays do not all pass through the optical center uniformly, so the true light ray direction cannot be obtained through distortion correction, and the light rays must be adjusted in three-dimensional space. In addition, starting from the basic principle of deflectometry, the only thing it cares about is the direction and position of the light rays, so the traditional models based on internal and external parameters such as the optical center, focal length, and distortion have a large number of redundant parameters. Second, in the above methods for solving the screen pose using a standard planar mirror, it is assumed that the planar mirror is an ideal plane. However, there is no ideal mirror surface in reality, and the surface shape error of the standard mirror will cause defocusing of the virtual image of the screen obtained by the camera. Therefore, the virtual image of the screen is no longer a plane, and ignoring it will introduce calibration errors.

[0006] In addition, the system calibration usually uses the reprojection error in machine vision as the evaluation index, which is different from the evaluation index of measurement. As a result, the calibration accuracy of the system cannot truly reflect the measurement accuracy, leading to a serious disconnection between calibration and measurement. In order to unify the evaluation systems of calibration and measurement, some scholars measure the standard parts after the system calibration is completed and reverse-optimize the system parameters. E. et al. [Kewei E, Li D, Yang L, et al. Novel method for high accuracy figure measurement of optical flat [J]. Optics and Lasers in Engineering, 2017, 88: 162-166.] use the calibrated deflectometry system and interferometer to measure the plane standard part respectively. Taking the measurement data of the interferometer as the true value, the deviation between the measurement results of the two devices is used as the measurement error introduced by the system calibration error, and the system error is compensated accordingly. The model of using constants to compensate the system error under the approximation of the pinhole model in this method is too simplified and cannot effectively compensate the measurement errors of different poses or different workpieces. Han et al. [Han H, Wu S, Song Z. An accurate calibration means for the phase measuring deflectometry system [J]. Sensors, 2019, 19(24): 5377.] also use the calibrated system to measure the standard part, and use the RMS of the measurement results and the deviation between the characteristic size and the true value as constraints to reverse-optimize the system parameters. However, for the system calibration model determined by the pinhole hypothesis, the coupling problem between multiple system parameters needs further analysis. Huang et al. [Huang L, Xue J, Gao B, et al. Modal phase measuring deflectometry [J]. Optics express, 2016, 24(21): 24649-24664.] do not choose to believe the calibrated system parameters, but optimize the calibration parameters as variables together with the measured surface shape during measurement. However, this method only adjusts the object-side light direction by adjusting the screen pose, ignoring the image-side light error, making the calculation and error transfer chain longer and the error more difficult to separate.

[0007] In summary, the following problems in the field of deflectometry measurement remain to be solved: problems such as excessive approximation of the calibration model, redundant parameters, ideal approximation of the standard mirror surface, and disconnection between the calibration and measurement evaluation systems. Summary of the Invention

[0008] The object of the present invention is to propose a consistency self-calibration method for a customized, high-precision, calibration and measurement integrated light constraint deflection system to solve problems such as over-approximation, parameter redundancy, ideal approximation of the standard mirror surface, and disconnection between the calibration and measurement evaluation systems of the existing deflection system.

[0009] The consistency self-calibration method for the light constraint deflection system provided by the present invention starts from the basic principle of deflection technology. First, the system parameters are simplified to the direction and position of the light rays in the screen coordinate system. The standard plane mirror is measured three times, and the surface shape error of the standard mirror surface and the defocused virtual image of the screen are modeled. Then, according to the principle of linear propagation of light rays, a light ray model of the system is established using the collinearity constraint between corresponding points of the virtual images. Then, the surface shape of the standard part is measured, a cost function that fuses the measurement error and the light ray collinearity error is established, and the system parameters are corrected. Finally, the sensitivity of the system is enhanced by a multi-frequency high-sensitivity encoding / decoding method (Servin M, Padilla M, Garnica G. Super-sensitive two-wavelength fringe projection profilometry with 2-sensitivities temporal unwrapping [J]. Optics and Lasers in Engineering, 2018, 106: 68-74.); the specific steps are as follows:

[0010] S1. Build a deflection system composed of a camera and a screen, and place a standard plane mirror (hereinafter referred to as the standard plane or standard part) in the field of view of the camera so that the camera can capture the screen pattern reflected by it.

[0011] In the deflection measurement system, the camera and the screen are placed on the same side. Therefore, the camera cannot directly image the screen and needs to rely on the mirror to assist in obtaining the virtual image of the screen to establish the connection between the image-side light rays and the screen.

[0012] S2. Model the standard plane, including modeling it as surface shape, normal direction, and curvature.

[0013] As can be seen from S1, what the camera directly obtains is the virtual image of the screen rather than the real screen. However, there is actually no ideal mirror. Due to the existence of its surface shape error, the virtual image of the screen reflected by the standard plane obtained by the camera will show a defocus phenomenon. In order to accurately model the virtual image, it is necessary to model the standard plane mirror in advance.

[0014] S3. Project a series of horizontal and vertical stripes on the screen, and the camera captures the stripe pattern reflected by the standard plane to establish the correspondence between the camera and the screen pixels.

[0015] Specifically, multi-frequency four-step phase-shifted fringes in the horizontal and vertical directions are sequentially displayed on the screen, and the fringe patterns reflected by the standard plane are captured by the camera. The truncated phases of multiple fringe frequencies are obtained respectively using the four-step phase-shifting algorithm, and then the truncated phase with a higher frequency is obtained by adding the truncated phases. The truncated phase with a lower frequency is obtained by subtracting the truncated phases. Finally, guide to perform phase unwrapping to obtain a high-sensitivity phase map, and then establish the correspondence between the camera and the screen pixels with the phase value as a bridge.

[0016] S4. Using S3, measure the standard flat mirrors in 3 non-coplanar and non-parallel poses respectively, and obtain the corresponding screen pixels of the camera pixels in the 3 poses of the standard parts.

[0017] S5. Obtain the initial poses of the 3 standard flat mirrors in the screen coordinate system through pre-calibration.

[0018] S6. According to the system parameters pre-calibrated in S5 and the modeling results of the standard plane in S2, model the defocused virtual image of the screen.

[0019] Specifically, obtain the initial ray model according to S5, obtain the intersection points of the rays and the standard plane through ray tracing, obtain the azimuth information of the corresponding virtual image points according to the direction of the traced rays and the normal vector at this point, and then obtain the image distance according to the Gaussian imaging formula, and then determine the corresponding virtual image points. The normal vector and curvature information of the intersection points are obtained from S2. Repeat the above steps to complete the modeling of the screen virtual images in the 3 poses of the standard parts in turn.

[0020] S7. According to the principle of linear propagation of light, use the collinearity of the three virtual image points corresponding to the same camera pixel to establish the ray model of the system and record the collinearity error.

[0021] S8. Measure the standard parts in 3 poses according to the system ray model and record the deviation relative to the modeling results in S2.

[0022] S9. Integrate the collinearity error in the ray system model calibration and the surface shape deviation in the measurement to establish a cost function.

[0023] S10. Reverse-optimize the pose of the standard part relative to the screen with the cost function to obtain the optimized system ray direction. Repeat S5 - S10 until the cost function reaches the minimum.

[0024] Furthermore:

[0025] In step S2, for modeling the standard plane, an interferometer can be used to provide the raw data required for modeling the standard plane, and Zernike polynomials can be used to model the surface shape, normal direction, and curvature of the standard plane. Alternatively, raw data can be collected using a high-precision measuring instrument such as a profilometer, and other mathematical tools such as B-splines and radial basis functions can be used for modeling.

[0026] In step S5, for obtaining the initial poses of the three standard plane mirrors in the screen coordinate system through pre-calibration, the initial pose of the standard part relative to the screen can be obtained using the camera pinhole model and the PnP algorithm; alternatively, other pre-calibration methods can be selected according to the actual working scenario. For example, a coordinate measuring machine can be used to pre-calibrate the system; for secondary calibration of a system that has been working for a long time, the previous system parameters can be directly used as priors, and the initial values can be obtained through bundle adjustment.

[0027] In the present invention, the poses of the three standard planes can uniquely constrain the system model, or multiple poses can be used for least squares solution, and practitioners can decide according to the actual situation.

[0028] In the present invention, the deflection measurement system may include two of the cameras. Another camera also performs the steps S1-S10, and calibrates the light ray models of the two cameras in the screen coordinate system, thereby completing the calibration of the binocular deflection system.

[0029] The present invention also provides an electronic device, including a memory storing executable program code and a processor coupled to the memory; wherein, the processor calls the executable program code stored in the memory and executes the self-calibration method for constraining the consistency of the deflection system as described in claim 1 or 2.

[0030] The present invention provides a general, customized, and high-precision self-calibration method for a phase measurement deflection system. Starting from the basic principle of deflection measurement, aiming at the problems of over-approximation, parameter redundancy, and disconnection between calibration and measurement in the traditional deflection system model, it returns to the essence of deflection measurement and directly models the light rays. The uniqueness constraint of the model is derived, and through three measurements of the standard plane, the high-precision calibration of the system can be completed. In addition, the method takes into account the surface shape error of the standard mirror surface ignored in the traditional method and models the defocus imaging problem caused by it. And a system evaluation system integrating calibration and measurement is established, effectively solving the problem of disconnection between calibration and measurement of the deflection system. Finally, the sensitivity of the system is improved through a multi-frequency high-sensitivity fringe encoding / decoding scheme, further improving the calibration accuracy. Description of the Drawings

[0031] Figure 1 It is a schematic diagram of the principle of the self-calibration model of the present invention. Among them, (a) is a schematic structural diagram and (b) is a schematic optical path diagram.

[0032] Figure 2 This is the standard planar mirror used for system calibration in the first embodiment of the present invention.

[0033] Figure 3 This is to model the standard plane using the Zygo interferometer in the first embodiment of the present invention.

[0034] Figure 4 These are two pieces of fringe data obtained under the pose of a standard part in the first embodiment of the present invention.

[0035] Figure 5 This is the schematic diagram of the defocused virtual image modeling in the first embodiment of the present invention.

[0036] Figure 6 These are the measurement data of the plane in the first embodiment of the present invention. Among them, (a) the method of the present invention (b) the traditional method.

[0037] Figure 7 This is the schematic diagram of the self-calibration model of the binocular deflection system in the second embodiment of the present invention.

[0038] Figure 8 This is the structural schematic diagram of the third embodiment of the present invention.

[0039] Figure 9 This is the flowchart of the method of the present invention. Detailed implementation manners

[0040] The following further describes the detailed implementation manners of the present invention with reference to the accompanying drawings. It should be noted here that the description of these implementation manners is for helping to understand the present invention, but does not constitute a limitation to the present invention. In addition, the technical features involved in the various implementation manners of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0041] It should be noted that in the description of the present invention, the orientation or positional relationship indicated by terms such as "upper", "lower", "left", "right", "front", "rear", etc. is the description of the structure of the present invention based on the accompanying drawings. It is only for the convenience of describing the present invention simply, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation to the present invention.

[0042] For the "first" and "second" in this technical solution, they are only the appellation distinctions for the same or similar structures, or the corresponding structures with similar functions, and do not represent the arrangement of the importance of these structures, nor do they have the meaning of sorting, comparing sizes, or others.

[0043] In addition, unless otherwise clearly specified and defined, the terms "installation" and "connection" should be understood in a broad sense. For example, a connection can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, and it can be the communication inside two structures. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to the general idea of the present invention and in connection with the specific circumstances of this solution's context.

[0044] Embodiment 1

[0045] A method for self-calibrating the consistency of a deflection system based on light ray constraint, where the system parameter is the light ray corresponding to each camera pixel. The system calibration method is as Figure 1 shown. By measuring the standard mirror three times, a light ray model is established with virtual images, and the light ray direction is reversely optimized with collinearity error and measurement error.

[0046] The method provided in this embodiment includes the following steps:

[0047] S1. Build a deflection system composed of a camera and a screen, and place the standard plane mirror in the camera's field of view so that the camera can capture the screen pattern reflected by the standard part, as Figure 1 shown.

[0048] S2. Model the standard plane, including: surface shape, normal direction, curvature, etc.

[0049] To model the standard plane, use the GPI XP / D interferometer of Zygo Corporation in the United States to measure the standard mirror as Figure 2 shown to obtain the original data, as Figure 3 shown. Then, model the surface shape, normal direction, and curvature of the standard part through Zernike polynomials.

[0050] S3. By projecting a series of horizontal and vertical stripes on the screen and having the camera capture the stripe pattern reflected by the standard plane, establish the correspondence between the camera and the screen pixels.

[0051] Specifically, the screen projects three-frequency four-step phase-shift stripe patterns with frequencies of 56, 63, and 64 in sequence, and measures the standard plane mirror in three non-coplanar and non-parallel poses to obtain the corresponding stripe patterns, as Figure 4 shown.

[0052] S4. Using S3, measure the standard plane mirror in three non-coplanar and non-parallel poses respectively, and obtain the screen pixels corresponding to the camera pixels in the three poses of the standard part.

[0053] S5. Obtain the initial poses of the three standard plane mirrors in the screen coordinate system through pre-calibration.

[0054] In this example, the initial poses between the screen and the standard plane mirror are estimated through Zhang Zhengyou calibration and the PnP algorithm. Specifically, the internal parameter matrix A of the camera is calibrated through the Zhang's algorithm. Through the PnP algorithm, the pose Ext1 of the standard part relative to the camera coordinate system is obtained using the marked points on the standard plane. Further, by taking pictures of the virtual images of the screen marked points by the camera, the pose Ext2 of the screen virtual image relative to the camera coordinate system is obtained. Finally, the initial pose between the screen and the standard part is obtained through plane mirror imaging and coordinate transformation.

[0055] S6. Model the defocused virtual image of the screen according to the system parameters pre-calibrated in S5 and the modeling results of the standard plane in S2, as Figure 5 shown.

[0056] Specifically, according to the initial ray model obtained in S5, the intersection points of the rays and the standard plane are obtained through ray tracing. According to the direction of the traced rays and the normal vector at this point, the azimuth information of the corresponding virtual image points is obtained. Furthermore, the image distance is obtained according to the Gaussian imaging formula, and then the corresponding virtual image points are determined. The normal vector and curvature information of the intersection points are obtained in S2. Repeat the above steps to complete the modeling of the screen virtual image under the poses of the three standard parts in sequence.

[0057] S7. According to the principle of linear propagation of light rays, use the collinearity of the three virtual image points corresponding to the same camera pixel to establish the ray model of the system and record the collinearity error.

[0058] S8. Measure the standard parts in the three poses according to the system ray model and record the deviation relative to the modeling results in S2.

[0059] S9. Integrate the collinearity error in the ray system model calibration and the surface shape deviation in the measurement to establish a cost function.

[0060] S10. Reverse-optimize the pose of the standard part relative to the screen with the cost function, and then obtain the optimized system ray direction. Repeat S5 - S10 until the cost function reaches the minimum.

[0061] Specifically, a group of experiments are carried out on the above method steps in this embodiment. A measurement system is constructed using a camera with a focal length of 75 mm, a resolution of 2048×2048, and a pixel size of 0.0055 mm and a screen with a resolution of 1536×2048 and a pixel size of 0.0784 mm, as shown in the figure. To provide an ideal initial value for the system, the actual poses of the three standard planes and the screen are given through Zhang Zhengyou calibration. With the screen coordinate system set as the global coordinate system, the poses of the standard parts in the global coordinate system are represented by T m2s_i , R m2s _i It is expressed as follows:

[0062]

[0063] where R is the rotation matrix, T is the translation vector, the subscript m represents the coordinate system of the standard part, s represents the screen coordinate system, and i = 1, 2, 3 represents the number of poses of the standard part. For example: R m2s _1 represents the rotation matrix from the standard plane coordinate system in the first pose to the screen coordinate system.

[0064] Model the surface shape, normal direction, and curvature information of the standard mirror surface using the measurement data of the interferometer. Specifically, first, fit the surface shape using Zernike polynomials, obtain the gradient of the surface shape by differentiating the polynomial, and then obtain the normal direction information. Secondly, the average curvature of the measured mirror surface is obtained through the following formula:

[0065]

[0066]

[0067]

[0068] where K is the average curvature, Z x and Z y are the first-order partial derivatives of the surface shape function with respect to the x and y directions, and Z xx , Z xy , Z yy are the second-order partial derivatives of the surface shape function with respect to the x and y directions. Due to the 2-ill-conditioned problem of the second derivative of the surface shape and the mode coupling problem between the terms of the Zernike polynomial, it is impossible to obtain accurate second derivatives by continuously differentiating the polynomial. Therefore, we obtain the second derivative of the surface shape through numerical differentiation of the gradient data, and then calculate the curvature K of the surface shape.

[0069] In the modeling of the virtual image, the image distance is obtained from the Gaussian imaging formula, where u is the object distance, v is the image distance, and R is the radius of curvature:

[0070]

[0071] After the calibration steps of S1 - S10, a commercial flat mirror with a flatness error less than λ / 10 was measured, and the measurement results are as Figure 6 (a) shown. To evaluate the accuracy, the traditional method was used to calibrate the system, and then the same method was used to process the same measurement data, as Figure 6 (b) shown. The accuracy comparison is shown in the following table. It can be seen from the table that calibrating the system using the proposed method can approximately double the measurement accuracy.

[0072] Table I Precision Evaluation Table

[0073]

[0074] Example 2

[0075] The present invention is applicable to any structure of the deflection system and can be very easily extended to a binocular system. This example provides a self-calibration method for a binocular-based deflection system, as Figure 7 shown. Since the proposed method simplifies the system model into a set of light rays in the screen coordinate system, there is no essential difference between the calibration of the binocular system and the monocular system, except for adding a set of light rays. Specifically, steps S1-S10 are respectively executed on the two cameras to achieve the calibration of the image-side light rays of the two cameras. Optionally, an item can be added to the cost function in S9: the deviation of binocular measurement of the standard part, which can better constrain the light rays and improve the accuracy of the system. The other specific processes are the same as those in Example 1 and will not be elaborated here.

[0076] It is easy to understand that by making minor modifications to the method steps disclosed in Example 1, various variants can also be constructed for various deflection measurement structures, such as a deflection measurement system composed of a single camera and two screens, etc.

[0077] Example 3

[0078] An electronic device, as Figure 8 shown, includes a memory storing executable program code and a processor coupled to the memory; wherein, the processor calls the executable program code stored in the memory and executes the method steps disclosed in the above examples. Among them, the processor can be any conventional processor, such as a CPU, FPGA, etc.; the storage module can be any conventional storage device, such as a memory card, hard disk, cloud server, etc.

[0079] In addition, as Figure 8 shown, when the device is used as a commercial measurement device, it further includes a data acquisition module, a display module, and a power supply module. Among them, the data acquisition module transmits the measurement data to the memory through a network cable or other means, and the memory inputs the received measurement data and the executable program code stored in advance to the processor. After being processed by the processor, the measurement results are then output to the storage module and displayed on the display module. The power supply module is used to supply power to other modules

[0080] This application is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices produce a means for implementing the functions specified in one or more flows and / or blocks. Figure 1 in one or more flows and / or blocks Figure 1 or a plurality of blocks.

[0081] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory produce a manufactured article including an instruction means that implements the functions specified in one or more flows and / or blocks. Figure 1 in one or more flows and / or blocks Figure 1 or a plurality of blocks.

[0082] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one or more flows and / or blocks. Figure 1 in one or more flows and / or blocks Figure 1 or a plurality of blocks.

[0083] The embodiments of the present invention have been described in detail above in conjunction with the accompanying drawings, but the present invention is not limited to the described embodiments. For those skilled in the art, without departing from the principles and spirit of the present invention, various changes, modifications, substitutions, and variations to these embodiments still fall within the protection scope of the present invention.

Claims

1. A method for self-calibrating the consistency of a light-based constrained deflection system, characterized in that The specific steps are as follows: S1. Build a deflection system composed of a camera and a screen, place a standard plane mirror within the camera's field of view so that the camera can capture the screen pattern reflected by it; Since the camera and the screen in the deflection measurement system are on the same side, the camera cannot directly image the screen. It is necessary to rely on the mirror to assist in obtaining the virtual image of the screen to establish the connection between the image-side light rays and the screen; S2. Model the standard plane, including modeling its surface shape, normal direction, and curvature; S3. Project a series of horizontal and vertical stripes on the screen, and the camera captures the stripe patterns reflected by the standard plane to establish the correspondence between the camera and the screen pixels; Specifically, multi-frequency four-step phase-shifted fringes in both horizontal and vertical directions are sequentially displayed on the screen, and the fringe pattern reflected by the standard plane is captured by a camera. The truncated phases of multiple fringe frequencies are obtained using the four-step phase-shift algorithm, and then the truncated phases of higher frequencies are obtained by adding the truncated phases. φ 1. The truncated phase of a lower frequency is obtained by subtracting the truncated phases. φ 2. Finally, φ 2 is used as a guide φ 1 to perform phase unwrapping to obtain a high-sensitivity phase map, and then a correspondence relationship between the camera and the screen pixels is established with the phase value as a bridge. S4. Use S3 to measure the standard plane mirrors in 3 non-coplanar and non-parallel poses respectively, and obtain the corresponding screen pixels of the camera pixels in the 3 poses of the standard components; S5. Obtain the initial poses of the 3 standard plane mirrors in the screen coordinate system through pre-calibration; S6. According to the system parameters pre-calibrated in S5 and the modeling results of the standard plane in S2, model the defocused virtual image of the screen; Specifically, obtain the initial light ray model according to S5, obtain the intersection points of the light rays and the standard plane through ray tracing, obtain the azimuth information of the corresponding virtual image points according to the direction of the traced rays and the normal direction at this point, and then obtain the image distance according to the Gaussian imaging formula to determine the corresponding virtual image points; the normal direction and curvature information of the intersection points are obtained from S2. Repeat the above steps to complete the modeling of the screen virtual images in the 3 poses of the standard components in sequence; S7. According to the principle of linear propagation of light rays, use the collinearity of the three virtual image points corresponding to the same camera pixel to establish the light ray model of the system and record the collinearity error; S8. Measure the standard components in the 3 poses according to the system light ray model and record the deviation relative to the modeling results in S2; S9. Integrate the collinearity error in the calibration of the light ray system model and the surface shape deviation in the measurement to establish a cost function; S10. Reverse-optimize the poses of the standard components relative to the screen with the cost function to obtain the optimized light ray direction of the system; repeat S5 - S10 until the cost function reaches the minimum.

2. The method according to claim 1, characterized in that, In S2, use an interferometer to provide the required original data for modeling the standard plane, and use Zernike polynomials to model the surface shape, normal direction, and curvature of the standard plane; or collect the original data through a profiler and use mathematical tools such as B-splines and radial basis functions for modeling.

3. The method according to claim 1, wherein In S5, use the camera pinhole model and the PnP algorithm to obtain the initial poses of the standard components relative to the screen, or pre-calibrate the system using a coordinate measuring machine according to the actual working scenario; If the system that has been working for a long time is calibrated for the second time, the previous system parameters can be directly used as a priori, and the initial value can be obtained through bundle adjustment.

4. The self-calibration method for the consistency of the constrained deflection system according to claim 1, wherein: The deflection system includes two cameras. The other camera also performs the steps of S1 - S10 to calibrate the light ray models of the two cameras in the screen coordinate system, and then the calibration of the binocular deflection system can be completed.

5. An electronic device for performing the method according to claim 1 or 2, characterized in that, It includes a memory storing executable program code and a processor coupled to the memory; wherein, the processor calls the executable program code stored in the memory to execute a method for self-calibrating the consistency of a constraint deflection system.

Images