Method and system for deformation measurement of large structure based on visual measurement

Abstract

The present application relates to the technical field of deformation visual measurement of large structural parts, and particularly relates to a large structural part deformation measurement method and system based on visual measurement. The method comprises the following steps: setting up a measurement system, completing camera calibration and coordinate system binding, collecting target images in real time, suppressing environmental light interference through active illumination and narrow bandpass lens, extracting feature points through image preprocessing, contour extraction and ellipse fitting, completing 2D-3D matching based on asymmetric topological relationship, obtaining a six-degree-of-freedom pose matrix through pose solution of the feature points, establishing a reference pose and operating with the real-time pose, and outputting the six-degree-of-freedom deformation variable of the structural part. The system comprises a measurement reference platform, a measurement camera, an active illumination unit, a cooperative target, an image processing and pose solution unit, and a deformation calculation unit. The advantages are as follows: strong anti-interference, high real-time performance, lightweight and adaptive on-orbit deployment, stable high-precision six-degree-of-freedom deformation monitoring at all times, and meeting the health monitoring requirements of aerospace large structural parts.

Description

Technical Field

[0001] This invention relates to the field of visual measurement technology for deformation of large structural components, and in particular to a method and system for measuring the deformation of large structural components based on visual measurement. Background Technology

[0002] Large structural components are core parts of spacecraft, such as large trusses, lander engine supports, and deployable space antennas. During their service in orbit, they are prone to localized deformation due to extreme temperature differences and vibrations. If the deformation exceeds a safety threshold, it may cause the on-orbit mission to deviate from its expected trajectory, or even trigger a major space safety accident. Therefore, real-time, high-precision deformation measurement of large structural components in orbit is crucial to ensuring the effective conduct of space missions.

[0003] Existing on-orbit structural deformation measurement technologies are mainly divided into two categories: contact and non-contact. Contact measurement suffers from the limitation of complex deployment, so the mainstream solution is mostly non-contact measurement. Among non-contact measurements, although equipment such as laser trackers and laser scanners have high measurement accuracy, they are bulky, heavy, consume a lot of power, and are expensive, and cannot simultaneously acquire six-degree-of-freedom deformation parameters. Monocular vision-based measurement methods, with their advantages of being non-contact, low-power, flexible in deployment, and capable of simultaneously calculating translational and rotational 6D deformation components, have become an important technical direction for on-orbit engineering parameter measurement. Among them, vision measurement schemes based on the PnP principle have extremely high engineering application value due to their simple system architecture, low cost, and high technological maturity.

[0004] However, existing on-orbit visual measurement schemes based on the PnP principle still have technical bottlenecks: they are susceptible to interference from extreme on-orbit lighting, making it difficult to achieve stable all-day measurement; conventional target features are singular, lacking redundancy optimization mechanisms, and cannot meet the high reliability and high precision requirements of on-orbit structural deformation monitoring. Therefore, developing a visual measurement method with strong resistance to environmental interference, high measurement accuracy, and the ability to achieve stable 6D deformation monitoring all day is of great significance to aerospace engineering. Summary of the Invention

[0005] To address the aforementioned problems, this invention provides a method and system for measuring the deformation of large structural components based on visual measurement.

[0006] The primary objective of this invention is to provide a method for measuring the deformation of large structural components based on visual measurement, comprising the following steps: S1. Construct a large-scale structural component deformation measurement system based on vision measurement, including a measurement reference platform, a measurement camera, an active illumination unit, a cooperative target, an image processing and pose calculation unit, and a deformation calculation unit; rigidly fix the measurement camera to the measurement reference platform, and rigidly fix the cooperative target on the surface of the large-scale structural component; calibrate the measurement camera, import the three-dimensional feature coordinates of the target, and complete the coordinate system binding; S2. The measuring camera acquires images containing the cooperative target in real time, simultaneously activates the active illumination unit, and suppresses ambient light interference through the measuring camera's optical lens to achieve stable imaging; S3. Perform adaptive threshold binarization, contour tracking and ellipse fitting on the acquired image to extract effective feature points, and sort them based on asymmetric topological relationships to complete 2D-3D feature point matching; S4. Perform pose calculation on the feature point matching results and output a six-degree-of-freedom pose matrix; S5. Establish a reference pose under the stable state of the structural component, perform matrix operations on the real-time pose and the reference pose, and calculate and output the six-degree-of-freedom deformation of the structural component.

[0007] Preferably, the cooperative target adopts an asymmetric topological multi-feature redundancy design, containing no less than 3 feature points; the feature points are circular features and / or dot features.

[0008] Preferably, in step S1, the calibration of the measuring camera specifically includes intrinsic parameters, distortion coefficients, and extrinsic parameters; the distortion coefficients include radial distortion and tangential distortion; the extrinsic parameters are the rotation matrix and translation vector of the measuring camera relative to the measuring reference platform.

[0009] Preferably, the active illumination unit is coaxially arranged with the measuring camera, and the illumination direction is aligned with the cooperative target area; in step S2, the active illumination unit adopts narrowband laser active illumination, and the optical lens of the measuring camera is a narrowband bandpass coated lens. The wavelength of the narrowband laser active illumination is consistent with the wavelength of the narrowband bandpass coated lens, so that the active illumination energy is much greater than the energy of sunlight in the corresponding wavelength band.

[0010] Preferably, step S3 specifically includes: using the Otsu method adaptive threshold segmentation algorithm to binarize the single-channel image, automatically adapting to illumination fluctuations, and segmenting the target feature foreground and background regions; using the Suzuki85 contour tracking algorithm to extract closed contours in the image and complete preliminary screening; performing ellipse fitting on the screened effective contours to remove false features; completing feature sorting based on topological relationships, establishing a one-to-one match between 2D image points and 3D target points, and establishing the correspondence between 2D and 3D feature points.

[0011] Preferably, the pose calculation in step S4 specifically includes: using the P3P algorithm to perform coarse pose calculation, and then using the LM algorithm to perform nonlinear optimization with the goal of minimizing the reprojection error, and outputting a six-degree-of-freedom pose matrix.

[0012] Preferably, in step S5, the reference pose is obtained by continuously collecting no less than 10 frames of pose data, removing outliers using the 3σ criterion, and then weighting the average.

[0013] Preferably, the six-degree-of-freedom deformation is calculated through the pose transformation matrix, and the matrix is ​​analyzed into translation vectors and Euler angles for output.

[0014] The second objective of this invention is to provide a large structural component deformation measurement system based on vision measurement, for performing the aforementioned large structural component deformation measurement method based on vision measurement, including a measurement reference platform, a measurement camera, an active illumination unit, a cooperative target, an image processing and pose calculation unit, and a deformation calculation unit; The measuring camera is fixed on the measuring reference platform; the active illumination unit is coaxially arranged with the measuring camera, and the illumination direction is aligned with the cooperative target area; the cooperative target is rigidly fixed to the surface of the large structural component. The image processing and pose calculation unit receives image data output from the measurement camera and sequentially performs image preprocessing, feature detection and matching, coarse pose calculation and fine pose calculation. The deformation calculation unit is used to establish the reference pose and calculate the six-degree-of-freedom deformation based on the real-time pose and the reference pose.

[0015] Preferably, the optical lens of the measuring camera is coated with a narrow bandpass coating, the band of which is matched with the active illumination band; The cooperative target has an asymmetric topological structure, and the three-dimensional coordinates of the feature points are calibrated using a coordinate measuring machine. The deformation calculation unit includes a reference pose establishment module and a real-time deformation calculation module; the reference pose establishment module uses the 3σ criterion to remove abnormal data, and the real-time deformation calculation module outputs six-degree-of-freedom deformation through pose matrix transformation.

[0016] Compared with the prior art, the present invention can achieve the following beneficial effects: (1) Adapt to the on-orbit space environment and achieve stable measurement all day long. Through narrowband laser active illumination and spectral matching imaging mechanism, it effectively suppresses interference from strong light, backlight, shadow and ambient stray light in orbit, overcomes the defects of traditional visual measurement that are easily limited by lighting conditions, and can maintain clear imaging and stable features under complex on-orbit lighting conditions, and truly achieve continuous and reliable measurement all day long and in all weather.

[0017] (2) It has strong real-time processing capabilities and is suitable for continuous dynamic monitoring in orbit. It adopts a lightweight and fast image processing and coarse-fine joint pose calculation strategy. The process is simple and the calculation efficiency is high. It can realize high frame rate pose and deformation calculation on a low-power hardware platform. It can track the dynamic deformation, instantaneous vibration and long-term slow change characteristics of structural components in real time, and meet the needs of long-term real-time monitoring in orbit.

[0018] (3) The system is lightweight and low-power, and is suitable for on-orbit deployment. It adopts monocular vision and rigid fixed architecture, with no moving parts, small size, light weight, low power consumption, strong resistance to vibration and temperature change, and can adapt to the harsh space environment of spacecraft. It is easy to integrate and install and does not occupy too much payload resources.

[0019] (4) Achieve full-dimensional deformation calculation with six degrees of freedom, and fully characterize the actual deformation state of the structure. Simultaneously output complete deformation information of three-dimensional translation and three-dimensional rotation, comprehensively reflect the spatial pose changes and local deformation characteristics of structural components, make up for the shortcomings of traditional measurement methods with single dimension and incomplete information, and provide comprehensive and accurate data support for fault early warning and safety control. Attached Figure Description

[0020] Figure 1 This is a flowchart of a method for measuring the deformation of large structural components based on visual measurement.

[0021] Figure 2 This is a schematic diagram of the measurement principle of a large structural component deformation measurement system based on vision measurement according to an embodiment of the present invention.

[0022] Figure 3 This is a schematic diagram of the cooperative target feature layout provided in an embodiment of the present invention.

[0023] Figure label: 1. Cooperative targets; 101. Circular features; 102. Column; 103. Target plane; 104. Dot features; 2. Measuring camera; 3. Measurement reference platform; 4. Structural components after deformation; 5. Structural components. Detailed Implementation

[0024] In the following description, embodiments of the invention will be described with reference to the accompanying drawings. In the description below, the same modules are denoted by the same reference numerals. Where the same reference numerals are used, their names and functions are also the same. Therefore, their detailed description will not be repeated.

[0025] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and do not constitute a limitation thereof.

[0026] This invention provides a method for measuring the deformation of large structural components based on vision measurement, the flowchart of which is shown below. Figure 1 This includes the following steps: S1. System setup and calibration; specifically including: S11. Construct a large-scale structural component deformation measurement system based on vision measurement, including a measurement reference platform, a measurement camera, an active illumination unit, a cooperative target, an image processing and pose calculation unit, and a deformation calculation unit; rigidly fix the cooperative target on the surface of the large-scale structural component, and fix the measurement camera on the measurement reference platform; the active illumination unit and the measurement camera are coaxially arranged, and the illumination direction is aligned with the cooperative target area; after the system is constructed, proceed to the system calibration stage; S12. System Calibration: Multiple images are acquired within the field of view of the measuring camera using a calibration board. The camera's intrinsic parameter matrix, radial distortion, and tangential distortion are calculated using Zhang's calibration method. The extrinsic parameters of the measuring camera relative to the measuring reference platform, including the rotation matrix and translation vector, are obtained through high-precision calibration. The three-dimensional world coordinates of the target feature points are imported to complete the binding of the target and the structural reference coordinate system.

[0027] The structure, function, and working process of each part of the measurement system are as follows: The measurement reference platform is a rigid base, providing a stable spatial reference for the entire measurement system; The measuring camera is rigidly mounted on the measuring reference platform, with no relative displacement after installation; the relative pose relationship between the measuring camera and the measuring reference platform (6-DOF relative pose) Where c represents the measuring camera and a represents the measuring reference platform; that is, the translation and rotation parameters of the measuring camera relative to the measuring reference platform are obtained through high-precision pre-calibration; the internal parameters and distortion coefficients of the measuring camera are pre-calibrated; the optical lens of the measuring camera is coated with a narrow bandpass coating, and its passband is matched with the emission band of the active illumination unit to suppress ambient light interference.

[0028] The active illumination unit is coaxially arranged with the measuring camera, and the illumination direction is towards the cooperative target. The illumination power is adaptively adjusted according to the image brightness to improve the imaging contrast of the target features and achieve all-day anti-strong light interference imaging. The active illumination unit employs a spectral matching scheme combining narrowband laser active illumination and narrowband coating on the optical lens. The core idea is to allow only energy within a narrow spectral range to enter the imaging target surface, ensuring that the active illumination energy is greater than the sunlight energy within this bandwidth. Specifically, the principle is as follows: assuming the energy in the direct sunlight region is... The energy in the shadow region produced by the sun is If there is no active lighting, The contrast between the two The energy is very large, which is detrimental to target recognition; if active laser illumination is used, the energy generated by active illumination is... Then the ratio of energy between the sun's direct rays and the shadow area is... Within a narrow spectral range, active illumination has a higher energy level than sunlight, thus significantly reducing the energy difference between light and shadow areas. Much larger Then the energy in the light and shadow areas is approximately equal. This means that solar interference can be completely ignored, enabling all-day operation of visual measurements.

[0029] The cooperative target is rigidly fixed to the surface of the large structural component being tested. Its six-degree-of-freedom pose relative to the structural component is obtained and solidified through precise calibration. The cooperative target has a multi-feature redundancy design and adopts an asymmetric topology structure, with N (N>3) circular features and / or dot features distributed on it. The target plane is coated with a matte black coating, and the feature areas are coated with a high-reflectivity matte white coating. The three-dimensional coordinates of all feature points on the cooperative target are pre-calibrated using a high-precision coordinate measuring machine.

[0030] The image processing and pose calculation unit receives image data output from the measurement camera and sequentially performs image preprocessing, feature detection and matching, and coarse-fine two-step pose calculation. (1) Image preprocessing: Adaptive threshold binarization, contour tracking, ellipse fitting, and removal of false features are performed on the image; Specifically, the Otsu method adaptive threshold segmentation algorithm is used to binarize the single-channel image, automatically adapting to illumination fluctuations and segmenting the target feature foreground and background regions; the Suzuki85 contour tracking algorithm is used to extract closed contours in the image, and preliminary screening is performed based on preset feature size range and contour moment threshold to filter out invalid interference contours caused by stray light and laser speckle; ellipse fitting is performed on the filtered effective contours to calculate the contour center coordinates and roundness parameters, and combined with the geometric design constraints of the target features, pseudo-features that do not meet the roundness and size requirements are filtered out to obtain the set of effective feature points; (2) Feature matching: Feature points are selected based on topological relationships, and a one-to-one correspondence between two-dimensional image feature points and three-dimensional target coordinates is established to complete 2D-3D feature point pair matching; (3) Pose calculation: The P3P algorithm is used to solve the initial pose value, and then the Levenberg-Marquardt (LM) algorithm is used to perform nonlinear optimization on the reprojection error to output a high-precision six-degree-of-freedom pose matrix; Specifically, the P3P algorithm is used to solve the initial coarse value of the pose, quickly determine the initial pose value and resolve the ambiguity of multiple pose solutions, and provide a reliable initial value for subsequent nonlinear optimization. A nonlinear optimization function is constructed with the goal of minimizing the sum of squared reprojection errors of all feature points. The Levenberg-Marquardt algorithm is used for iterative optimization, and after convergence, the high-precision six-degree-of-freedom pose matrix corresponding to the frame image is obtained.

[0031] The deformation calculation unit is data-connected to the image processing and pose calculation unit, and is used to realize the establishment of the reference pose and the calculation of six degrees of freedom deformation, including the reference pose establishment module and the real-time deformation calculation module. (1) Reference pose establishment module: used to establish a high-precision initial reference pose when the large structural component under test is in a stable and deformation-free state; the specific workflow is as follows: continuously acquire no less than 10 frames of cooperative target images and solve the target pose frame by frame; use the 3σ criterion to remove outliers from the multi-frame pose data, that is, first calculate the mean and standard deviation of all pose data, take the mean ± 3σ as the effective data interval, and remove the abnormal pose results that exceed the interval; perform weighted average calculation on the retained effective pose data to obtain the high-precision initial reference pose matrix. As a benchmark for subsequent deformation measurements, it effectively reduces the benchmark establishment error caused by on-orbit micro-vibration and random noise.

[0032] (2) Real-time deformation calculation module: used to perform matrix operations on the target pose obtained in real time and the reference pose, and output the true six-degree-of-freedom deformation of the structural component; specifically, the pose matrix obtained in real time is used to calculate the target pose. With reference pose matrix Substitute into the deformation calculation formula: ; The pose transformation matrix of the structural component before and after deformation is obtained; then the pose transformation matrix is ​​analytically converted into three-dimensional translation vectors and three-dimensional Euler angles, and finally the complete six-degree-of-freedom deformation of the structural component is output.

[0033] S2. Image Acquisition and Laser Anti-interference Imaging: The measuring camera acquires images containing the cooperative target in real time, and simultaneously turns on the narrowband laser active illumination. The narrowband filter suppresses the interference of sunlight and ambient stray light, achieving stable imaging all day long. The image quality is judged, and overexposed, underexposed and blurred frames are discarded.

[0034] S3. Fast image processing and feature extraction; specifically including: S31. Perform adaptive threshold binarization on the image to segment the target feature region; Specifically, the Otsu method adaptive threshold segmentation algorithm is used to binarize the single-channel image, automatically adapt to illumination fluctuations, and segment the target feature foreground and background regions. S32. Extract effective dot features or ring features through contour tracking, ellipse fitting, and geometric constraint judgment; Specifically, the Suzuki85 contour tracking algorithm is used to extract closed contours from the image, and preliminary screening is performed based on the preset feature size range and contour moment threshold to filter out invalid interference contours caused by stray light and laser speckle. Ellipse fitting is performed on the selected effective contours to calculate the contour center coordinates and roundness parameters. Combined with the geometric design constraints of the target features, pseudo-features that do not meet the roundness and size requirements are filtered out to obtain a set of effective feature points. S33. Based on topological relationships, complete feature sorting, establish one-to-one matching between 2D image points and 3D target points, and establish the correspondence between 2D and 3D feature points; Specifically, the topological relationship is an asymmetric topological relationship.

[0035] S4. Pose calculation: The P3P algorithm is used to solve the initial pose value, and then the Levenberg-Marquardt algorithm is used to perform nonlinear optimization on the reprojection error to output a high-precision six-degree-of-freedom pose matrix. Specifically, the P3P algorithm is used to solve the initial coarse value of the pose, quickly determine the initial pose value and resolve the ambiguity of multiple pose solutions, and provide a reliable initial value for subsequent nonlinear optimization. A nonlinear optimization function is constructed with the goal of minimizing the sum of squared reprojection errors of all feature points. The Levenberg-Marquardt algorithm is used for iterative optimization, and after convergence, the high-precision six-degree-of-freedom pose matrix corresponding to the frame image is obtained.

[0036] S5. Deformation Calculation: Under stable structural conditions, continuously acquire N frames (N≥10) of cooperative target images and calculate the target pose for each frame; remove abnormal pose data, and perform a weighted average of the effective single-frame poses to obtain a high-precision initial reference pose matrix. The pose calculated in real time As a measurement value, it is compared with the reference value; the change in pose of the structural component before and after deformation is calculated, i.e., the deformation; finally, the pose transformation matrix is ​​converted into translation vectors and Euler angles to obtain the true six-degree-of-freedom deformation of the structural component (6D deformation). Specifically, the 3σ criterion is used to remove abnormal pose data; The formula for calculating deformation is as follows: ; It outputs and stores deformation data in real time, automatically alarms when thresholds are exceeded, and enables long-term monitoring.

[0037] Example 1 This invention provides a method for measuring the deformation of large structural components based on visual measurement, comprising the following steps: S1. System Setup and Calibration: A large-scale structural component deformation measurement system based on vision measurement is set up, including a measurement reference platform 3, a measurement camera 2, an active illumination unit, a cooperative target 1, an image processing and pose calculation unit, and a deformation calculation unit; the cooperative target 1 is rigidly fixed on the surface of the structural component 5 to be measured, and the measurement camera 2 is fixed on the measurement reference platform 3; see Figure 2 for a schematic diagram of the measurement principle. See Figure 3The feature markers of the cooperative target 1 are distributed on the target plane 103 and the pillar 102. The target plane 103 is 60mm×60mm in size, and the pillar 102 is 30mm in height. Five high-precision circular features 101 are designed on the target plane 103, adopting an asymmetric topological layout. The feature point coordinates have no rotational polarity ambiguity, which can realize the rapid sorting and unique matching of feature points. Three circular features 104 are designed on the pillar 102 to further improve the solution accuracy. There is a certain distance between the feature points, which reserves sufficient space for feature point edge extraction and avoids feature adhesion. The surface of the collaborative target 1 is coated with a coating adapted to the space environment. The base area is coated with a matte black coating, and the areas of the annular feature 101 and the dot feature 104 are coated with a matte white coating with high reflectivity to maintain the high contrast and recognizability of the feature points. The three-dimensional coordinates of all feature points of the collaborative target 1 are calibrated by a high-precision coordinate measuring machine, providing accurate three-dimensional reference target coordinates for subsequent PnP pose calculation.

[0038] The measuring camera 2 is a monocular vision camera. It employs a 120mm focal length optical lens and an image sensor resolution of 2048×2048, capable of covering a deformation range of ±5mm for a 60mm target at a working distance of 1200mm. The measuring camera 2 is integrated with the active illumination unit and coaxially positioned, meaning it is equipped with 808nm VCSEL laser active illumination, with power adaptively adjusted according to the image processing algorithm. To match the laser illumination energy, the optical lens of the measuring camera 2 undergoes an 808nm±5nm bandpass coating to suppress solar energy at 808nm.

[0039] S2. Image Acquisition and Laser Anti-interference Imaging: The measuring camera 2 acquires images containing the cooperative target 1 in real time, and simultaneously turns on the narrowband laser active illumination, suppressing interference from sunlight and ambient stray light through the narrowband filter; S3. Fast Image Processing and Feature Extraction: The Otsu method adaptive threshold segmentation algorithm is used to binarize the single-channel image, automatically adapting to illumination fluctuations and segmenting the target feature foreground and background regions; the Suzuki85 contour tracking algorithm is used to extract closed contours in the image, and preliminary screening is performed based on preset feature size range and contour moment threshold to filter out invalid interference contours caused by stray light and laser speckle; ellipse fitting is performed on the filtered effective contours to calculate the contour center coordinates and roundness parameters, and pseudo-features that do not meet the roundness and size requirements are filtered out in combination with the geometric design constraints of the target features to obtain a set of effective feature points; feature sorting is completed based on asymmetric topological relationships to establish a one-to-one matching between 2D image points and 3D target points, and to establish the correspondence between 2D and 3D feature points; S4. Pose calculation: The P3P algorithm is used to solve the initial pose value, and then the Levenberg-Marquardt algorithm is used to perform nonlinear optimization on the reprojection error to output a high-precision six-degree-of-freedom pose matrix. S5. Deformation Calculation: Under the stable state of structural component 5, continuously acquire no less than 10 frames of cooperative target images, and calculate the target pose for each frame; use the 3σ criterion to remove abnormal pose data, and perform a weighted average of the effective single-frame poses to obtain a high-precision initial reference pose matrix. The pose calculated in real time As a measurement value, it is compared with the reference value; the pose change of structural component 5 after deformation (i.e., structural component 4 after deformation) is calculated, that is, the deformation amount; finally, the pose transformation matrix is ​​converted into translation vector and Euler angle form, that is, the true six-degree-of-freedom deformation amount of structural component 5 is obtained. The formula for calculating deformation is as follows: ; It outputs and stores deformation data in real time, automatically alarms when thresholds are exceeded, and enables long-term monitoring.

[0040] To demonstrate the effectiveness of the technical solution of this invention, the system of this embodiment underwent a physical accuracy test at a distance of 1200mm ± 10mm between the measuring camera 2 and the cooperative target 1. In the test system, the cooperative target 1 was mounted on a six-legged parallel adjustment platform with a motion accuracy of 3µm, which could serve as a 6-DOF motion generator to simulate the deformation of large structural components. The platform's motion pose was used as a reference value and compared with the measured pose of the measuring camera 2. A total of 9 independent measurement experiments were completed, and the accuracy test results are shown in Table 1.

[0041] Table 1 Measurement Error Results

[0042] In the translation dimension, the maximum absolute errors in the tx, ty, and tz directions are 0.0467 mm, 0.0551 mm, and 0.1651 mm, respectively; in the rotational deformation measurement dimension, the maximum absolute errors in the rx, ry, and rz directions are 0.0151°, 0.0236°, and 0.0100°, respectively. Test results show that the measurement method proposed in this invention achieves high-precision measurement in all 6D dimensions, with small error fluctuations in multiple sets of repeated measurements, demonstrating excellent measurement stability and robustness. It can fully meet the engineering application requirements for local deformation measurement of large on-orbit structural components.

[0043] It should be understood that the various forms of processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this invention disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this invention can be achieved, and this is not limited herein.

[0044] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for measuring the deformation of large structural components based on vision measurement, characterized in that: Includes the following steps: S1. Construct a large-scale structural component deformation measurement system based on vision measurement, including a measurement reference platform, a measurement camera, an active illumination unit, a cooperative target, an image processing and pose calculation unit, and a deformation calculation unit; rigidly fix the measurement camera to the measurement reference platform, and rigidly fix the cooperative target on the surface of the large-scale structural component; calibrate the measurement camera, import the three-dimensional feature coordinates of the target, and complete the coordinate system binding; S2. The measuring camera acquires images containing the cooperative target in real time, simultaneously activates the active illumination unit, and suppresses ambient light interference through the measuring camera's optical lens to achieve stable imaging; S3. Perform adaptive threshold binarization, contour tracking and ellipse fitting on the acquired image to extract effective feature points, and sort them based on asymmetric topological relationships to complete 2D-3D feature point matching; S4. Perform pose calculation on the feature point matching results and output a six-degree-of-freedom pose matrix; S5. Establish a reference pose under the stable state of the structural component, perform matrix operations on the real-time pose and the reference pose, and calculate and output the six-degree-of-freedom deformation of the structural component.

2. The method for measuring the deformation of large structural components based on vision measurement according to claim 1, characterized in that: The cooperative target adopts an asymmetric topological multi-feature redundancy design, containing no less than 3 feature points; the feature points are circular features and / or dot features.

3. The method for measuring the deformation of large structural components based on vision measurement according to claim 1, characterized in that: In step S1, the calibration of the measuring camera specifically includes intrinsic parameters, distortion coefficients, and extrinsic parameters; the distortion coefficients include radial distortion and tangential distortion; the extrinsic parameters are the rotation matrix and translation vector of the measuring camera relative to the measuring reference platform.

4. The method for measuring the deformation of large structural components based on vision measurement according to claim 1, characterized in that: The active illumination unit is coaxially arranged with the measuring camera, and the illumination direction is aligned with the cooperative target area; In step S2, the active illumination unit uses narrowband laser active illumination, and the optical lens of the measuring camera is a narrowband bandpass coated lens. The wavelength of the narrowband laser active illumination is consistent with the wavelength of the narrowband bandpass coated lens, so that the active illumination energy is much greater than the energy of sunlight in the corresponding wavelength band.

5. The method for measuring the deformation of large structural components based on vision measurement according to claim 1, characterized in that: Step S3 specifically includes: using the Otsu method adaptive threshold segmentation algorithm to binarize the single-channel image, automatically adapting to illumination fluctuations, and segmenting the target feature foreground and background regions; using the Suzuki85 contour tracking algorithm to extract closed contours in the image and complete preliminary screening; performing ellipse fitting on the screened effective contours to remove false features; completing feature sorting based on topological relationships, establishing a one-to-one match between 2D image points and 3D target points, and establishing the correspondence between 2D and 3D feature points.

6. The method for measuring the deformation of large structural components based on vision measurement according to claim 1, characterized in that: The pose calculation in step S4 specifically includes: using the P3P algorithm to perform coarse pose calculation, and then using the LM algorithm to perform nonlinear optimization with the goal of minimizing the reprojection error, and outputting a six-degree-of-freedom pose matrix.

7. The method for measuring the deformation of large structural components based on vision measurement according to claim 1, characterized in that: In step S5, the reference pose is obtained by continuously collecting no less than 10 frames of pose data, removing outliers using the 3σ criterion, and then weighting the average.

8. The method for measuring the deformation of large structural components based on vision measurement according to claim 7, characterized in that: The six-degree-of-freedom deformation is calculated through the pose transformation matrix, and the matrix is ​​analyzed into translation vectors and Euler angles for output.

9. A vision-based deformation measurement system for large structural components, used to perform the vision-based deformation measurement method for large structural components as described in claim 1, characterized in that: It includes a measurement reference platform, a measurement camera, an active illumination unit, a cooperative target, an image processing and pose calculation unit, and a deformation calculation unit; The measuring camera is fixed on the measuring reference platform; the active illumination unit is coaxially arranged with the measuring camera, and the illumination direction is aligned with the cooperative target area. The collaborative target is rigidly fixed to the surface of a large structural component; The image processing and pose calculation unit receives image data output from the measurement camera and sequentially performs image preprocessing, feature detection and matching, coarse pose calculation and fine pose calculation. The deformation calculation unit is used to establish the reference pose and calculate the six-degree-of-freedom deformation based on the real-time pose and the reference pose.

10. A large structural component deformation measurement system based on vision measurement according to claim 9, characterized in that: The optical lens of the measuring camera is coated with a narrow bandpass coating, and the band of the narrow bandpass coating is matched with the active illumination band. The cooperative target has an asymmetric topological structure, and the three-dimensional coordinates of the feature points are calibrated using a coordinate measuring machine. The deformation calculation unit includes a reference pose establishment module and a real-time deformation calculation module; The baseline pose establishment module uses the 3σ criterion to eliminate abnormal data, and the real-time deformation calculation module outputs six-degree-of-freedom deformation through pose matrix transformation.

Images