A general thermal drift suppression method for visual displacement measurement and visual measurement device

Abstract

The application relates to the technical field of visual measurement, and discloses a general thermal drift suppression method for visual displacement measurement and a visual measurement device. The general thermal drift suppression method for visual displacement measurement comprises the following steps: acquiring reference information and current information of common imaging of a reference calibration board and a target object; obtaining a thermal drift correction constraint model according to the reference information and the current information; and obtaining coordinate information of the target object according to the current information and the thermal drift correction constraint model. The application can improve the measurement precision in actual complex environments.

Description

Technical Field

[0001] This application relates to the field of visual measurement technology, and in particular to a general thermal drift suppression method and visual measurement device for visual displacement measurement. Background Technology

[0002] Visual measurement technology is applicable to structural displacement health monitoring systems for industrial and civil infrastructure such as wind power, nuclear power, lifting equipment, bridges, buildings, and tunnels. Structural displacement is a crucial monitoring indicator for assessing structural operating status, identifying abnormal behavior, and ensuring the long-term safe operation of engineering systems, possessing significant engineering significance and application value. In recent years, with the rapid development of computer vision technology, image-based non-contact displacement measurement methods have been widely applied in various engineering scenarios. These methods offer advantages such as non-contact operation, long distance, high precision, fast response speed, and relatively low cost, and have gradually become a research hotspot and development direction in the field of structural health monitoring. With the widespread application of vision-based structural displacement measurement technology in engineering monitoring, addressing the performance fluctuations of imaging systems in complex environments has become a key challenge in improving the measurement accuracy and stability of the system. Especially in long-term outdoor monitoring tasks, camera systems are significantly affected by changes in ambient temperature, easily generating image thermal drift errors, which directly affect the positioning and tracking accuracy of target points.

[0003] To address the problem of image thermal drift error, existing research mainly focuses on two compensation approaches: parametric modeling-based methods and data-driven methods. Parametric modeling methods assume that the camera's intrinsic and extrinsic parameters (such as focal length, principal point coordinates, and camera attitude) change deterministically with temperature, thus establishing an analytical function model of thermal drift. Data-driven methods, on the other hand, collect a large amount of image drift data under varying temperatures, constructing a statistical mapping relationship between temperature and pixel drift. These models are typically trained using machine learning algorithms (such as linear regression, support vector machines, and neural networks) to replace explicit modeling of the complex physical processes.

[0004] However, existing thermal drift compensation methods often struggle to balance modeling accuracy, method versatility, and engineering applicability. On the one hand, parametric modeling methods heavily rely on idealized assumptions (such as linear response to thermal stability) and are highly sensitive to multi-factor interference in real-world environments. On the other hand, while data-driven methods can address nonlinear variations to some extent, their model generalization ability is limited, typically requiring retraining and calibration for specific cameras, making them difficult to apply in engineering settings with diverse equipment types and drastic temperature variations.

[0005] Therefore, existing technologies still need to be improved and developed. Summary of the Invention

[0006] The main objective of this application is to provide a universal thermal drift suppression method and visual measurement device for visual displacement measurement, aiming to solve the problem that in the prior art, camera systems are prone to image thermal drift errors due to changes in ambient temperature during outdoor monitoring, which affects the positioning and tracking accuracy of target points and results in low accuracy of visual displacement measurement.

[0007] The first aspect of this application provides a general thermal drift suppression method for visual displacement measurement. The general thermal drift suppression method for visual displacement measurement includes the following steps: obtaining reference information and current information of the joint imaging of a reference calibration plate and a target object; obtaining a thermal drift correction constraint model based on the reference information and the current information; and obtaining the coordinate information of the target object based on the current information and the thermal drift correction constraint model.

[0008] Optionally, in one embodiment of this application, the reference calibration plate has multiple calibration points, the target object has measurement points, the reference information includes reference data of multiple calibration points under the corrected optical path, and the current information includes current data of multiple calibration points under the corrected optical path and data to be corrected of measurement points under the measurement optical path; the step of obtaining the reference information and current information of the reference calibration plate and the target object jointly imaging specifically includes: obtaining a reference frame image of the reference calibration plate and the target object jointly imaging at an initial temperature, and obtaining reference data of multiple calibration points under the corrected optical path based on the reference frame image; obtaining a current frame image of the reference calibration plate and the target object jointly imaging at the current temperature, and obtaining current data of multiple calibration points under the corrected optical path and data to be corrected of measurement points under the measurement optical path based on the reference frame image.

[0009] Optionally, in one embodiment of this application, acquiring a reference frame image jointly formed by the reference calibration plate and the target object at an initial temperature specifically includes: acquiring an initial image jointly formed by the reference calibration plate and the target object at an initial temperature; if multiple calibration points in the initial image do not overlap with the measurement points, then using the initial image as a reference frame image; if there are calibration points in the initial image that overlap with the measurement points, then after moving the visual measurement device, acquiring an updated image jointly formed by the reference calibration plate and the target object until multiple calibration points in the updated image do not overlap with the measurement points, then using the updated image as a reference frame image.

[0010] Optionally, in one embodiment of this application, the reference data includes a reference corner coordinate set, the current data includes a current corner coordinate set, and the data to be corrected includes measurement point coordinates; obtaining reference data for multiple calibration points under the corrected optical path based on the reference frame image specifically includes: dividing the reference frame image into regions to obtain multiple reference local network sub-regions; extracting feature points for each reference local network sub-region to obtain a reference corner coordinate set corresponding to each reference local network sub-region; obtaining current data for multiple calibration points under the corrected optical path and measurement point data to be corrected under the measurement optical path based on the reference frame image specifically includes: dividing the current frame image into regions to obtain multiple current local network sub-regions, wherein the current local network sub-regions correspond to the reference local network sub-regions; extracting feature points for each current local network sub-region to obtain a current corner coordinate set and measurement point coordinates corresponding to each current local network sub-region.

[0011] Optionally, in one embodiment of this application, the thermal drift correction constraint model includes multiple geometric transformation models; the step of obtaining the thermal drift correction constraint model based on the reference information and the current information specifically involves: for multiple reference local network sub-regions and corresponding multiple current local network sub-regions, obtaining multiple geometric transformation models based on multiple reference corner coordinate sets and multiple current corner coordinate sets.

[0012] Optionally, in one embodiment of this application, when the geometric transformation model is a local homography matrix, the local homography matrix is ​​represented as:

[0013] ;

[0014] in, Indicates the first The local homography matrix of a current local network sub-region Indicates the horizontal scaling factor. Indicates horizontal shear strength. Indicates the amount of horizontal translation. Indicates vertical shear strength. This represents the scaling factor in the vertical direction. This indicates the vertical translation amount. Indicates the perspective distortion coefficient in the horizontal direction. This represents the perspective distortion coefficient in the vertical direction. Represents the normalization factor. Represents the set of real numbers;

[0015] When the geometric transformation model is a local affine transformation matrix, the local affine transformation matrix is ​​expressed as:

[0016] ;

[0017] in, Indicates the first i A local unit matrix of a current local website sub-region. These represent the scaling factors in the horizontal and vertical directions, respectively. and These represent the shear coefficients in the horizontal and vertical directions, respectively. and These represent the translation amounts in the horizontal and vertical directions, respectively.

[0018] When the geometric transformation model is a high-order nonlinear distortion model, the modeling method for the high-order nonlinear distortion model is as follows:

[0019] ;

[0020] in, The x-coordinate of the pixel in the reference frame, The vertical coordinate of the pixel in the reference frame. The x-coordinate after drifting in the current frame. The vertical coordinate after drifting in the current frame. The polynomial coefficients are used to map the x-coordinate of a pixel in the reference frame to the shifted x-coordinate in the current frame. The polynomial coefficients are the coordinates of a pixel in the reference frame mapped to its shifted coordinate in the current frame, where N is the polynomial order; or

[0021] The modeling method for the high-order nonlinear distortion model is as follows:

[0022] ;

[0023] ;

[0024] in, , Indicates the radial distance to the center of the image; , , Radial distortion coefficient; , denoted as the tangential distortion coefficient.

[0025] Optionally, in one embodiment of this application, the coordinate information includes the coordinates of multiple target points;

[0026] The step of obtaining the coordinate information of the target object based on the current information and the thermal drift correction constraint model specifically involves:

[0027] Based on the coordinates of all the measured points and the corresponding geometric transformation models, the coordinates of multiple target points are obtained;

[0028] The coordinates of the target point are represented as follows:

[0029] ;

[0030] in, This indicates the target point coordinates after the measurement point coordinates have been corrected. Indicates the first Geometric transformation model of a current local network sub-region The inverse mapping, the geometric transformation model Local homography matrix Local affine matrix Or a higher-order nonlinear distortion model, This indicates the coordinates of the measurement point in the current frame image.

[0031] Optionally, in one embodiment of this application, after obtaining the coordinate information of the target object based on the current information and the thermal drift correction constraint model, the method further includes: obtaining an image scale factor, and obtaining the actual displacement based on the target point coordinates, the measurement point coordinates, and the image scale factor.

[0032] A second aspect of this application provides a visual measurement device for implementing a universal thermal drift suppression method for visual displacement measurement as described in any of the above-described solutions. The visual measurement device includes a processor, a camera, a beam splitter, a collimating lens, and a reference calibration plate. The processor is communicatively connected to the camera. The camera, the beam splitter, the collimating lens, and the reference calibration plate are arranged sequentially. The two sides of the beam splitter face the reference calibration plate and the target structure, respectively. The reference calibration plate has a plurality of calibration points arranged in an array, and the target structure has measurement points.

[0033] The camera is used to acquire reference information and current information of the joint imaging of the reference calibration plate and the target object, and sends them to the processor; the processor obtains a thermal drift correction constraint model based on the reference information and the current information, and obtains the coordinate information of the target object based on the current information and the thermal drift correction constraint model.

[0034] Optionally, in one embodiment of this application, the beam splitter is provided with an infrared bandpass filter on the side facing the target structure, and the camera, the beam splitter, the infrared bandpass filter, the collimating lens and the reference calibration plate are disposed inside an optical dark box, with the target structure located outside the optical dark box.

[0035] Beneficial effects: This application provides a general thermal drift suppression method and visual measurement device for visual displacement measurement. This application eliminates the cumulative error caused by environmental changes due to the separation of traditional calibration and measurement through real-time common field of view correction. Furthermore, it achieves sub-pixel level thermal drift correction accuracy through block homography transformation (a type of thermal drift correction constraint model), thereby improving the measurement accuracy in real complex environments. Moreover, it does not depend on specific camera models or temperature sensors and can be adapted to different industrial camera and lens combinations, ensuring long-term stability in real complex environments. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0037] Figure 1 This is a three-dimensional structural diagram of a preferred embodiment of the visual measurement device of this application;

[0038] Figure 2 This is a schematic diagram of the common-view imaging module in a preferred embodiment of the visual measurement device of this application;

[0039] Figure 3 This is a schematic diagram of the optical coupling module in a preferred embodiment of the visual measurement device of this application;

[0040] Figure 4 This is a schematic diagram of the planar calibration marks on the reference calibration plate in a preferred embodiment of the visual measurement device of this application;

[0041] Figure 5 This is a flowchart of a preferred embodiment of the general thermal drift suppression method for visual displacement measurement in this application;

[0042] Figure 6 This is a schematic diagram showing the correspondence between calibration points, measurement points, and local homography matrix in a preferred embodiment of the general thermal drift suppression method for visual displacement measurement in this application;

[0043] Figure 7 This is a schematic diagram of the specific implementation steps in a preferred embodiment of the general thermal drift suppression method for visual displacement measurement in this application.

[0044] Explanation of reference numerals in the attached figures:

[0045] 1. Optical dark box; 2. Camera; 3. Beam splitter prism; 4. Infrared bandpass filter; 5. Measurement point; 6. Collimating lens; 7. Reference calibration plate; 8. Coaxial quick-release plate; 9. Complementary lighting device.

[0046] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0047] To make the objectives, technical solutions, and effects of this application clearer and more explicit, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. The described embodiments are only possible technical implementations of this application and not all possible implementations. Based on the embodiments in this application, those skilled in the art can obtain other embodiments without creative effort, and these embodiments are also within the protection scope of this application.

[0048] To address the issue that camera systems are prone to image thermal drift errors due to changes in ambient temperature during outdoor monitoring, which affects the positioning and tracking accuracy of target points and leads to low accuracy in visual displacement measurement, this application eliminates the cumulative errors caused by environmental changes due to the separation of traditional calibration and measurement through real-time common field of view correction. Furthermore, by using block homography transformation (a type of thermal drift correction constraint model), the thermal drift correction accuracy reaches the sub-pixel level, thereby improving the measurement accuracy in real-world complex environments. Moreover, it is not dependent on specific camera models or temperature sensors and can be adapted to different industrial camera and lens combinations, ensuring long-term stability in real-world complex environments.

[0049] The technical solutions of this application will be described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0050] like Figure 1 As shown in the preferred embodiment of this application, the visual measurement device is used to implement a general thermal drift suppression method for visual displacement measurement. The visual measurement device includes a processor, a camera 2, a beam splitter 3, a collimating lens 6, and a reference calibration plate 7. The processor is communicatively connected to the camera 2. The camera 2, the beam splitter 3, the collimating lens 6, and the reference calibration plate 7 are arranged sequentially. The two sides of the beam splitter 3 face the reference calibration plate 7 and the target structure, respectively. The reference calibration plate 7 has multiple calibration points arranged in an array, and the target structure has measurement points 5.

[0051] The camera 2 is used to acquire reference information and current information of the reference calibration plate 7 and the target object being imaged together, and sends them to the processor; the processor obtains a thermal drift correction constraint model based on the reference information and the current information, and obtains the coordinate information of the target object based on the current information and the thermal drift correction constraint model.

[0052] It is understood that the visual measurement device of this application embodiment has a stable structure and compact size, and can realize the accurate correction of thermal drift error in the visual measurement system, and is suitable for long-term displacement monitoring tasks of various engineering structures.

[0053] Specifically, see Figure 1 and Figure 2 The industrial camera 2, the beam splitter 3, and the collimating lens 6 (i.e., the convex lens) together constitute the common-view imaging visual measurement module (common-view imaging module). Figure 2 The main channel is used for displacement measurement of the target structure (measurement point 5 is located on the target structure), and the secondary channel is used to observe the embedded calibration plane (i.e., the plane with multiple calibration points on the reference calibration plate 7). Both channels are projected onto the same image sensor via an optical coupling module. See also... Figure 3 The optical coupling module includes a beam splitter 3 for splitting and combining light paths, a collimating lens 6 for adjusting the consistency of the imaging focal plane, and an infrared bandpass filter 4 for filtering out stray light. The collimating lens 6 and the filter SM01 (model identifier of the optical element fixing device) are fixed in a lens sleeve and connected to the beam splitter 3 fixed in the prism cube by threads. The entire module is only 4.9cm×4.9cm×4.1cm in size, with a compact structure. It can realize the optical path coupling of the main channel and the sub-channel, and ensure that the two fields of view are co-viewed on the same image sensor.

[0054] In this embodiment, a passive planar calibration mark (calibration point) is used on the reference calibration plate 7. This mark is fixed in the secondary channel optical path to provide a stable reference feature point, assisting in thermal drift modeling and calibration. It is made of materials with low thermal expansion coefficient and strong geometric stability (such as glass-ceramic substrate, metal etched plate, composite structure plate) to ensure dimensional stability and no deformation under different ambient temperatures. The mark pattern (i.e., the image of the calibration plane) can adopt a checkerboard pattern. Figure 4 (a) of the circular dot matrix calibration plate Figure 4 (b) of the above), or an optical mask with a high-contrast edge structure ( Figure 4 (c) These patterns facilitate high-precision extraction of corner points or centroids in images, and can form a set of feature points that are spatially uniform and highly sensitive to thermal drift.

[0055] In one embodiment of this application, the beam splitter 3 is provided with an infrared bandpass filter 4 on the side facing the target structure. The camera 2, the beam splitter 3, the infrared bandpass filter 4, the collimating lens 6 and the reference calibration plate 7 are disposed inside the optical dark box 1, and the target structure is located outside the optical dark box 1.

[0056] Specifically, the bottom wall of the optical dark box 1 is detachably equipped with a coaxial quick-release plate 8. The camera 2, beam splitter prism 3, collimating lens 6, and reference calibration plate 7 are connected to the coaxial quick-release plate 8, which has corresponding limiting grooves to facilitate rapid assembly of the vision measurement device. A supplementary lighting device 9 is installed above the reference calibration plate 7 to provide supplementary lighting for the industrial camera 2's imaging of the calibration plane. It is understood that the optical dark box 1, infrared bandpass filter 4, and supplementary lighting device 9 can suppress interference from changes in day and night illumination, ensuring stable extraction of feature points, and are suitable for low-light scenarios such as tunnels and under bridges.

[0057] The visual measurement device of this application embodiment has a stable structure and compact size, ensuring deployability in complex real-world environments and enabling long-term stable operation under varying ambient temperatures.

[0058] Understandably, in the optical coupling module, the collimating lens 6 can be adjusted to change the imaging distance of the sub-channel reference plane by replacing it with a collimating lens 6 of different focal lengths or by using multiple lens groups, thereby adapting to the needs of different working distances or reference plane deployment depths; the filters can be flexibly configured. In addition to the fixed-band infrared bandpass filter 4, different band filters (such as near-infrared, short-wave infrared, etc.) can be selected to adapt to different light source environments, or combined filters (such as bandpass + long-wave pass composite filters) can be used to improve the signal-to-noise ratio and anti-interference capability; in specific environments, polarizers or polarization beam splitters can be used to separate reflected light and direct light, improving the imaging contrast of the reference mark, which is particularly suitable for structural scenarios with strong surface reflection interference; under conditions where some optical axes cannot be collinear or sensor size is limited, miniature reflectors, steering prisms, or refracting wedges can be set to slightly deflect the sub-channel optical path to achieve image alignment or field-of-view center alignment, improving imaging consistency. The planar calibration marker offers a variety of planar marker point options. Different types of patterned marker points (such as circular dot matrix, checkerboard, coded marker points, optical masks, etc.) can be selected for the reference area and measurement area to enhance the difference and stability of feature point recognition, and improve the matching accuracy and automation of the algorithm. When it is necessary to perform three-dimensional structural deformation monitoring or multi-angle attitude estimation, the planar marker can be extended to a three-dimensional target dot matrix (such as spherical target points, spatial target columns, 3D marker blocks, etc.). Combined with stereo vision or multi-view reconstruction technology, the displacement measurement of the structural target in the XYZ three-dimensional direction can be realized, thus expanding the functional boundaries of the system.

[0059] In this application, a planar marker design based on non-uniform distribution is adopted to adapt to various complex scenarios. Traditional reference marker boards mostly use regular dot matrices (such as checkerboard or circular dot matrices). However, in actual engineering, due to uneven temperature changes and uneven spatial distribution of marker points, a non-uniform distribution planar marker strategy is adopted: high-density reference points are placed in key areas, and low-density reference points are placed in non-critical areas or areas with poor visibility; a hybrid nested design of large and small patterns is adopted to balance overall rigid alignment and local fine modeling; the reference point design can be targeted to the thermally sensitive areas and important structural parts of the target structure. Through reasonable irregular marker pattern design, the modeling stability and data recovery integrity of the system under actual complex working conditions such as local fine calibration and strong thermal drift spatial non-uniformity can be significantly enhanced.

[0060] To address the technical problems of existing visual measurement systems being susceptible to temperature changes during long-term monitoring, leading to image drift and decreased measurement accuracy, this application presents a universal and high-precision thermal drift suppression method for visual displacement measurement. Based on a measurement-correction common field-of-view imaging configuration, an invariant planar reference calibration plate is embedded within the measurement field of view. A compact dual-channel common-field-of-view optical path is used to achieve spatial fusion of the measurement and monitoring areas, fundamentally eliminating the error accumulation problem caused by the separation of calibration and measurement in traditional visual systems. This application constructs a thermal drift correction constraint model using a stable reference plane in the measurement field of view and employs a block-based correction strategy on the imaging plane to model and compensate for image point drift, thereby effectively suppressing thermally induced spatial non-uniform image distortion. Specifically, this method describes the thermal drift characteristics of different regions through local homography transformation / distortion model construction, achieving high-precision zonal correction of full-field image drift.

[0061] This application significantly improves the thermal robustness and displacement measurement accuracy of visual measurement systems in complex and variable environments. It exhibits stronger robustness and adaptability, and is independent of camera model and lens type, making it particularly suitable for long-term structural health monitoring of outdoor infrastructure such as wind power, nuclear power, bridges, and tunnels. This application features a compact structure, high stability, and easy deployment. Experiments have demonstrated that it maintains good measurement stability and correction accuracy even under large temperature differences, improving long-term stability and measurement accuracy in real-world complex environments, and has broad practical application prospects.

[0062] The preferred embodiment of this application describes a general thermal drift suppression method for visual displacement measurement, such as... Figure 5 As shown, the general thermal drift suppression method for visual displacement measurement includes the following steps:

[0063] In step S101, reference information and current information of the joint imaging of the reference calibration plate and the target object are obtained.

[0064] In one possible implementation, the reference calibration plate has multiple calibration points, the target object has measurement points, the reference information includes reference data for the multiple calibration points under the corrected optical path, and the current information includes current data for the multiple calibration points under the corrected optical path and data to be corrected for the measurement points under the measurement optical path. A reference frame image is acquired, showing the reference calibration plate and the target object together at an initial temperature, and the reference data for the multiple calibration points under the corrected optical path is obtained based on the reference frame image. A current frame image is acquired, showing the reference calibration plate and the target object together at the current temperature, and the current data for the multiple calibration points under the corrected optical path and the data to be corrected for the measurement points under the measurement optical path are obtained based on the reference frame image.

[0065] Understandably, the initial temperature is the temperature at which the camera initially operates when it is turned on, while the current temperature is the real-time temperature after the camera has been running continuously for a period of time.

[0066] This application's thermal drift modeling can capture and model local image drift behavior caused by temperature changes. During thermal drift modeling, the entire image is divided into several fixed grid regions (e.g., 3×3, 5×5, etc.), each sub-region covering a limited pixel range. The region division method can be flexibly adjusted according to the field of view size, reference point density, and expected non-uniform drift scale. In each frame, stable feature points (e.g., checkerboard or dot array) embedded in the calibration plane are extracted. The least squares method or the RANSAC algorithm (RANdom Sampling Consensus) is used to fit the reference feature points within each sub-region, estimating the geometric transformation model (local homography matrix) of that region relative to the initial time step. This approximates the local geometric deformation of the region under temperature changes; the simplest moments corresponding to all sub-regions are... Together, they constitute the thermal drift deformation field of the entire image, providing a modeling basis for subsequent point-by-point correction of target point positions. This module fully considers the non-uniform distribution characteristics of thermal drift in image space, and compared with traditional modeling methods that assume overall rigid body drift of the image, it has higher fitting accuracy and adaptability.

[0067] In one possible implementation, an initial image is acquired whereby the reference calibration plate and the target object are jointly imaged at an initial temperature. If multiple calibration points in the initial image do not overlap with the measurement points, the initial image is used as a reference frame image. If there are calibration points in the initial image that overlap with the measurement points, an updated image is acquired whereby the reference calibration plate and the target object are jointly imaged after the vision measurement device is moved, until multiple calibration points in the updated image do not overlap with the measurement points, and the updated image is used as a reference frame image.

[0068] In one possible implementation, the reference data includes a reference corner coordinate set, the current data includes a current corner coordinate set, and the data to be corrected includes measurement point coordinates. The reference frame image is divided into regions to obtain multiple reference local network sub-regions; feature points are extracted from each reference local network sub-region to obtain the reference corner coordinate set corresponding to each reference local network sub-region. The current frame image is divided into regions to obtain multiple current local network sub-regions, wherein each current local network sub-region corresponds to a reference local network sub-region; feature points are extracted from each current local network sub-region to obtain the current corner coordinate set and measurement point coordinates corresponding to each current local network sub-region. The current local network sub-region is the same as the reference local network sub-region.

[0069] Specifically, in the process of image region segmentation, a frame of image is assumed to be a two-dimensional pixel array. ,in The image is divided into n×m local grid sub-regions, where n is the number of grid rows and m is the number of grid columns. Let the nth sub-region be the nth grid row. Each sub-region is . For a two-dimensional pixel array image function, it represents the image in coordinates. Pixel value at; The domain of an image is the set of coordinates of all pixels in the image. For the first A local grid sub-region, For sub-region indexing, ( ).

[0070] Understandably, region partitioning can employ adaptive partitioning, dynamically determining region boundaries based on factors such as reference corner density, image texture complexity, or image brightness gradient. For example, a region with high corner density can be partitioned into smaller areas for finer local modeling, while a region with low corner density can be processed as a larger area to improve robustness and computational efficiency. Region partitioning can also be based on temperature gradients. When the system can acquire external or image temperature distribution information (such as infrared images), image partitioning can be set according to temperature change gradients. Regions with drastic temperature changes are subdivided more densely, while regions with gradual temperature changes are merged, ensuring that the region partitioning more closely reflects actual thermal deformation behavior.

[0071] During feature point extraction, stable feature points (such as checkerboard corner points) on the reference plane are extracted in each frame of the image. The sub-regions in the initial frame (reference frame) are also extracted. The set of coordinates of the inner corner points is:

[0072] ;

[0073] in, This represents the reference corner coordinate set, i.e., the first corner coordinate set in the reference frame. The set of feature point coordinates for each sub-region (superscript 0 indicates the initial reference frame); For the reference frame Sub-region Homogeneous coordinates of the feature points; For the reference frame Sub-region Feature points coordinate; For the reference frame Sub-region Feature points coordinate, It is the first Number of available reference points in each sub-region This represents the transpose operator;

[0074] The set of corresponding corner points in the current frame (after temperature drift) is as follows:

[0075] ;

[0076] in, This represents the current set of corner coordinates, specifically the i-th corner point in the current frame (after temperature drift). The set of corresponding feature points of each sub-region (superscript) Indicates time The set of corresponding feature points for each sub-region at time (time). For the current frame, the first Sub-region Homogeneous coordinates of feature points.

[0077] In step S102, a thermal drift correction constraint model is obtained based on the reference information and the current information.

[0078] It should be noted that, see Figure 7 The definition is quite broad. Homography is one method in thermal drift modeling. Depending on the applicable scenario, thermal drift modeling can choose affine transformation, higher-order nonlinear distortion models, or homography. Note that the versatility of thermal drift modeling lies in the fact that the appropriate method can be determined based on the thermal drift modeling time, the magnitude of image point drift, and the pattern of local image point drift. First, we will explain thermal drift modeling using local homography.

[0079] In one possible implementation, the thermal drift correction constraint model includes multiple geometric transformation models. For multiple reference local network sub-regions and corresponding multiple current local network sub-regions, multiple geometric transformation models are obtained based on multiple reference corner coordinate sets and multiple current corner coordinate sets.

[0080] It should be noted that the geometric transformation model is one of the following: local homography transformation, local affine transformation, and higher-order nonlinear distortion model. In this embodiment, the geometric transformation model is local homography transformation (local homography matrix).

[0081] Specifically, see Figure 4 and Figure 6 The red markers represent calibration points. A point surrounded by multiple calibration points is a measurement point. A measurement point can be surrounded by multiple (e.g., 4, 5, or 6) calibration points. In this embodiment, 4 calibration points, 1 measurement point, and 1 local homography matrix correspond to each other. Figure 6 (a) in the middle has , , , , Figure 6 (b) in the middle has , , , and .

[0082] In the process of local homography matrix estimation, for each sub-region The local homography matrix is ​​fitted using the least squares method or the RANSAC method. , so that:

[0083] .

[0084] In one possible implementation, when the geometric transformation model is a local affine transformation matrix, the affine transformation matrix is ​​represented as:

[0085] ;

[0086] in, Indicates the first i A local unit matrix of a current local website sub-region. These represent the scaling factors in the horizontal and vertical directions, respectively. These represent the shear coefficients in the horizontal and vertical directions, respectively. These represent the translation amounts in the horizontal and vertical directions, respectively.

[0087] When the geometric transformation model is a high-order nonlinear distortion model, the high-order nonlinear distortion modeling method is as follows:

[0088] ;

[0089] in, For the pixel coordinates in the reference frame, These are the coordinates after the drift in the current frame. The coefficient matrix elements of the polynomial to be fitted, where N is the order of the polynomial (e.g., second, third, fourth, or higher).

[0090] Alternatively, radial and tangential distortion can be jointly modeled. The higher-order nonlinear distortion modeling method is as follows:

[0091] ;

[0092] ;

[0093] in, , Indicates the radial distance to the center of the image; The radial distortion coefficient is... , , Radial distortion coefficient; The tangential distortion coefficient is... , denoted as the tangential distortion coefficient.

[0094] Secondly, the modeling of thermal drift is explained using local affine transformation.

[0095] In applications where temperature changes are relatively gradual and image deformation is mainly manifested as translation and linear transformations, affine transformation models can be used instead of homography matrices for local modeling. Affine transformations have fewer parameters and lower computational complexity, making them suitable for drift estimation in scenarios with limited computing resources or high real-time requirements. In other words, within a local region of an image, if thermal drift is mainly manifested as approximately linear changes (such as translation, scaling, and shearing), affine transformations can be used instead of homography transformations. Affine models have fewer parameters and higher computational efficiency, making them suitable for conditions with small drift amplitudes or uniformly heated structures.

[0096] Next, we will explain the thermal drift modeling using high-order polynomial fitting.

[0097] For scenarios where thermal drift trajectories exhibit stable and regular characteristics, higher-order nonlinear distortion models (such as third- or fourth-order polynomials) can be introduced to perform pixel-level fitting of corner points or drift vectors. This method can more precisely express complex thermal deformation trends and is suitable for system environments with well-defined or pre-defined temperature change patterns, improving the accuracy of local modeling. In other words, using higher-order nonlinear distortion models to model local image thermal drift is suitable for situations where optical systems exhibit nonlinear imaging distortion under complex thermal fields.

[0098] Specifically, in the modeling scheme of the high-order nonlinear distortion model, the initial reference frame is used as a reference, and the drift of each reference marker point on the image plane under temperature perturbation is represented as a high-order polynomial function mapping of the input coordinates, such as:

[0099] ;

[0100] in, For the pixel coordinates in the reference frame, These are the coordinates after the drift in the current frame. The elements of the polynomial coefficient matrix to be fitted are: The polynomial coefficients are used to map the x-coordinate of a pixel in the reference frame to the shifted x-coordinate in the current frame. The polynomial coefficients are used to map the pixel ordinates in the reference frame to the shifted ordinates in the current frame, where N is the order of the polynomial (e.g., second, third, fourth, or higher).

[0101] Alternatively, a combined modeling approach using radial and tangential distortion can be employed.

[0102] ;

[0103] ;

[0104] in, , Indicates the radial distance to the center of the image; The radial distortion coefficient is... , , Radial distortion coefficient; The tangential distortion coefficient is... , denoted as the tangential distortion coefficient.

[0105] Compared to local homography correction methods, this modeling method can more accurately describe the nonlinear drift trend of image points when the imaging system undergoes higher-order distortion behaviors such as micro-displacement inside the lens, lens deformation, and thermal expansion of the CCD (Charge-Coupled Device) under the influence of thermal stress. Compared with geometric transformation-based models, higher-order distortion models have higher fitting ability and stronger expressive power, and are suitable for situations where the image edge region is unevenly drifted or has "barrel" / "pincushion" deviations. It is also suitable for visual system thermal drift compensation in long-term, high-temperature difference, and wide-angle lens scenarios.

[0106] In step S103, the coordinate information of the target object is obtained based on the current information and the thermal drift correction constraint model.

[0107] In one possible implementation, the coordinate information includes the coordinates of multiple target points. The coordinates of multiple target points are obtained based on all the measured point coordinates and the corresponding multiple geometric transformation models.

[0108] It should be noted that thermal drift calibration applies the local homography transformation results estimated in thermal drift modeling to the reverse drift correction of each pixel within the target measurement area. By constructing a local mapping function and compensating for thermally induced drift point by point, the true structural displacement in the image is recovered, thus improving measurement accuracy.

[0109] Specifically, to determine the target point mapping region, firstly, based on the image coordinate position, determine the sub-region to which the target point belongs in the current frame. The region index is determined by a preset grid division method. Taking a checkerboard mask as an example, quadrilaterals are divided based on the coordinates of the checkerboard corner points. Each sub-region corresponds to a local homography matrix. Let the pixel coordinates of the target point in the current frame image be:

[0110] ;

[0111] in, This indicates the coordinates of the measurement point in the current frame image. Indicates the target point in the current frame (time). In the image coordinate, Indicates the target point in the current frame image Coordinates. Based on the image's preset grid division rules, the system determines the local region number to which this point belongs. And determine the homography matrix corresponding to this region. Each local area The spatial extent is determined by the quadrilateral formed by reference feature points (such as checkerboard corner points), and the region index is obtained by judging the geometric fall of points.

[0112] Then, temperature drift compensation is performed for each sub-region. Extract the spatial variation between the checkerboard corner points in the reference frame (initial temperature state) and the current frame (heated drift state). In each sub-region... Within, extract its value in the reference frame. With the current frame The reference corner coordinates are used to calculate the homography matrix from the reference frame to the current frame using a formula.

[0113] Using the inverse matrix of the local homography matrix Measure the points in the current frame The position of the target point, which is back-projected onto the reference frame based on the thermal drift coordinates, is expressed as follows:

[0114] ;

[0115] in, This indicates the target point coordinates after the measurement point coordinates have been corrected. Indicates the first Geometric transformation model of a current local network sub-region The inverse mapping, the geometric transformation model It can be represented as a local homography matrix Local affine matrix Or a higher-order nonlinear distortion model, This indicates the coordinates of the measurement point in the current frame image.

[0116] This mapping process achieves spatial geometric compensation for the thermal drift error at that point, yielding the true position coordinates after removing temperature disturbances. This compensation process only offsets the image geometric drift caused by changes in ambient temperature, without affecting the actual displacement of the structure due to real loads or motion states, thus ensuring the accuracy and physical meaning of the measurement data.

[0117] In one possible implementation, an image scale factor is obtained, and the actual displacement is obtained based on the target point coordinates, the measurement point coordinates, and the image scale factor.

[0118] Specifically, to obtain the actual structural displacement at this point, its position relative to the reference position in the initial frame can be calculated. Differences between them:

[0119] ;

[0120] in, Indicates the difference value. The coordinates of the target point, The coordinates of the measurement points are used. An image scale factor is introduced. (Unit: mm / pixel) Convert pixel displacement to real-world displacement:

[0121] ;

[0122] in, This represents the actual displacement.

[0123] This application embodiment has an abnormal region handling and redundancy mechanism. If there are abnormal situations such as insufficient number of reference points or insufficient accuracy of thermal drift modeling in a certain sub-region, the system will trigger an adaptive strategy, such as: interpolation filling, estimating the transformation of the current region through weighted interpolation of the models of adjacent sub-regions; region merging, merging with the surrounding regions to fit the joint homography transformation; residual detection, judging whether there are unreasonable jumps in the position change of the compensated target point, and removing the measurement values ​​of the region if necessary.

[0124] This module can perform region-by-region, point-by-point displacement correction on all target points in the image, compensating not only for overall thermal drift but also adapting to the non-uniformity and local features of thermal drift within the image space, significantly improving the spatial accuracy and robustness of the system measurement. It is particularly suitable for long-cycle visual measurement tasks in scenarios with high requirements such as drastic temperature changes, large field of view, and high measurement accuracy.

[0125] It should be noted that, compared with the temperature compensation methods commonly used in existing vision measurement systems, this application has significant innovations in structural design, thermal drift modeling, correction strategy, and adaptability. The key differences and the resulting technological breakthroughs are as follows:

[0126] First, it possesses excellent system versatility and adaptability. The key difference lies in the fact that traditional thermal drift compensation techniques typically rely on specific camera models, lens structures, or the stability of temperature changes, requiring modeling using camera intrinsic parameters or external temperature sensors, resulting in poor adaptability in complex environments. In contrast, the measurement-calibration common-field visual measurement method proposed in this invention introduces a fixed reference plane within the same imaging field of view and combines it with local region segmentation and modeling strategies for the image plane. It directly corrects thermal drift based on geometric changes between images, without relying on specific camera structures or temperature change patterns. This provides excellent adaptability to different camera models, lenses, and non-constant temperature fields, significantly improving the system's versatility and practicality. Technical benefits include: insensitivity to camera model, lens parameters, and structural design, exhibiting excellent versatility; applicability to different types of industrial camera and lens combinations; and guaranteed reliability and accuracy even in complex scenarios with large dynamic temperature changes and harsh environments.

[0127] Second, the thermal drift modeling method based on local image blocks differs from traditional methods. Traditional thermal drift correction methods often rely on global rigid body models or simple function fitting between temperature and parameters. When faced with the prevalent non-uniform thermal distribution in actual imaging, these methods often fail to effectively describe local image deformation, leading to insufficient correction accuracy. This application innovatively proposes a method combining image plane segmentation and local homography transformation modeling. By dividing the image into multiple sub-regions and modeling their respective thermal deformation characteristics, it achieves a fine depiction of spatially heterogeneous thermal drift. Technical effects: This method can accurately model the local non-uniformity of thermal drift on the image plane, significantly improving the spatial resolution and compensation accuracy of thermal drift correction compared to traditional methods. It is particularly suitable for long-term visual monitoring systems operating under conditions such as multi-source thermal interference and complex thermal field distributions.

[0128] Third: All-weather environmental adaptability design. A key difference: Traditional visual temperature compensation systems are typically sensitive to external lighting conditions, easily affected by strong light, backlight, and low nighttime illumination, leading to unstable image features and drift compensation failure. This application effectively overcomes interference from changes in ambient light by introducing an all-weather observation optical design. The entire imaging system is placed in a closed optical dark box, equipped with a highly uniform LED array supplementary light source, and an infrared bandpass filter is introduced into the measurement marker path, allowing only specific wavelengths of light (such as 850nm) to image, thereby achieving high-contrast, interference-resistant marker extraction. Technical effects: Stable observation day and night, without relying on natural lighting conditions; suppression of strong external light or reflection interference, improving the robustness of image feature point extraction; long-term operation in low-light environments such as tunnels, under bridges, and at night, ensuring measurement continuity and stability.

[0129] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0130] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0131] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or N executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0132] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable storage medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable storage medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable storage medium could be paper or other suitable media on which the program can be printed, since the program can be obtained electronically by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0133] It should be understood that the various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0134] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0135] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0136] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.

[0137] It should be understood that the application of this application is not limited to the examples above. Those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

[0138] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

Claims

1. A universal thermal drift suppression method for visual displacement measurement, characterized in that, The general thermal drift suppression method for visual displacement measurement includes: Obtain reference information and current information of the joint imaging of the reference calibration plate and the target object; Based on the reference information and the current information, a thermal drift correction constraint model is obtained; Based on the current information and the thermal drift correction constraint model, the coordinate information of the target object is obtained; The thermal drift correction constraint model includes multiple geometric transformation models; The process of obtaining the thermal drift correction constraint model based on the reference information and the current information is as follows: For multiple reference local network sub-regions and corresponding multiple current local network sub-regions, multiple geometric transformation models are obtained based on multiple reference corner coordinate sets and multiple current corner coordinate sets; When the geometric transformation model is a local homography matrix, the local homography matrix is ​​represented as: ； in, Indicates the first The local homography matrix of a current local network sub-region This represents the horizontal scaling factor. Indicates horizontal shear strength. Indicates the horizontal translation amount. Indicates vertical shear strength. This represents the scaling factor in the vertical direction. This indicates the vertical translation amount. Indicates the perspective distortion coefficient in the horizontal direction. This represents the perspective distortion coefficient in the vertical direction. Represents the normalization factor. Represents the set of real numbers; When the geometric transformation model is a local affine transformation matrix, the local affine transformation matrix is ​​expressed as: ； in, Indicates the first i A local unit matrix of a current local website sub-region. These represent the scaling factors in the horizontal and vertical directions, respectively. and These represent the shear coefficients in the horizontal and vertical directions, respectively. and These represent the translation amounts in the horizontal and vertical directions, respectively. When the geometric transformation model is a high-order nonlinear distortion model, the modeling method for the high-order nonlinear distortion model is as follows: ； in, The x-coordinate of the pixel in the reference frame, The vertical coordinate of the pixel in the reference frame. The x-coordinate after drifting in the current frame. The vertical coordinate after drifting in the current frame. The polynomial coefficients are used to map the x-coordinate of a pixel in the reference frame to the shifted x-coordinate in the current frame. The polynomial coefficients are the coordinates of a pixel in the reference frame mapped to its shifted coordinate in the current frame, where N is the polynomial order; or The modeling method for the high-order nonlinear distortion model is as follows: ； ； in, , Indicates the radial distance to the center of the image; , , Radial distortion coefficient; , The tangential distortion coefficient; The coordinate information includes the coordinates of multiple target points; The step of obtaining the coordinate information of the target object based on the current information and the thermal drift correction constraint model specifically involves: Based on the coordinates of all the measured points and the corresponding geometric transformation models, the coordinates of multiple target points are obtained; The coordinates of the target point are represented as follows: ； in, This indicates the target point coordinates after the measurement point coordinates have been corrected. Indicates the first Geometric transformation model of a current local network sub-region The inverse mapping, the geometric transformation model Local homography matrix Local affine matrix Or a higher-order nonlinear distortion model, This indicates the coordinates of the measurement point in the current frame image.

2. The universal thermal drift suppression method for visual displacement measurement according to claim 1, characterized in that, The reference calibration plate has multiple calibration points, the target object has measurement points, the reference information includes reference data of multiple calibration points under the correction optical path, and the current information includes current data of multiple calibration points under the correction optical path and data to be corrected of measurement points under the measurement optical path. The acquisition of reference information and current information of the joint imaging of the reference calibration plate and the target object specifically includes: Acquire a reference frame image of the reference calibration plate and the target object at the initial temperature, and obtain reference data of multiple calibration points under the corrected optical path based on the reference frame image; Acquire the current frame image of the reference calibration plate and the target object at the current temperature, and obtain the current data of multiple calibration points under the correction optical path and the data to be corrected of the measurement points under the measurement optical path based on the reference frame image.

3. The general thermal drift suppression method for visual displacement measurement according to claim 2, characterized in that, The acquisition of the reference frame image jointly imaged by the reference calibration plate and the target object at the initial temperature specifically includes: Acquire an initial image of the reference calibration plate and the target object at the initial temperature; If the multiple calibration points in the initial image do not overlap with the measurement points, then the initial image is used as a reference frame image; If the calibration point overlaps with the measurement point in the initial image, then after moving the visual measurement device, an updated image is acquired that is jointly imaged by the reference calibration plate and the target object, until multiple calibration points in the updated image do not overlap with the measurement point, and the updated image is used as a reference frame image.

4. The universal thermal drift suppression method for visual displacement measurement according to claim 2, characterized in that, The reference data includes a reference corner point coordinate set, the current data includes a current corner point coordinate set, and the data to be corrected includes the coordinates of the measurement point. The step of obtaining reference data for multiple calibration points under the corrected optical path based on the reference frame image specifically includes: The reference frame image is divided into regions to obtain multiple reference local network sub-regions; Feature points are extracted from each of the reference local network sub-regions to obtain the reference corner coordinate set corresponding to each of the reference local network sub-regions; The step of obtaining the current data of multiple calibration points under the corrected optical path and the data to be corrected of the measurement points under the measurement optical path based on the reference frame image specifically includes: The current frame image is divided into regions to obtain multiple current local network sub-regions, wherein the current local network sub-regions correspond to the reference local network sub-regions; Feature points are extracted for each current local network sub-region to obtain the current corner coordinate set and measurement point coordinates corresponding to each current local network sub-region.

5. The universal thermal drift suppression method for visual displacement measurement according to claim 1, characterized in that, The step of obtaining the coordinate information of the target object based on the current information and the thermal drift correction constraint model further includes: Obtain the image scale factor, and based on the target point coordinates, the measurement point coordinates, and the image scale factor, obtain the true displacement.

6. A visual measurement apparatus for implementing the universal thermal drift suppression method for visual displacement measurement according to any one of claims 1 to 5, characterized in that, The visual measurement device includes a processor, a camera, a beam splitter, a collimating lens, and a reference calibration plate. The processor is communicatively connected to the camera. The camera, the beam splitter, the collimating lens, and the reference calibration plate are arranged sequentially. The two sides of the beam splitter face the reference calibration plate and the target structure, respectively. The reference calibration plate has multiple calibration points arranged in an array, and the target structure has measurement points. The camera is used to acquire reference information and current information of the joint imaging of the reference calibration plate and the target object, and send them to the processor; The processor obtains a thermal drift correction constraint model based on the reference information and the current information, and obtains the coordinate information of the target object based on the current information and the thermal drift correction constraint model.

7. The visual measurement device according to claim 6, characterized in that, The beam splitter has an infrared bandpass filter on the side facing the target structure. The camera, the beam splitter, the infrared bandpass filter, the collimating lens, and the reference calibration plate are arranged inside the optical dark box, and the target structure is located outside the optical dark box.

Images