Temperature compensation methods, equipment, storage media, and software products for image measurement equipment

By constructing a global affine incremental temperature model through the instantaneous calibration method of thermal reference, the problem of thermal expansion and thermal drift error of image measurement equipment in non-constant temperature environment is solved, and the accurate temperature compensation effect is achieved, which is suitable for measurement scenarios that have not reached thermal stability.

CN123123694APending Publication Date: 2026-11-17NANJING MUMUSILI TECH CO LTD +2
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611097365.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-07-23
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing image measurement equipment has difficulty in effectively compensating for errors caused by thermal expansion and thermal drift in non-constant temperature environments. In particular, when the equipment has not reached a thermally stable state, the temperature sensor readings are difficult to accurately reflect the thermal deformation state of the entire machine.

Method used

A temperature compensation method based on instantaneous calibration of the thermal reference is adopted. By collecting the residual of the thermal reference mark at a non-base temperature, a global affine incremental temperature model is constructed. The stability of the thermal reference is judged in multiple rounds of observation, the effective temperature offset is optimized in reverse, and a global affine correction model is fitted to achieve instantaneous calibration.

Benefits of technology

In environments where thermal stability has not been achieved, accurate temperature compensation is realized, reducing the impact of inconsistencies between local temperature sensor readings and the actual thermal state of the entire device. It avoids miscompensation of local marker points and forms a closed loop between the temperature compensation result and the actual measurement state of the equipment, making it suitable for measurement scenarios where thermal stability has not been achieved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure SMS_5
    Figure SMS_5
  • Figure QLYQS_1
    Figure QLYQS_1
  • Figure QLYQS_2
    Figure QLYQS_2
Patent Text Reader

Abstract

This invention discloses a temperature compensation method, device, storage medium, and program product for image measurement equipment. The invention first enables existing temperature compensation, then observes the baseline residual after temperature compensation using preset thermal reference markers on the device; next, it selects the effective temperature from candidate temperature offsets that best fits the current baseline residual; based on the effective temperature, it fits a global affine correction, and decides whether to enable it based on its reliability. This invention does not require the device to achieve complete thermal stability beforehand, nor does it rely entirely on the temperature sensor's installation location and single-point readings. It infers the current thermal state from actual visual measurement results, creating a closed-loop correlation between the compensation result and the current device state. Furthermore, this invention only corrects errors that can be explained by a low-order global model, avoiding the erroneous expansion of a small number of reference points into local compensation tables, thereby improving the controllability and traceability of the correction.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of image measurement, and specifically relates to a temperature compensation technology for image testing equipment. Background Technology

[0002] Image measurement equipment is typically used for measuring micrometer-level dimensions, and its accuracy is easily affected by the thermal deformation of materials. Using the linear expansion coefficient of common steel, approximately 11.5 × 10⁻⁶ / ℃, it can be estimated that if the effective length of the moving platform or structural component is 1 m, the theoretical length change is approximately 11.5 μm for every 1℃ change in temperature, which can usually be estimated in the order of approximately 12 μm.

[0003] Existing image measurement equipment typically requires users to operate at a constant temperature, such as 20°C or 23°C. When the equipment is not used at the standard temperature, or when the ambient temperature fluctuates significantly, the basic calibration table cannot fully compensate for errors caused by thermal expansion and thermal drift. Some equipment adds temperature sensors and uses temperature compensation modules for compensation. This method is usually a positive compensation: first, the temperature at one or more locations is measured, and then the motion coordinate correction is calculated based on the temperature compensation model. However, material thermal deformation is a process of transmission, accumulation, and distribution over time. The temperature sensor measures a local temperature, which may not be equivalent to the actual thermal deformation state of the entire device at present.

[0004] For example, when the equipment is first started, the light source and motor begin to work, the machine body experiences localized temperature rise, the ambient temperature fluctuates continuously, or the on-site temperature cannot be kept constant, the equipment may remain in a state of thermal instability or difficulty in achieving thermal stability for an extended period. In such cases, even if the temperature sensor readings are accurate, these readings may not fully reflect the thermal state formed by the guide rails, worktable, support structure, and camera system together, resulting in global proportional, translation, rotation, shearing, or orthogonality errors even after temperature compensation. Summary of the Invention

[0005] To address the technical problems mentioned in the background section, this invention proposes a temperature compensation method, device, storage medium, and program product for image measurement equipment.

[0006] To achieve the above-mentioned technical objectives, the technical solution of the present invention is as follows:

[0007] A temperature compensation method for image measurement equipment based on instantaneous calibration using a thermal reference includes the following steps:

[0008] (1) Collect the target coordinates and real motion coordinates of multiple grid points on the calibration plate at the base temperature, fit the global affine parameters of the base, calculate the local residuals of the base, and construct the base compensation model;

[0009] (2) Collect temperature sampling points under non-base temperature, calculate the difference between the real motion coordinates and the base compensation result under non-base temperature, and use it as temperature error. Based on the temperature error, fit the global affine increment of temperature and construct a temperature compensation model containing the global affine increment of temperature.

[0010] (3) Set N hot reference marks on the workbench or measurement area, read the nominal coordinates and current temperature t0 of the N hot reference marks, start basic compensation and temperature compensation, and obtain the initial true motion coordinates of each hot reference mark, N≥3;

[0011] (4) Control the motion platform to move near each hot reference mark, use the camera to collect images and identify the mark center, and calculate the reference residual of a single observation for each hot reference mark; perform multiple rounds of observation for each hot reference mark, each round of observation includes a single observation by controlling the motion platform to approach the hot reference mark from different directions, average the reference residuals obtained from all single observations in each round of observation to obtain the comprehensive reference residual of the round of observation, compare the change in the comprehensive reference residual of the current round and the previous round of observation for the same hot reference mark, and take the maximum value of the change in the comprehensive reference residual of the current round and the previous round of observation for N hot reference marks as the repeated observation drift of the current round of observation. If the repeated observation drift of the current round of observation exceeds the preset threshold T1, then proceed to the next round of observation for each hot reference mark; otherwise, proceed to step (5).

[0012] (5) Replace the current temperature t0 with each candidate temperature t in the preset candidate temperature set, calculate the candidate temperature compensation change corresponding to each thermal reference mark, and select the candidate temperature that minimizes the difference between the comprehensive reference residual in the last round of observation of each thermal reference mark in step (4) and the candidate temperature compensation change corresponding to the thermal reference mark, as the effective temperature t*. Then deduct the temperature compensation change caused by the effective temperature t* from the comprehensive reference residual in the last round of observation of N thermal reference marks in step (4) to obtain the temperature compensation residual corresponding to each thermal reference mark.

[0013] (6) Fit the residual after temperature compensation obtained in step (5) into a global affine correction model. Based on the low-order fitting residual, the magnitude of the global correction and the repeated observation drift in the last round of observation in step (4), determine whether to use the global affine correction model for instantaneous calibration.

[0014] Furthermore, in step (4), the calculation method for the reference residual of a single observation of each thermal reference mark is as follows:

[0015]

[0016]

[0017] In the above formula, d_img,i,k represents the image deviation of the kth observation of the i-th reference point, c_i,k represents the marker center obtained by image recognition of the kth observation of the i-th reference point, c_0 represents the camera field of view center, d_i,k represents the reference residual of the kth observation of the i-th reference point, M_pix2stage represents the conversion matrix from pixel to platform coordinates, β_k represents the directional deviation corresponding to the kth approach direction, and i∈[1,N].

[0018] Furthermore, in step (4), the averaging process is a simple average, a weighted average, or a robust average.

[0019] Further, in step (6), the global affine correction model is subjected to coordinate normalization.

[0020] Further, in step (6), the low-order fitting residual, the magnitude of the global correction, and the repeated observation drift of the last round of observation in step (4) are weighted and summed to obtain a confidence score. If the confidence score does not exceed the preset threshold T2, the global affine correction model is used for instantaneous calibration.

[0021] Secondly, the present invention also provides a computer device, including a processor and a memory, wherein the memory stores a computer program, and the processor executes the computer program to implement the above-described temperature compensation method for image measurement equipment based on instantaneous calibration of thermal reference.

[0022] Furthermore, the present invention also provides a computer storage medium storing a computer program, which, when executed on a processor, implements the above-described temperature compensation method for image measurement equipment based on instantaneous calibration of thermal reference.

[0023] Finally, the present invention also provides a computer program product, including a computer program or instructions, which, when executed by a processor, implement the above-mentioned temperature compensation method for image measurement equipment based on instantaneous calibration of thermal reference.

[0024] The beneficial effects of adopting the above technical solution are as follows:

[0025] 1. This invention is applicable to measurement scenarios where thermal stability has not been achieved or cannot be achieved on-site, and is not limited to the time when the equipment is turned on.

[0026] 2. This invention optimizes the effective temperature offset by reverse optimization of the thermal reference residual, thereby reducing the impact of inconsistency between the local readings of the temperature sensor and the actual thermal state of the whole machine.

[0027] 3. This invention fits a global affine correction based on the effective temperature, so that the temperature compensation result forms a closed loop with the actual measurement state of the current equipment.

[0028] 4. The instantaneous calibration of the hot reference designed in this invention only corrects the global influence terms, avoiding the misuse of a small number of marker points for local compensation.

[0029] 5. This invention can determine the reliability of the thermal reference itself through multiple data acquisitions and constraint optimizations, thus avoiding the incorrect correction of temperature compensation using deformed references.

[0030] 6. This invention does not contaminate the basic local residual table and the temperature local residual table, making it easy to restore, trace, and recalibrate. Detailed Implementation

[0031] The overall approach of this embodiment is as follows: first, enable the existing temperature compensation, then use the preset thermal reference mark on the device to observe the reference residual after temperature compensation; then select the effective temperature that best fits the current reference residual from the candidate temperature offsets; based on the effective temperature, fit the global affine correction, and decide whether to enable it based on the confidence level.

[0032] 1. Basic Compensation Modeling

[0033] At the base temperature, the target coordinates and actual motion coordinates of multiple grid points on the calibration plate are collected, the base global affine model is fitted, and the base local residuals are calculated.

[0034]

[0035] In the above formula, P_i represents the nominal coordinates of the i-th calibration point; P_real,i represents the real motion coordinates corresponding to the i-th calibration point; A_base(P_i) represents the basic global affine compensation result of the point; L_base(P_i) represents the basic local residual after deducting the global affine term.

[0036] The basic global affine parameters and the basic local residual table are stored separately, and together they constitute the basic compensation during subsequent runtime.

[0037] 2. Temperature compensation modeling

[0038] Temperature sampling points are collected at non-base temperatures, and the difference between the actual motion coordinates and the base compensation results at that temperature is calculated as the temperature error.

[0039]

[0040] In the above formula, ΔT_err(P_i,T) represents the temperature error of the i-th sampling point relative to the base compensation when the temperature is T; P_real,i(T) represents the actual motion coordinates measured when the temperature is T; and C_base(P_i) represents the base compensation result.

[0041] The temperature error is fitted with a global affine increment. After subtracting the global affine increment, an optional local temperature residual increment is obtained. If multi-temperature experiments cannot prove that the local residual changes stably with temperature, the local temperature residual increment is not used, and temperature compensation only uses the global affine increment.

[0042] 3. Activate basic compensation and determine the initial position.

[0043] This embodiment sets up four thermal reference markers. Before performing instantaneous calibration, the system first enables basic compensation and current temperature compensation based on the current temperature t0 to obtain the initial true motion coordinates that the motion platform should reach.

[0044]

[0045] In the above formula, M_0(P_ref,i,t0) represents the initial motion coordinates of the i-th thermal reference mark calculated at the current temperature t0; P_ref,i represents the nominal coordinates of the i-th thermal reference mark; A_base, L_base, A_temp, and L_temp represent the basic global term, the basic local term, the temperature global term, and the temperature local term, respectively.

[0046] 4. Take images from multiple directions and calculate the residual of a single observation.

[0047] The motion platform is moved to the vicinity of each hot reference marker, and images are acquired using a camera to identify the marker center. To reduce the backlash introduced by unidirectional approach, the same marker can be approached from multiple directions.

[0048]

[0049]

[0050] In the above formula, d_img,i,k represents the image deviation of the k-th observation at the i-th reference point; c_i,k represents the marker center obtained by image recognition from the k-th observation at the i-th reference point; c_0 represents the camera field of view center; d_i,k represents the single reference residual of the k-th observation at the i-th reference point; M_pix2stage represents the pixel-to-platform coordinate conversion matrix; β_k represents the directional deviation corresponding to the k-th approach direction. If the device hysteresis is very small, β_k can be set to zero.

[0051] 5. Integrate multiple rounds of observations and determine the stability of the benchmark.

[0052] Data is collected in multiple rounds. In each round, four thermal reference markers are observed multiple times. For the i-th thermal reference marker, K single-point reference residuals d_i,k(m) are obtained in the m-th round, where k=1..K. The system averages, weights, or robustly averages the K single-point reference residuals of this round to obtain the comprehensive reference residual r_i(m) of the m-th round. After the m-th round of collection is completed, the comprehensive reference residual r_i(m) of each thermal reference marker is compared with the corresponding result r_i(m-1) of the previous round. The change between adjacent rounds is calculated, and the value with the largest change among the four thermal reference markers is taken as the repeated observation drift ρ of this round. If ρ exceeds a preset threshold, it is considered that the current thermal state of the equipment is still changing rapidly, or the thermal reference observation results are unstable. In this case, instantaneous calibration correction is not activated, and the next round of observation begins.

[0053]

[0054]

[0055] In the above formula, r_i (m) represents the comprehensive baseline residual of the i-th reference point obtained in the m-th round of acquisition; RobustMean_k represents the robust averaging over different number of observations or similar directions; ρ represents the drift of repeated observations; m represents the current acquisition round; m-1 represents the previous acquisition round; Δt represents the time interval between two acquisition rounds.

[0056] Multiple rounds of data acquisition are used to determine whether the thermal baseline residuals continue to change over time. A single round of acquisition only reflects the observation results at a certain moment and cannot determine whether the current thermal state of the equipment is still drifting. By comparing the combined baseline residuals of two adjacent rounds, the amount of repeated observation drift ρ can be obtained, thereby determining whether the equipment has entered a relatively stable state. If the changes in the results of multiple rounds of acquisition are small, it indicates that the current thermal baseline residuals are repeatable and can be used to back-calculate the effective temperature shift and fit global affine corrections. If the changes in the results of multiple rounds of acquisition are consistently large, it indicates that the thermal state of the equipment has not yet stabilized, or that the baseline observation results are unreliable, and correction should be delayed or re-acquisition should be performed.

[0057] 6. Back-calculate the effective temperature offset

[0058] For multiple candidate temperatures t, calculate the additional compensation change that would occur at the thermal reference mark position after changing the current temperature input t0 to a candidate temperature t, and compare it with the current observation residual. Select the candidate temperature t* that minimizes the difference between the comprehensive reference residual in the last round of observations for each thermal reference mark in the previous step and the compensation change of the candidate temperature corresponding to that thermal reference mark.

[0059]

[0060]

[0061]

[0062] In the above formula, H(P,t) represents the temperature compensation model when the temperature input is t; q_i(t;T0) represents the compensation change at the i-th thermal reference mark after changing the temperature input from t0 to t; t* represents the effective temperature.

[0063] 7. Recalculate the temperature compensation using the effective temperature and obtain the remaining residual.

[0064] The temperature compensation change from t0 to t* is recalculated using the effective temperature t*, and this change is subtracted from the observation residuals to obtain the remaining residuals that need to be explained by the instantaneous calibration.

[0065]

[0066]

[0067] In the above formula, q_i(t*;t0) represents the temperature compensation change of the effective temperature t* relative to the current temperature input t0; e_i represents the remaining residual after deducting the temperature compensation change, which is used for subsequent global affine correction fitting.

[0068] 8. Global affine correction during fitting runtime

[0069] The remaining residuals are fitted to a global affine correction model. This model only expresses the global low-order error and does not express the local complex error.

[0070]

[0071]

[0072] In the above formula, b represents the global translation term; A represents the global affine matrix; ε_i represents the residuals not explained by the global affine model; and G(P) represents the global correction of the thermal reference at any target point P.

[0073] In practice, the coordinates can be normalized to reduce the risk of ill-conditioned matrices. Let u_i=(x_i-x_c) / L_x, v_i=(y_i-y_c) / L_y, then the solution can be obtained by linear least squares as follows.

[0074]

[0075]

[0076]

[0077] In the above formula, G_x and G_y represent the global corrections in the X and Y directions, respectively; u_i and v_i represent the normalized coordinates; x_c and y_c represent the normalization centers; L_x and L_y represent the normalization scales; a_0, a_1, a_2, b_0, b_1, and b_2 are the affine parameters to be determined; θ represents the parameter vector; and λ represents the regularization weights.

[0078] Four hot reference markers provide eight two-dimensional residual equations, sufficient to solve for six affine parameters.

[0079] 9. Credibility assessment and modification of activation conditions

[0080] The confidence score is calculated based on the low-order fitting residuals, the magnitude of the global correction, and the drift rate of multiple observations.

[0081]

[0082]

[0083]

[0084] In the above formula, Score represents the confidence score; FitRes represents the low-order fitting residual; CorrectionSize represents the magnitude of the global correction; ρ represents the repeated observation drift; w1, w2, and w3 represent the weights of each evaluation item, and w1+w2+w3=1.

[0085] If the confidence score is within a preset threshold, global correction to the thermal baseline is enabled. Otherwise, instantaneous calibration correction is not enabled, and a wait, downgrade, or rescale strategy is output.

[0086] This embodiment also relates to a computer device, which includes a processor and a memory. The memory stores a computer program, and the processor executes the computer program to implement the above-described contents. This embodiment also relates to a computer storage medium storing a computer program, which, when executed on a processor, implements the above-described contents. This embodiment also relates to a computer program product, including a computer program or instructions, which, when executed by a processor, implements the above-described contents.

[0087] The embodiments are merely illustrative of the technical concept of the present invention and should not be construed as limiting the scope of protection of the present invention. Any modifications made to the technical solution based on the technical concept proposed in this invention shall fall within the scope of protection of this invention.

Claims

1. A temperature compensation method for image measurement equipment based on instantaneous calibration using a thermal reference, characterized in that, Includes the following steps: (1) Collect the target coordinates and real motion coordinates of multiple grid points on the calibration plate at the base temperature, fit the global affine parameters of the base, calculate the local residuals of the base, and construct the base compensation model; (2) Collect temperature sampling points under non-base temperature, calculate the difference between the real motion coordinates and the base compensation result under non-base temperature, and use it as temperature error. Based on the temperature error, fit the global affine increment of temperature and construct a temperature compensation model containing the global affine increment of temperature. (3) Set N hot reference marks on the workbench or measurement area, read the nominal coordinates and current temperature t0 of the N hot reference marks, start basic compensation and temperature compensation, and obtain the initial true motion coordinates of each hot reference mark, N≥3; (4) Control the motion platform to move near each hot reference mark, use the camera to collect images and identify the mark center, and calculate the reference residual of a single observation for each hot reference mark; perform multiple rounds of observation for each hot reference mark, each round of observation includes a single observation by controlling the motion platform to approach the hot reference mark from different directions, average the reference residuals obtained from all single observations in each round of observation to obtain the comprehensive reference residual of the round of observation, compare the change in the comprehensive reference residual of the current round and the previous round of observation for the same hot reference mark, and take the maximum value of the change in the comprehensive reference residual of the current round and the previous round of observation for N hot reference marks as the repeated observation drift of the current round of observation. If the repeated observation drift of the current round of observation exceeds the preset threshold T1, then proceed to the next round of observation for each hot reference mark; otherwise, proceed to step (5). (5) Replace the current temperature t0 with each candidate temperature t in the preset candidate temperature set, calculate the candidate temperature compensation change corresponding to each thermal reference mark, and select the candidate temperature that minimizes the difference between the comprehensive reference residual in the last round of observation of each thermal reference mark in step (4) and the candidate temperature compensation change corresponding to the thermal reference mark, as the effective temperature t*. Then deduct the temperature compensation change caused by the effective temperature t* from the comprehensive reference residual in the last round of observation of N thermal reference marks in step (4) to obtain the temperature compensation residual corresponding to each thermal reference mark. (6) Fit the residual after temperature compensation obtained in step (5) into a global affine correction model. Based on the low-order fitting residual, the magnitude of the global correction and the repeated observation drift in the last round of observation in step (4), determine whether to use the global affine correction model for instantaneous calibration.

2. The temperature compensation method for image measurement equipment based on instantaneous calibration of a thermal reference according to claim 1, characterized in that, In step (4), the calculation method for the reference residual of a single observation of each thermal reference mark is as follows: , , In the above formula, This represents the image deviation of the k-th observation at the i-th reference point. This represents the marker center obtained from the k-th observation image of the i-th reference point. Indicates the center of the camera's field of view. This represents the baseline residual of the k-th observation at the i-th baseline point. This represents the conversion matrix from pixel to platform coordinates. Let i represent the directional deviation corresponding to the k-th approach direction, i∈[1,N].

3. The temperature compensation method for image measurement equipment based on instantaneous calibration of a thermal reference according to claim 1, characterized in that, In step (4), the averaging process is a simple average, a weighted average, or a robust average.

4. The temperature compensation method for image measurement equipment based on instantaneous calibration of a thermal reference according to claim 1, characterized in that, In step (6), the global affine correction model is subjected to coordinate normalization.

5. The temperature compensation method for image measurement equipment based on instantaneous calibration of a thermal reference according to claim 1, characterized in that, In step (6), the low-order fitting residual, the global correction amount and the repeated observation drift amount of the last round of observation in step (4) are weighted and summed to obtain a confidence score. If the confidence score does not exceed the preset threshold T2, the global affine correction model is used for instantaneous calibration.

6. A computer device comprising a processor and a memory, wherein the memory stores a computer program, characterized in that: The processor is used to execute the computer program to implement the image measurement device temperature compensation method based on thermal reference instantaneous calibration as described in any one of claims 1-5.

7. A computer storage medium storing a computer program, characterized in that: When the computer program is executed on the processor, it implements the image measurement equipment temperature compensation method based on thermal reference instantaneous calibration as described in any one of claims 1-5.

8. A computer program product, comprising a computer program or instructions, characterized in that: When the computer program or instructions are executed by the processor, they implement the image measurement equipment temperature compensation method based on thermal reference instantaneous calibration as described in any one of claims 1-5.