Calibration and error correction method and system based on stereoscopic vision measurement system

Abstract

The invention discloses a calibration and error correction method and system based on a stereoscopic vision measurement system, and relates to the technical field of measurement system optimization, and the method comprises the steps: carrying out the scanning of an object with a standard size, and collecting data points; calculating an error between a measured length and an actual length for each data point, calculating an average value and a standard deviation of all measured values, and excluding the measured values exceeding the errors; and calculating a calibration coefficient and offset to obtain corrected side length data, and dynamically adjusting the calibration coefficient and the offset. Through detailed algorithm steps and a closed-loop feedback control mechanism, the machine measurement system can realize automatic data collection, real-time error analysis and dynamic adjustment of calibration parameters. Therefore, the accuracy and the stability of measurement are improved, and the high efficiency and the reliability of the system in long-term operation are ensured. Through regular calibration and real-time error correction, the algorithms can significantly improve the overall performance of the measurement system.

Description

Technical Field

[0001] The present invention relates to the technical field of measurement system optimization, and in particular to a method and system for calibrating and correcting errors of a measurement system based on stereo vision. Background Art

[0002] In the field of high-precision measurement, it is crucial to obtain accurate and reliable measurement data. Especially in machine vision systems, the accuracy of data directly affects the effect of subsequent processing and application. Traditional calibration methods are often unable to adapt to complex and changeable environments and measurement conditions, resulting in inaccurate measurement results. The present invention provides a method and system for calibrating and correcting a stereoscopic vision measurement system, which can improve the efficient calibration and error correction rate of the measurement system. Summary of the invention

[0003] In view of the above-mentioned problems, the present invention is proposed.

[0004] Therefore, the technical problem solved by the present invention is: how to ensure efficient calibration of the measurement system, improve the error correction rate, and enhance the measurement precision and accuracy.

[0005] To solve the above technical problems, the present invention provides the following technical solutions: a method for calibration and error correction of a stereoscopic vision measurement system, comprising: scanning an object of standard size to collect data points; calculating the error between the measured length and the actual length for each data point, calculating the average and standard deviation of all measured values, and excluding those measured values ​​with errors exceeding; calculating the calibration coefficient and offset to obtain corrected side length data, and dynamically adjusting the calibration coefficient and offset.

[0006] As a preferred solution of the method for calibrating and correcting the error of the stereoscopic vision measurement system described in the present invention, the acquisition includes selecting three objects of the same shape with different sizes, performing a 360-degree rotation scan on each object, and sequentially acquiring multiple data points at the same time and angle intervals to obtain the side length data L ( m i) easured Recorded in the system database, where (i) is the measurement sequence number.

[0007] As a preferred solution of the method for calibrating and correcting the error of the stereoscopic vision measurement system described in the present invention, the exclusion of the measured value includes calculating the error in each measurement. L actual After the measurement is completed, the average error is calculated Standard deviation σ, when the calculated error ΔL (i) >2σ, exclude the measured value. The calibration parameters are updated using linear regression when a preset threshold is exceeded.

[0008] As a preferred solution of the method for calibration and error correction of a stereoscopic vision measurement system according to the present invention, the calculation of the calibration coefficient and the offset includes initializing the calibration coefficient a=1 and the offset b=0, and iteratively updating the calibration coefficient a and the offset b using linear regression until convergence or reaching a predetermined maximum number of iterations:

[0009]

[0010] Among them, j represents the measurement number, underseta, bargmin represents the optimization expression of parameters a and b that minimize the sum of squared errors of the objective function.

[0011] As a preferred solution of the method for calibrating and correcting the error of the stereoscopic vision measurement system described in the present invention, the updated calibration coefficient a and offset b are used to calculate the new side length data.

[0012]

[0013] Monitor the post-calibration error of consecutive measurements, if the new average error If the error still exceeds the threshold, the system will automatically adjust the scanning parameters of the laser scanner and rescan the calculation to obtain an error within the threshold.

[0014] As a preferred solution of the method for calibrating and correcting the error of the stereoscopic vision measurement system described in the present invention, the dynamic adjustment includes:

[0015] Initialize the calibration parameters a and offset, set the learning rate η and error threshold, and calculate the measurement value based on the measurement data With the actual value L actual The error between:

[0016] Update the parameters using online gradient descent:

[0017]

[0018] The specific partial derivative calculation is expressed as:

[0019]

[0020] Substituting the above partial derivatives into the update formula, we can get:

[0021] b′=b+2·η·ΔL (i) .

[0022] As a preferred solution of the stereoscopic vision measurement system calibration and error correction method described in the present invention, the dynamic adjustment also includes dynamically adjusting the error threshold according to the most recent error data, by calculating the mean and standard deviation of the past n errors, and setting a new error threshold = mean + k × standard deviation, where k is a coefficient.

[0023] In the second aspect, another object of the present invention is to provide a calibration and error correction system based on a stereoscopic vision measurement system, comprising: a scanning acquisition module, a data processing module and a dynamic adjustment module; the scanning acquisition module is used to scan and collect data points of objects of standard size; the data processing module is used to calculate the error, average value and standard deviation between the measured length and the actual length; the dynamic adjustment module is used to calculate the calibration coefficient and offset correction side length data, and dynamically adjust the calibration coefficient and offset in real time.

[0024] In a third aspect, a computer device includes a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above-mentioned method for calibrating and correcting an error based on a stereoscopic vision measurement system when executing the computer program.

[0025] In a fourth aspect, a computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the stereoscopic vision measurement system calibration and error correction method as described above.

[0026] Beneficial effects of the invention: The calibration and error correction method and system based on stereoscopic vision measurement system provided by the invention can realize automatic data collection, real-time error analysis and dynamic adjustment of calibration parameters through detailed algorithm steps and closed-loop feedback control mechanism. This not only improves the accuracy and stability of measurement, but also ensures the efficiency and reliability of the system in long-term operation. Through regular calibration and real-time error correction, these algorithms can significantly improve the overall performance of the measurement system;

[0027] Linear calibration algorithms are simple and efficient, and are suitable for scenarios where the errors are small and close to linear. Nonlinear error correction algorithms can handle more complex error situations, especially when the error pattern of the measurement system varies greatly with the measured value. Through regular calibration and real-time error correction, these algorithms can significantly improve the overall performance of the measurement system. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0029] Figure 1 An overall flow chart of a method for calibrating and correcting errors in a stereoscopic vision measurement system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0030] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0031] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0032] Example 1

[0033] Reference Figure 1 , is an embodiment of the present invention, and provides a method for calibrating and correcting errors based on a stereoscopic vision measurement system, comprising:

[0034] S1: Scan objects of standard size to collect data points;

[0035] Furthermore, select standard objects: select three standard objects of different sizes (small, medium, and large), such as cubes, whose side lengths are 10 cm, 20 cm, and 30 cm respectively. Perform a 360-degree rotation scan on each object, collect multiple data points at the same time and angle, and obtain side length data. Recorded in the system database, where (i) is the measurement sequence number. For example, a 360-degree rotation scan is performed on each object, and a data point is collected every 15 degrees, for a total of 24 data points.

[0036] S2: Calculate the error between the measured length and the actual length for each data point, calculate the mean and standard deviation of all measurements, and exclude those measurements with errors exceeding;

[0037] Furthermore, the error is calculated at each measurement. After the measurement is completed, the average error is calculated Standard deviation σ, when the calculated error ΔL (i) >2σ, exclude the measured value. The calibration parameters are updated using linear regression when a preset threshold is exceeded. is the calibrated value of the i-th measurement. is the value obtained from the actual measurement of the ith time. By calibrating the parameters a and b, the measured value is corrected to improve accuracy.

[0038] S3: Calculate the calibration coefficient and offset, obtain the corrected side length data, and dynamically adjust the calibration coefficient and offset.

[0039] Furthermore, the calibration coefficient a=1 and the offset b=0 are initialized, and the calibration coefficient a and the offset b are iteratively updated using linear regression until convergence or a predetermined maximum number of iterations is reached:

[0040]

[0041] Among them, j represents the measurement number, underseta, bargmin represents the optimization expression of parameters a and b that minimize the sum of squared errors of the objective function.

[0042] The updated calibration coefficient a and offset b are used to calculate the new side length data

[0043] Monitor the post-calibration error of consecutive measurements, if the new average error If the error still exceeds the threshold, the system will automatically adjust the scanning parameters of the laser scanner and rescan the calculation to obtain an error within the threshold.

[0044] Furthermore, the calibration coefficient and offset are adjusted dynamically, the calibration parameter a and the offset are initialized, the learning rate η and the error threshold are set, and the measurement value is calculated based on the measurement data. With the actual value L actual The error between:

[0045] Update the parameters using online gradient descent:

[0046]

[0047] The specific partial derivative calculation is expressed as:

[0048]

[0049] Substituting the above partial derivatives into the update formula, we can get:

[0050] b′=b+2·η·ΔL (i) .

[0051] The error threshold is adjusted dynamically according to the most recent error data. By calculating the mean and standard deviation of the past n errors, a new error threshold is set = mean + k × standard deviation, where k is a coefficient.

[0052] Furthermore, the measurement data is collected regularly and the error is calculated to analyze the measurement performance and error trend of the system. Error analysis can evaluate the effect of calibration parameters by calculating the average error. Performance criteria, such as error thresholds, are set. If the average error measured by the system exceeds a preset threshold (generally, for example, 0.05 cm), parameter adjustment is triggered.

[0053] Assume that the current calibration coefficients a = 1.01 and b = -0.05 cm, and the error threshold is set to 0.1 cm. Based on the data analysis of the past week, the online learning model recommends updating a = 1.02 and b = -0.03 cm, and adjusting the error threshold to 0.08 cm.

[0054] Example 2

[0055] As one embodiment of the present invention, a stereoscopic vision measurement system calibration and error correction system is provided, comprising:

[0056] Scanning and acquisition module, data processing module and dynamic adjustment module;

[0057] The scanning and acquisition module is used to scan and collect data points of objects of standard size;

[0058] The data processing module is used to calculate the error, average value and standard deviation between the measured length and the actual length;

[0059] The dynamic adjustment module is used to calculate the calibration coefficient and the offset correction side length data, and dynamically adjust the calibration coefficient and the offset in real time.

[0060] Example 3

[0061] As an embodiment of the present invention, a method for calibrating and correcting errors based on a stereoscopic vision measurement system is provided. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0062] Suppose a standard object (a cube with a side length of 20 cm) is measured and the series of measurements obtained are:

[0063] <![CDATA[Measured value (L measured )]]> <![CDATA[Actual value (L actual )]]> 19.8cm 20cm 20.1cm 20cm ...... 20cm 19.7cm 20cm

[0064] Error calculation: Calculate the error (ΔL) for each measurement. For example, for a measurement of 19.8 cm, (ΔL = 19.8 cm – 20 cm = -0.2 cm).

[0065] Outlier exclusion: Assuming the mean error is (-0.1 cm) and the standard deviation is (0.15 cm), exclude those measurements with an error greater than (0.3 cm) ((2×0.15 cm)).

[0066] Real-time calibration: When the mean error is found to be greater than 0.1 cm, the system updates the calibration parameters using the least squares method.

[0067] Closed-loop feedback: Apply new calibration parameters, the system automatically adjusts the scan configuration, and remeasures to verify the calibration effect.

[0068] In this way, the machine measurement system can achieve a high degree of automation and precision, effectively self-calibrating to ensure the accuracy of the measurement results and the long-term stability of the system. This closed-loop operation method significantly improves the efficiency and reliability of machine measurement, in stark contrast to traditional manual measurement and calibration methods.

[0069] Assuming we have a set of historical measurement data and actual dimension values, here is a code example to build a simple online linear regression model to update the calibration parameters in real time.

[0070] import numpy as np

[0071] from sklearn.linear_model import SGDRegressor

[0072] # Example data: measured and actual values

[0073] measured_values ​​= np.array([19.8,20.1,19.9,20.2,19.7])#measured values

[0074] actual_values ​​= np.array([20.0,20.0,20.0,20.0,20.0]) #actual value (standard size)

[0075] # Initialize the online learning model

[0076] model=SGDRegressor(learning_rate='constant',eta0=0.01)

[0077] #Simple online learning loop

[0078] for iin range(len(measured_values)):

[0079] # Extract a single data point

[0080] X=measured_values[i].reshape(-1,1)

[0081] y = actual_values[i]

[0082] #Update the model

[0083] model.partial_fit(X,[y])

[0084] #Get the updated coefficients and intercept

[0085] a = model.coef_[0]

[0086] b = model.intercept_[0]

[0087] #Print updated parameters

[0088] print(f"Updated calibration parameters: a={a:.2f},b={b:.2f}")

[0089] #Use the updated model for prediction and calibration

[0090] new_measurement = 19.9

[0091] corrected_value=model.predict([[new_measurement]])[0]

[0092] print(f"New measurement value: {new_measurement}, calibrated value: {corrected_value:.2f}")

[0093] This code first uses a simple linear model (gradient descent regression) to learn the calibration parameters, and then implements online learning through the partial_fit method to gradually update the model parameters. Every time new measurement data arrives, the model updates its parameters to reflect the latest data trends.

[0094] In practical applications, data may come from real-time measurement systems and need to be acquired and processed in real time. The choice of model and parameters (such as learning rate) need to be adjusted and optimized according to the specific circumstances of the actual application.

[0095] This approach ensures that the measurement system is not only calibrated based on current data, but also adapts to long-term environmental changes and system drift through continuous learning and optimization.

[0096] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for calibrating and correcting errors of a stereoscopic vision measurement system, characterized in that: include: Scan objects of standard size to collect data points; The error between the measured length and the actual length was calculated for each data point, and the mean and standard deviation of all measurements were calculated, excluding those measurements with errors exceeding; The calibration coefficient and the offset are calculated to obtain the corrected side length data, and the calibration coefficient and the offset are dynamically adjusted.

2. The method for calibrating and correcting errors based on a stereoscopic vision measurement system according to claim 1, characterized in that: The acquisition includes selecting three objects of the same shape with different sizes, performing a 360-degree rotation scan on each object, and sequentially collecting multiple data points at the same time and angle intervals to obtain side length data. Recorded in the system database, where (i) is the measurement sequence number.

3. The method for calibrating and correcting errors based on a stereoscopic vision measurement system according to claim 2, characterized in that: The exclusion of the measured values ​​includes the error calculated in each measurement After the measurement is completed, the average error is calculated Standard deviation σ, when the calculated error ΔL (i) >2σ, exclude the measured value. The calibration parameters are updated using linear regression when a preset threshold is exceeded.

4. The method for calibrating and correcting errors based on a stereoscopic vision measurement system according to claim 3, characterized in that: The calculation of the calibration coefficient and the offset includes initializing the calibration coefficient a=1 and the offset b=0, and iteratively updating the calibration coefficient a and the offset b using linear regression until convergence or a predetermined maximum number of iterations is reached: Among them, j represents the measurement number, underseta, bargmin represents the optimization expression of parameters a and b that minimize the sum of squared errors of the objective function.

5. The method for calibrating and correcting errors based on a stereoscopic vision measurement system according to claim 4, characterized in that: The updated calibration coefficient a and offset b are used to calculate the new side length data Monitor the post-calibration error of consecutive measurements, if the new average error If the error still exceeds the threshold, the system will automatically adjust the scanning parameters of the laser scanner and rescan the calculation to obtain an error within the threshold.

6. The method for calibrating and correcting errors based on a stereoscopic vision measurement system according to claim 5, characterized in that: The dynamic adjustment includes: Initialize the calibration parameters a and offset, set the learning rate η and error threshold, and calculate the measurement value based on the measurement data With the actual value L actual The error between: Update the parameters using online gradient descent: The specific partial derivative calculation is expressed as: Substituting the above partial derivatives into the update formula, we can get: ,b′=b+2·η·ΔL (i) 。 7. The method for calibrating and correcting errors based on a stereoscopic vision measurement system according to claim 6, characterized in that: The dynamic adjustment also includes dynamically adjusting the error threshold according to the most recent error data, by calculating the mean and standard deviation of the past n errors, and setting a new error threshold = mean + k × standard deviation, where k is a coefficient.

8. A system using the method for calibrating and correcting errors based on a stereoscopic vision measurement system according to any one of claims 1 to 7, characterized in that: include: Scanning and acquisition module, data processing module and dynamic adjustment module; The scanning and acquisition module is used to scan and acquire data points from objects of standard size; The data processing module is used to calculate the error, average value and standard deviation between the measured length and the actual length; The dynamic adjustment module is used to calculate the calibration coefficient and the offset correction side length data, and dynamically adjust the calibration coefficient and the offset in real time.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the stereo vision measurement system calibration and error correction method described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for calibrating and correcting errors of a stereoscopic vision measurement system according to any one of claims 1 to 7 are implemented.

Images