A calibration method and device for an industrial machine vision measurement system

Abstract

The present invention relates to the field of machine vision technology, and specifically to a calibration method and device for an industrial machine vision measurement system, which includes the following steps: using a glass linear scale and a circular target, collecting error data to construct a standardized calibration data set, obtaining a global error gradient field, a residual trend map, and a residual weight coefficient table, fusing a size compensation value and a directional correction value, constructing a calibration matrix, and obtaining a system calibration result. In the present invention, through the fusion modeling of multi-source error data, the recognition accuracy of the error source of the system is improved. Combining the construction of the gradient field and the extraction of the trend map, the directionality and spatial continuity of error compensation are enhanced. Using the weight coefficient distribution mechanism, differential control of the influence of dynamic and static errors is achieved. A calibration matrix with multi-dimensional parameter fusion is adopted to optimize the consistency and adaptability of the overall measurement accuracy, enabling the system to maintain stable error correction ability and calibration accuracy in a complex environment.

Description

Technical Field

[0001] The present invention relates to the technical field of machine vision, and particularly to a calibration method and device for an industrial machine vision measurement system. Background Art

[0002] The technical field of machine vision includes technologies such as image processing, pattern recognition, and automatic control, and is widely used in multiple industries such as electronics, logistics, and chemical industry for automated detection, measurement, and control. An industrial machine vision system consists of an industrial camera, a lens, a light source, a computer, and related software, focusing on simulating the visual function of humans, quickly extracting the features of an object, such as size, position, and angle, for accurate measurement and detection. The core lies in analyzing and calculating the collected images through image processing technology, judging the state or features of the target object, and providing a decision-making basis to improve production efficiency and product quality.

[0003] Among them, the calibration method for an industrial machine vision measurement system refers to effectively calibrating the measurement errors existing in the existing machine vision measurement system through various technical means to ensure the measurement accuracy of the system. The patent theme mainly involves methods for systematically calibrating an industrial machine vision system, solving the measurement error problems caused by factors such as environmental changes and equipment aging during the use of the existing vision system. The method implements regular or on-demand calibration according to specific equipment performance and working environment by designing special calibration devices and calibration procedures, ensuring the accuracy of the measurement results output by the system under various operating conditions, and improving the stability and reliability of the system.

[0004] Traditional calibration technologies for industrial machine vision measurement systems rely on error mean comparison and correction methods under a fixed path, lacking the identification of residual structures and directional errors, and lacking clear distinction and processing rules for various error sources during the calibration process, resulting in compensation offsets when the scene changes or the equipment state changes. Due to the relatively flat calibration data structure, sparse acquisition points, and the lack of a cross-scene comparison mechanism, it is difficult to identify changes in directional errors under dynamic interference. When there are conditions such as equipment vibration and light fluctuations, the stability of the measurement results decreases. When measuring in the edge area, for example, without introducing a direction correction factor, the errors often show aggregated offsets, affecting the overall calibration accuracy. The error weight processing does not form a differential judgment mechanism, and the weights of static and dynamic measurement points are the same in compensation processing, resulting in a fuzzy judgment of the system's sensitivity to different errors and reducing the pertinence and effectiveness of error correction. Summary of the Invention

[0005] In order to solve the technical problems existing in the prior art, embodiments of the present invention provide a calibration method and device for an industrial machine vision measurement system. The technical solutions are as follows:

[0006] To achieve the above object, the present invention adopts the following technical solutions. A calibration method for an industrial machine vision measurement system includes the following steps:

[0007] S1: Using a glass linear scale, collect multiple size error measurement values at multiple calibration positions and calculate the average value. Using a circular target image, synchronously collect the residual extreme difference values of multiple measurement points in a dynamic scene and the residual mean value in a static scene to obtain a standardized calibration data set;

[0008] S2: Call the standardized calibration data set, use the size error measurement value data set to calculate the average error value of each calibration position, analyze the gradient change and calculate the smoothing correction parameter, and use discrete points to construct a continuous error field of the entire workbench to generate a global error gradient field;

[0009] S3: Call the standardized calibration data set, use the dynamic residual extreme difference data set, with the center of the fitted circle as the origin, divide the residual distribution area into multiple sector units according to a fixed angle, calculate the sector residual extreme difference by aggregating the residual values within each sector, analyze the change trend of the radial distance of the measurement points, and generate a residual trend map;

[0010] S4: Call the standardized calibration data set, call the dynamic residual extreme difference data set and the static residual mean data set, calculate the weight coefficients of the dynamic residual and the static residual by analyzing the influence degree of the dynamic residual fluctuation data and the static residual stability on the system error, and generate a residual weight coefficient table.

[0011] As a further solution of the present invention, the global error gradient field specifically includes an error change direction layer, a smoothing correction parameter layer, and a path continuity fitting layer. The residual trend map specifically refers to a radial expansion trend layer, a sector extreme difference distribution layer, and an angle mapping layer. The residual weight coefficient table includes a dynamic error response weight, a static error stability weight, and a direction sensitivity coefficient.

[0012] As a further solution of the present invention, the steps of using a glass linear scale to collect multiple size error measurement values at multiple calibration positions and calculate the average value, and using a circular target image to synchronously collect the residual extreme difference values of multiple measurement points in a dynamic scene and the residual mean value in a static scene to obtain a standardized calibration data set are specifically as follows:

[0013] S101: Using a glass linear scale, perform multiple size measurements at each calibration position, call the multiple size error measurement values corresponding to each position, calculate the average value between each group of size error measurement values, and establish a size error average value set in combination with the position number;

[0014] S102: Call the set of average dimensional error values, use the circular target image to collect the residual extreme difference values of each measurement point in the dynamic scene, obtain the residual mean values of each group of measurement points in the static scene, and generate a multi-scene measurement point residual data set;

[0015] S103: According to the multi-scene measurement point residual data set, combined with the measurement position information, standardize the dimensional error measurement values, dynamic residual extreme difference values, and static residual mean values, and respectively construct measurement data sets, including a dimensional error measurement value data set, a dynamic residual extreme difference value data set, and a static residual mean value data set, to obtain a standardized calibration data set.

[0016] As a further solution of the present invention, the steps of calling the standardized calibration data set, using the dimensional error measurement value data set to calculate the average error value of each calibration position, analyzing the gradient change and calculating the smoothing correction parameter, and using discrete points to construct a continuous error field of the entire workbench and generating a global error gradient field are specifically as follows:

[0017] S201: Call the standardized calibration data set, use the dimensional error measurement value data set to extract each group of measurement coordinate values and error values, calculate the average error value of each group of coordinate points, establish the corresponding relationship between the error value and the coordinate number in combination with the position information, and generate a coordinate error mean value list;

[0018] S202: According to the coordinate error mean value list, obtain the gradient change rate of the error data at each position according to the error change trend of adjacent coordinate points, and calculate the smoothing correction parameter;

[0019] S203: Call the smoothing correction parameter, combined with spatial interpolation, to construct a continuous error field of the entire workbench and obtain a global error gradient field.

[0020] As a further solution of the present invention, the specific formula for constructing the continuous error field of the entire workbench is:

[0021] ;

[0022] Calculate the coordinate interpolation prediction value;

[0023] where represents the error interpolation prediction value of the coordinate point , represents the smoothing correction parameter of the th measurement point, represents the residual weight factor of the th measurement point, represents the Euclidean distance between the th measurement point and the target point , represents the smoothing control factor, represents the total number of measurement points participating in the interpolation calculation, represents the measurement point number, represents the X-axis coordinate value of the target interpolation point, represents the Y-axis coordinate value of the target interpolation point.

[0024] As a further solution of the present invention, the steps of calling the standardized calibration data set, using the dynamic residual extreme difference data set, taking the fitting center as the origin, dividing the residual distribution area into multiple sector units according to a fixed angle, calculating the sector residual extreme difference by aggregating the residual values in each sector, analyzing the change trend of the radial distance of the measurement points, and generating a residual trend map are specifically as follows:

[0025] S301: Call the standardized calibration data set, use the dynamic residual extreme difference data set, extract the coordinate position data of each measurement point relative to the fitting center, and divide the residual distribution area into multiple sector units in sequence according to the coordinate angle of each measurement point to generate a sector division coordinate group;

[0026] S302: Based on the sector division coordinate group, aggregate all the residual measurement point values in each sector, record the residual extreme difference corresponding to each sector, analyze and mark the residual fluctuation direction to obtain a sector residual fluctuation parameter set;

[0027] S303: According to the sector residual fluctuation parameter set, use the residual extreme difference and fluctuation direction of each sector, combine with the radial distance data from each measurement point to the center, extract the change amount of the radial distance of the measurement points in each sector, analyze the change trend of the radial distance of the measurement points, and establish a residual trend map.

[0028] As a further solution of the present invention, the steps of calling the standardized calibration data set, calling the dynamic residual extreme difference data set and the static residual mean data set, calculating the weight coefficients of the dynamic residual and the static residual by analyzing the influence degrees of the dynamic residual fluctuation data and the static residual stability on the systematic error, and generating a residual weight coefficient table are specifically as follows:

[0029] S401: Call the standardized calibration data set, call the dynamic residual extreme difference data set and the static residual mean data set, calculate the extreme differences of multiple measurement points in the dynamic class to analyze the dynamic residual fluctuation intensity, calculate the mean square deviation values of multiple measurement points in the static class to identify the static residual stability, and generate a residual characteristic data group;

[0030] S402: Based on the residual characteristic data group, perform normalization processing on the fluctuation intensity index and the stability index, and establish a residual influence analysis result by analyzing the influence degrees of the fluctuation intensity index and the stability index on the visual measurement system error;

[0031] S403: According to the residual impact analysis result, set the dynamic residual coefficient and the static residual coefficient according to the impact degree, and obtain the residual weight coefficient table.

[0032] As a further solution of the present invention, the method further includes:

[0033] S5: Call the global error gradient field, the residual trend map and the residual weight coefficient table, use the gradient field as the size error compensation reference value, use the trend map as the directional error correction value, combine the residual weight coefficient, dynamically adjust the application intensity of the directional correction value, and according to the coordinate data, fuse the adjusted correction value and the size error compensation value, construct a calibration matrix to calibrate the vision measurement system, and obtain the system calibration result;

[0034] The system calibration result is specifically the coordinate mapping compensation value, the directional error adjustment coefficient, and the fused correction output value.

[0035] As a further solution of the present invention, the steps of calling the global error gradient field, the residual trend map and the residual weight coefficient table, using the gradient field as the size error compensation reference value, using the trend map as the directional error correction value, combining the residual weight coefficient, dynamically adjusting the application intensity of the directional correction value, and according to the coordinate data, fusing the adjusted correction value and the size error compensation value, constructing a calibration matrix to calibrate the vision measurement system, and obtaining the system calibration result are specifically as follows:

[0036] S501: Call the global error gradient field, the residual trend map and the residual weight coefficient table, calculate the basic compensation value and the directional correction value of multiple coordinate points according to the size error compensation value and the directional error correction value, and generate a coordinate mapping error value group;

[0037] S502: According to the coordinate mapping error value group, adjust the directional correction value according to the dynamic and static residual weight values corresponding to each coordinate point, and obtain a set of correction adjustment application values;

[0038] S503: Based on the set of correction adjustment application values, according to the coordinate data, fuse the adjusted correction value and the size error compensation value, calculate the error correction parameters at multiple positions, construct a calibration matrix and calibrate the vision measurement system, and obtain the system calibration result;

[0039] The specific formula for calculating the error correction parameters at multiple positions is:

[0040] ;

[0041] Calculate the fused normalized error correction parameter;

[0042] Wherein, represents the fusion standard error correction parameter value of the th coordinate position, represents the fusion weight coefficient, represents the th coordinate position's directional correction adjustment value, represents the th coordinate position's dimensional error compensation value, is the total number of coordinates of the calibration points to be fused and calibrated, is the number of the coordinate point currently being calculated.

[0043] On the other hand, an industrial machine vision measurement system calibration device is provided. This device is applied to an industrial machine vision measurement system calibration method. The device includes:

[0044] The data acquisition module, based on a glass linear scale, collects multiple dimensional error measurement values at multiple calibration positions and calculates the average value. Using circular target images, it synchronously obtains the residual measurement data of multiple measurement points in dynamic and static scenarios, extracts the residual range of dynamic measurement points and the residual variance of static measurement points, and generates a standardized calibration data set;

[0045] The error field construction module, based on the standardized calibration data set, uses the dimensional error measurement value data set to extract the coordinates of multiple calibration points and the corresponding error average values. According to the error change amplitude and spatial distance between adjacent points, it performs gradient analysis, calculates the smoothing correction parameter, and fills the error data in the area through spatial interpolation to establish a global error gradient field;

[0046] The residual analysis module, based on the standardized calibration data set, uses the dynamic residual range value data set, calls the fitted center position as the origin of angle division, divides the residual distribution area into multiple equal-angle sector units, aggregates the residual measurement points in multiple sectors and calculates the range, and extracts the change trend in multiple sectors in combination with radial distance information to establish a residual trend map;

[0047] The weight calculation module, based on the standardized calibration data set, uses the dynamic residual range value data set and the static residual average data set to perform normalization processing on the two groups of residual data, analyzes the influence degree of dynamic residual fluctuation data on the system error, analyzes the influence degree of static residual stability on the system error, obtains the weight coefficients of dynamic and static residuals, and obtains a residual weight coefficient table;

[0048] The compensation and fusion module calls the global error gradient field, the residual trend map, and the residual weight coefficient table, extracts the error compensation value, direction correction value, and application intensity coefficient of each coordinate point, fuses the direction correction value and the dimensional compensation value, constructs calibration matrix parameters, calibrates the vision measurement system, and obtains the system calibration result.

[0049] The beneficial effects brought by the technical solutions provided in the embodiments of the present invention at least include:

[0050] Through the fusion modeling of multi-source error data, the recognition accuracy of the error sources by the system is improved. Combining the construction of the gradient field and the extraction of the trend map enhances the directivity and spatial continuity of error compensation. By using the weight coefficient allocation mechanism, differential control over the influence of dynamic and static errors is achieved. A calibration matrix with multi-dimensional parameter fusion is adopted to optimize the consistency and adaptability of the overall measurement accuracy, enabling the system to maintain stable error correction ability and calibration accuracy in complex environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0052] Figure 1 It is a schematic diagram of the working process of the present invention;

[0053] Figure 2 It is a flowchart of the device of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0054] The following describes the technical solutions in the present invention with reference to the drawings.

[0055] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to represent examples, illustrations or explanations. Any embodiment or design solution described as an "example" in the present invention should not be construed as being more preferred or having more advantages than other embodiments or design solutions. Exactly speaking, the use of the word "example" aims to present concepts in a specific way. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one of the two.

[0056] In the embodiments of the present invention, "image" and "picture" can sometimes be used interchangeably. It should be noted that when the difference is not emphasized, the meanings they express are the same. "(of)", "corresponding", and "corresponding" can sometimes be used interchangeably. It should be noted that when the difference is not emphasized, the meanings they express are the same.

[0057] In the embodiments of the present invention, sometimes subscripts such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meanings they express are the same.

[0058] To make the technical problems, technical solutions, and advantages to be solved by the present invention clearer, the following will be described in detail with reference to the accompanying drawings and specific embodiments.

[0059] Please refer to Figure 1 , the present invention provides a technical solution, a calibration method for an industrial machine vision measurement system, including the following steps:

[0060] S1: Using a glass linear scale, collect multiple size error measurement values at multiple calibration positions and calculate the average value. Using a circular target image, synchronously collect the residual extreme difference values of multiple measurement points in a dynamic scene and the residual mean value in a static scene to obtain a standardized calibration data set;

[0061] S2: Call the standardized calibration data set, use the size error measurement value data set to calculate the average error value of each calibration position, analyze the gradient change and calculate the smoothing correction parameter, and use discrete points to construct a continuous error field for the entire workbench area to generate a global error gradient field;

[0062] S3: Call the standardized calibration data set, use the dynamic residual extreme difference data set, with the center of the fitted circle as the origin, divide the residual distribution area into multiple sector units according to a fixed angle, calculate the sector residual extreme difference by aggregating the residual values within each sector, analyze the change trend of the radial distance of the measurement points, and generate a residual trend map;

[0063] S4: Call the standardized calibration data set, call the dynamic residual extreme difference data set and the static residual mean data set, calculate the weight coefficients of the dynamic residual and the static residual by analyzing the influence degree of the dynamic residual fluctuation data and the static residual stability on the system error, and generate a residual weight coefficient table;

[0064] S5: Call the global error gradient field, the residual trend map, and the residual weight coefficient table. Use the gradient field as the reference value for size error compensation, use the trend map as the directional error correction value, combine the residual weight coefficient, dynamically adjust the application intensity of the directional correction value, and according to the coordinate data, fuse the adjusted correction value and the size error compensation value to construct a calibration matrix to calibrate the vision measurement system and obtain the system calibration result.

[0065] The global error gradient field specifically includes an error change direction layer, a smoothing correction parameter layer, and a path continuity fitting layer. The residual trend map specifically refers to a radial expansion trend layer, a sector extreme difference distribution layer, and an angle mapping layer. The residual weight coefficient table includes a dynamic error response weight, a static error stability weight, and a direction sensitivity coefficient. The system calibration result specifically includes a coordinate mapping compensation value, a directional error adjustment coefficient, and a fused correction output value.

[0066] Using a glass linear scale, multiple size error measurement values are collected at multiple calibration positions and the average value is calculated. Using a circular target image, the residual extreme difference values of multiple measurement points in a dynamic scene and the residual mean value in a static scene are collected synchronously. The steps to obtain the standardized calibration data set are as follows:

[0067] S101: Using a glass linear scale, multiple size measurements are performed at each calibration position, the multiple size error measurement values corresponding to each position are called, the average value between each group of size error measurement values is calculated, and combined with the position number, a size error average value set is established;

[0068] Select 4 calibration positions on the system workbench surface, including 1 position on each of the two diagonals ( and direction) along the measurement range, and 1 position in the direction parallel to axis and axis respectively. At each calibration position, use a standard glass linear scale as the calibration reference, and perform no less than 3 size measurements according to the requirement that the minimum interval does not exceed the measurement stroke and the maximum interval is not less than the measurement stroke ( axis direction) or (diagonal direction). Subsequently, data acquisition of three repeated measurements is performed on each group of measurement values, the difference between its average value and the actual value of the glass linear scale is calculated to obtain the size measurement error at each position, and the error average value is classified and stored according to the calibration position based on the position number to form a size error average value set.

[0069] S102: Call the size error average value set, use the circular target image, collect the residual extreme difference value of each measurement point in the dynamic scene, obtain the residual mean value of each group of measurement points in the static scene, and generate a multi-scene measurement point residual data group;

[0070] Based on the size error average value set, further perform multi-scene data acquisition through the circular target image. In the dynamic scene, fix the circular target horizontally on the workbench, uniformly select 25 measurement points within the field of view by moving the workbench, collect data successively using the single-point measurement method, fit the least squares circle using all the measurement point data and calculate the center of the circle, and then obtain the extreme difference value of the radius of each measurement point ( ) as the dynamic residual. In the static scene, select a circular target with an image diameter occupying of the field of view, keep the workbench stationary, perform multi-point measurements on the circumference of the target from different positions in the field of view, and calculate the residual mean value of each group of measurement point data. Finally, integrate the dynamic residual extreme difference and the static residual mean value to generate a multi-scene measurement point residual data group containing the dynamic extreme difference and the static mean value.

[0071] S103: Based on the multi-scenario measurement point residual data set, combined with the measurement position information, standardize the measured dimension error values, dynamic residual extreme difference values, and static residual mean values, and respectively construct measurement data sets, including the measured dimension error value data set, dynamic residual extreme difference value data set, and static residual mean value data set, to obtain a standardized calibration data set;

[0072] Combined with the spatial distribution information of the measurement positions, standardize the three types of data: the average dimension error, dynamic residual extreme difference, and static residual mean value. This includes normalizing the dimension error values to eliminate the dimension differences, converting the dynamic residual extreme difference into a standardized difference related to the measurement travel, and scaling the static residual mean value according to the field of view ratio. Subsequently, respectively construct the measured dimension error value data set, dynamic residual extreme difference value data set, and static residual mean value data set corresponding one-to-one with the calibration positions, and realize data association through position numbers. Finally, integrate the three standardized data sets to form a comprehensive calibration data set containing spatial position information, providing standardized input for system error analysis and parameter optimization.

[0073] The steps of calling the standardized calibration data set, using the measured dimension error value data set to calculate the average error value of each calibration position, analyzing the gradient change and calculating the smoothing correction parameter, and using discrete points to construct a continuous error field across the workbench and generating a global error gradient field are as follows:

[0074] S201: Call the standardized calibration data set, use the measured dimension error value data set to extract each set of measurement coordinate values and error values, calculate the average error value of each set of coordinate points, combine the position information to establish the corresponding relationship between the error value and the coordinate number, and generate a list of coordinate error mean values;

[0075] When calling the standardized calibration data set, it is necessary to extract the spatial coordinate information of the measurement points and their corresponding error values from the original measured dimension error value data set. Each set of data is indexed by the point position number P, and records the actual position of this point in the workbench coordinate system and the corresponding measured error value , and perform an average calculation on the multiple error measurement values under the same numbered point position to filter out the error offsets caused by environmental fluctuations and image noise. Taking the coordinate point P1 as an example, the position is (100, 200), and its three error values are 0.08mm, 0.10mm, and 0.07mm. Using the average error calculation, execute the following formula:

[0076] ;

[0077] where is the average error of the th coordinate point, is the error value of this point under the th measurement, is the number of sampling times at this point. Set , , , , substitute into the calculation:

[0078] ;

[0079] The calculation result is the average measurement error of point P1 in the current calibration cycle. This value is bound to the coordinate number of this point and is recorded as (P1, 100, 200, 0.0833). The same process is sequentially executed for all coordinate points to construct a correspondence table between error values and spatial coordinates, forming a structured error list with the point number as the index, the coordinate position as the spatial reference, and the error mean as the data field, which is used for subsequent spatial gradient modeling operations, and finally a list of coordinate error means is generated.

[0080] S202: According to the list of coordinate error means, based on the error change trend of adjacent coordinate points, obtain the gradient change rate of the error data at each position and calculate the smoothing correction parameter;

[0081] According to the list of coordinate error means, perform grid mapping on the distribution relationship of all points in space, arrange the coordinate points in rows and columns according to the spatial sequence, extract the difference in error values between each pair of adjacent coordinate points, and calculate the error change rate per unit distance as the error gradient rate parameter in this direction. In actual processing, taking two points P1(100, 200) and P2(120, 200) as examples, the error values are respectively recorded as 0.08 mm and 0.14 mm. The two points are 20 mm apart in the X-axis direction, then the error change rate per unit spacing is (0.14 - 0.08) / 20 = 0.003 mm / mm. The error gradient change rate adopts the following formula:

[0082] ;

[0083] Among them, is the error gradient change rate between point positions i and j, is the error mean of the two points, is the spatial distance between the two points. Set , , , substitute into the calculation:

[0084] ;

[0085] Compare the calculated gradient value with the set error fluctuation reference value. The set reference value is 0.002. If the gradient value is greater than the reference value, it is recorded as an abnormal gradient segment, and a smoothing parameter needs to be added in the interpolation modeling for correction. The correction parameter can be dynamically assigned according to the gradient ratio in the abnormal gradient area. In the above example, since the error change rate is 0.003, which exceeds the reference value, it is marked as a segment requiring smoothing adjustment. After calculating the gradient rate for all points, the results are stored according to the point pair coordinates, and the directionality (X-direction or Y-direction) is marked, which is subsequently used to generate the smoothing weight in the interpolation fitting, and finally the smoothing correction parameter is calculated.

[0086] S203: Call the smoothing correction parameter, combine with spatial interpolation, construct the continuous error field of the entire workbench, and obtain the global error gradient field;

[0087] The specific formula for constructing the continuous error field of the entire workbench is:

[0088] ;

[0089] Calculate the coordinate interpolation prediction value;

[0090] Among them, represents the error interpolation prediction value of the coordinate point , represents the th smoothing correction parameter of the measurement point, represents the th residual weight factor of the measurement point, represents the th measurement point and the target point Euclidean distance, represents the smoothing control factor, represents the total number of measurement points participating in the interpolation calculation, represents the measurement point number, represents the X-axis coordinate value of the target interpolation point, represents the Y-axis coordinate value of the target interpolation point.

[0091] Formula:

[0092] ;

[0093] Detailed explanation of the formula and the derivation process of the formula calculation:

[0094] The formula is used to calculate the error interpolation prediction value of each coordinate point, and the result is used to construct the continuous error field of the workbench area and generate the global error gradient field;

[0095] Parameter meaning and setting value:

[0096] is the The smoothing correction parameter for each measurement point is set to ;

[0097] is the residual weight factor for the th measurement point, which is set to ;

[0098] is the Euclidean distance between the th measurement point and the target interpolation point . Assuming the coordinates of the measurement points are , and the target point is , the corresponding distances are, , , ;

[0099] is the smoothing control factor, which is set to ;

[0100] is the number of measurement points participating in interpolation, which is set to 3;

[0101] is the target interpolation coordinate, which is set to .

[0102] Substitute the parameters into the formula for calculation:

[0103] ;

[0104] ;

[0105] ;

[0106] The result 0.2536 represents the error interpolation prediction value of the target point, which is used to generate a continuous error field subsequently and serves as the basic interpolation data source for the global error gradient field. The calculation results can be extended to other coordinate points in the same way to construct the error field distribution of the complete grid area.

[0107] The steps of calling the standardized calibration dataset, using the dynamic residual extreme value dataset, taking the fitting center as the origin, dividing the residual distribution area into multiple sector units according to a fixed angle, calculating the sector residual extreme value by aggregating the residual values in each sector, and analyzing the change trend of the radial distance of the measurement points to generate the residual trend map are as follows:

[0108] S301: Call the standardized calibration dataset, use the dynamic residual extreme value dataset, extract the coordinate position data of each measurement point relative to the fitting center, and divide the residual distribution area into multiple sector units in sequence according to the coordinate angle of each measurement point to generate a sector division coordinate group;

[0109] Call the standardized calibration data set and utilize the dynamic residual extreme difference data set. First, fit the contour boundary of the circular target through image processing, identify the coordinate positions of each measurement point, and take the fitted center coordinate as the origin to convert the coordinates of all measurement points into polar coordinate form with the center as the origin. Among them, the angle value represents the polar angle position of the measurement point relative to the center of the circle. The measurement point numbers are sorted according to the polar angle size to construct a continuous angle sequence from 0° to 360°. To classify the measurement points according to the angle, divide the 360° circumference into several sector units with a fixed angle width. For example, if each sector is set to 15°, then a total of 24 sectors are divided. Each sector contains all the measurement points within the polar angle range, extract the coordinate data of each measurement point , and combine its polar angle relative to the center of the circle to assign it to the corresponding sector. Taking the center of the circle as (200, 200) and the measurement point A as (210, 210), it is calculated that

[0110] S302: Based on the sector division coordinate group, aggregate all the residual measurement point values within each sector, record the residual extreme difference corresponding to each sector, analyze and mark the residual fluctuation direction to obtain the sector residual fluctuation parameter set;

[0111] Based on the sector division coordinate group, the system performs an aggregation operation on all the residual values within each sector, extracts the residual value of each measurement point in the dynamic scenario, and the residual value is defined as the absolute value of the difference between the radius of the current frame of the measurement point and the average radius of the fitted circle. Perform the range calculation on the residual value sequence under each sector, that is, the difference between the maximum residual and the minimum residual. The calculation process uses the following formula:

[0112] ;

[0113] where is the residual range of the s-th sector, is the residual value of the j-th measurement point within the sector. Suppose the residual values of the measurement points in the 5th sector are 0.04, 0.06, 0.09, 0.03 mm, then:

[0114] ;

[0115] When the residual range of a certain sector exceeds the set fluctuation reference value, the sector is marked as having a fluctuation direction. According to the distribution direction of the measurement points and the polar angle position, the direction vector of this fluctuation direction is defined. The judgment basis for the residual fluctuation direction is the product sign of the difference between the residuals of two adjacent measurement points and their polar angle directions. If the product is greater than zero, it is defined as an expansion trend; if it is less than zero, it is a contraction trend. The system processes each sector in sequence according to the angle serial number, records the residual range and the fluctuation direction, and integrates all the results into a three-field data table with "sector number - polar difference - fluctuation trend", and finally obtains the sector residual fluctuation parameter set.

[0116] S303: According to the sector residual fluctuation parameter set, using the residual polar difference and the fluctuation direction of each sector, combined with the radial distance data from each measurement point to the center of the circle, extract the change amount of the radial distance of the measurement points in each sector, analyze the change trend of the radial distance of the measurement points, and establish a residual trend map;

[0117] According to the sector residual fluctuation parameter set, the system further calls the actual coordinate data of each measurement point and the center coordinate to calculate its radial distance, denoted as Each measurement point in each sector constitutes a radial distance set. By calculating the difference between the maximum radial distance and the minimum radial distance, it is judged whether there is a non-uniform distribution in the radial structure within the sector. Then, combined with the residual polar difference and the directional annotation obtained in the previous stage, the radial distances within each sector are sorted according to the polar angle sequence of the measurement points, and the trend of the radial change curve is observed. If it shows a monotonically increasing trend and the residual direction is marked as expansion, it is confirmed that there is an error expansion behavior in this sector; otherwise, it is a contraction behavior. If the change is irregular, it is marked as having no significant direction. Organize this analysis logic into a directional trend data structure, and the trend value is calculated using the following formula:

[0118] ;

[0119] where is the radial change trend value of the s-th sector, is the radial distance of the j-th measurement point in ascending order of polar angle, is the number of measurement points in the s-th sector. Suppose the radial distances of five points in the 6th sector are 8.0, 8.3, 8.5, 8.6, 8.7 mm, then:

[0120] ;

[0121] The value is positive and continuously increasing, marked as expansion. Summarize the trend results of all sectors into a residual trend curve mapping diagram, which reflects the spatial structural evolution state of the error in polar coordinates, and finally forms a residual trend map.

[0122] Call the standardized calibration data set, call the dynamic residual extreme value data set and the static residual mean data set. By analyzing the influence degrees of the dynamic residual fluctuation data and the static residual stability on the system error, calculate the weight coefficients of the dynamic residual and the static residual, and the steps for generating the residual weight coefficient table are specifically as follows:

[0123] S401: Call the standardized calibration data set, call the dynamic residual extreme value data set and the static residual mean data set. By calculating the extreme values of multiple measurement points in the dynamic category, analyze the dynamic residual fluctuation intensity. By calculating the mean square deviation values of multiple measurement points in the static category, identify the static residual stability, and generate a residual characteristic data group;

[0124] Call the standardized calibration data set, call the dynamic residual extreme value data set and the static residual mean data set. First, for the dynamic residual part, extract the maximum value and the minimum value of each measurement point in the continuous acquisition frames, and record its extreme value. The extreme value calculation adopts the formula:

[0125] ;

[0126] Among them, is the dynamic residual extreme value of a certain measurement point, is the id-th measurement radius value of this measurement point during the dynamic acquisition process. Suppose the continuous measurement radii of measurement point A are 8.22, 8.19, 8.31, 8.25 mm, then:

[0127] ;

[0128] The system performs the same operation on all dynamic measurement points to obtain the set of dynamic residual extreme values of all-domain measurement points. At the same time, for the measurement point data in the static scenario, the system collects the residual value sequences of each measurement point in different static frames, calculates the mean square deviation, and uses the formula:

[0129] ;

[0130] Among them, is the static residual mean square deviation, is the jt-th measurement value, is the average value of the residual values of all frames of this measurement point. Suppose the measurement values of measurement point B in the static scenario are 0.02, 0.01, 0.03, 0.01 mm, then the average value is:

[0131] ;

[0132] ;

[0133] The system aggregates the dynamic extreme differences and static mean square deviations of all measurement points into a dynamic residual fluctuation intensity data series and a static residual stability data series respectively, and integrates them into a residual characteristic data group according to the corresponding measurement point numbers.

[0134] S402: Based on the residual characteristic data group, perform normalization processing on the fluctuation intensity index and the stability index, and establish a residual influence analysis result by analyzing the influence degree of the fluctuation intensity index and the stability index on the visual measurement system error;

[0135] Based on the residual characteristic data group, the system performs normalization processing on the dynamic residual extreme difference value and the static residual mean square deviation value respectively. The normalization reference is set as the maximum value of the residual values of all measurement points in the current scene, and the normalization formula is:

[0136] ;

[0137] Among them, is the residual index value of the ic-th measurement point after normalization, is the original residual value, is the maximum residual reference value. If the maximum dynamic residual is 0.14 mm and the measurement point A is 0.12 mm, then:

[0138] ;

[0139] Similarly, the static residual is normalized in the same way. After normalization, the system calls the normalized dynamic and static residual indexes to construct an error influence matrix. The dynamic normalization value and the static normalization value of each measurement point are used as two dimensions, and the corresponding coordinate points are the error influence distribution points. Depending on the two-dimensional distribution, calculate the offset angle and influence amplitude of each measurement point in the error map. Use the method of subtracting the static weight value from the dynamic weight value to initially judge the main source direction of the system error. The system extracts the high-frequency band as the main direction of the residual influence according to the distribution density of the offset angle concentration area, and establishes a residual influence analysis result.

[0140] S403: According to the residual influence analysis result, set the dynamic residual coefficient and the static residual coefficient according to the influence degree, and obtain the residual weight coefficient table;

[0141] According to the residual influence analysis result, the system sets the residual influence proportionality coefficient, establishes an association relationship between the normalized values of the dynamic and static residuals and the influence degree. The dynamic residual weight coefficient and the static residual weight coefficient are set to satisfy the following constraints:

[0142] ;

[0143] According to the ratio of the normalized values, the weight coefficient is calculated by the proportional method. If the normalized dynamic residual of a measuring point is 0.8 and the static residual is 0.2, then:

[0144] ;

[0145] ;

[0146] The system stores the dynamic and static residual weight coefficients corresponding to each measuring point in a tabular form, forming a three-field structure of "measuring point number - dynamic weight - static weight". After loop processing all the measuring points, it is integrated and output as a residual weight coefficient table.

[0147] Call the global error gradient field, residual trend map, and residual weight coefficient table. Use the gradient field as the benchmark value for dimensional error compensation, the trend map as the directional error correction value, combine the residual weight coefficient, dynamically adjust the application intensity of the directional correction value. According to the coordinate data, fuse the adjusted correction value and dimensional error compensation value, and the steps to calibrate the vision measurement system by constructing a calibration matrix and obtaining the system calibration result are as follows:

[0148] S501: Call the global error gradient field, residual trend map, and residual weight coefficient table. According to the dimensional error compensation value and directional error correction value, calculate the basic compensation value and directional correction value of multiple coordinate points, and generate a coordinate mapping error value group;

[0149] The coordinate error calculation sub-module extracts the numbers, horizontal and vertical axis indices, and corresponding spatial position coordinates of all measurement coordinate points one by one according to the dimensional error compensation value established in the global error gradient field and the directional error correction value marked in the residual trend map. Perform a dual-value reading operation on each coordinate point, that is, obtain the basic compensation value E1 of the point in the error gradient field and the directional correction value E2 in the trend map respectively, and perform an average merging operation on these two error values for subsequent error fusion processing. During the process, it is necessary to ensure the accurate correspondence of values through the system coordinate index matching method. In addition, the corresponding compensation source mark needs to be written into the mapping table for subsequent weight calling and adjustment processing. This process binds each coordinate number to its corresponding merged error value, and finally generates a complete coordinate mapping error value group for weighted correction in the next stage; The formula is:

[0150] ;

[0151] Calculate the coordinate merged error value, where, is the basic merged compensation value of the coordinate point, is the dimensional error compensation value, is the directional correction value. Set to 10, Substitute 6 for calculation:

[0152] ;

[0153] The calculation results show that the combined error value is 8 in space. Subsequently, the error values of all points are uniformly transferred into the coordinate mapping table and stored in sequence according to the number and the actual position number, completing the data operation of the error distribution structure. This data set will be used as the data input port for subsequent weight superposition and correction to establish a coordinate mapping error value group.

[0154] S502: According to the coordinate mapping error value group, adjust the directional correction value according to the dynamic and static residual weight values corresponding to each coordinate point to obtain a set of corrected adjustment application values;

[0155] The direction correction weighting sub-module receives the directional correction values of all position points in the coordinate mapping error value group and performs weighted adjustment processing based on the dynamic residual coefficient α and the static residual coefficient β. In this processing step, the α and β values corresponding to each coordinate number are first extracted from the residual weight coefficient table, and the direction correction values are weighted and summed. A normalization coefficient λ is introduced to balance the influence of extreme disturbance measurement points on the overall correction result. Specifically, weighted multiplication processing is performed on each group of measurement points and written into the structure data set. Each item in this data set includes the coordinate number, the original direction correction value, the weighted adjustment value, and the corrected error value; The formula is used:

[0156] ;

[0157] Calculate the direction correction weighted value, where is the weighted adjustment application value for directional error correction, is the dynamic residual weight coefficient, is the static residual weight coefficient, is the original directional correction value, is the normalization coefficient. Set to 0.3, to 0.4, to 0.85, Substitute 6 for calculation:

[0158] ;

[0159] After the calculation is completed, remap each weighted adjustment value to the corresponding coordinate number, overwrite the original direction correction item in the mapping value table through the coordinate index method, update it to the weight correction result, and obtain a set of corrected adjustment application values.

[0160] S503: Based on the corrected adjusted application value set, fuse the adjusted correction value and the dimensional error compensation value according to the coordinate data, calculate the error correction parameters at multiple positions, construct a calibration matrix and calibrate the vision measurement system to obtain the system calibration result;

[0161] The specific formula for calculating the error correction parameters at multiple positions is:

[0162] ;

[0163] Calculate the standardized error correction parameter after fusion;

[0164] Among them, represents the fused standardized error correction parameter value at the th coordinate position, represents the fusion weight coefficient, represents the directional correction adjustment value at the th coordinate position, represents the dimensional error compensation value at the th coordinate position, is the total number of coordinate points to be fused and calibrated, is the serial number of the currently calculated coordinate point.

[0165] Formula:

[0166] ;

[0167] Detailed explanation of the formula and the derivation process of the formula calculation:

[0168] The formula is used to calculate the fused standardized error correction parameter for each coordinate point, and the parameter is used to construct the calibration matrix of the vision measurement system.

[0169] Meaning and setting value of the parameter:

[0170] is the fusion weight coefficient, set to 0.6;

[0171] is the directional correction adjustment value at the th coordinate point, set to , , , and a total of 3 position points are extracted;

[0172] is the dimensional error compensation value at the th coordinate point, set to , , ;

[0173] is the total number of coordinate points, set to 3;

[0174] Substitute the parameters into the formula for calculation:

[0175] ;

[0176] ;

[0177] The fusion error correction parameter of the first position point is calculated as:

[0178] ;

[0179] The result shows that the standardized error correction parameter after fusion is 1.855, corresponding to the final error correction input value of the first measurement point, which will be used as the error adjustment basis for this position point during the construction of the calibration matrix. The parameters of other position points are processed in the same way and uniformly written into the calibration matrix to complete the error correction process of the vision measurement system.

[0180] Please refer to Figure 2 , an industrial machine vision measurement system calibration device. The industrial machine vision measurement system calibration device is used to execute the above-mentioned industrial machine vision measurement system calibration method. The device includes:

[0181] The data acquisition module, based on the glass linear scale, collects multiple size error measurement values at multiple calibration positions and calculates the average value. Using the circular target image, it synchronously obtains the residual measurement data of multiple measurement points in dynamic and static scenarios, extracts the residual range of dynamic measurement points and the residual variance of static measurement points, and generates a standardized calibration data set;

[0182] The error field construction module, based on the standardized calibration data set, uses the size error measurement value data set to extract the coordinates of multiple calibration points and the corresponding average error values. According to the error change amplitude and spatial distance between adjacent points, it performs gradient analysis, calculates the smoothing correction parameter, and fills the error data in the area through spatial interpolation to establish a global error gradient field;

[0183] The residual analysis module, based on the standardized calibration data set, uses the dynamic residual range value data set, calls the fitted center position of the circle as the origin of angle division, divides the residual distribution area into multiple equal-angle sector units, aggregates the residual measurement points in multiple sectors and calculates the range, and extracts the change trend in multiple sectors in combination with the radial distance information to establish a residual trend map;

[0184] The weight calculation module, based on the standardized calibration data set, uses the dynamic residual extreme difference data set and the static residual mean data set to normalize the two groups of residual data, analyzes the influence degree of the dynamic residual fluctuation data on the system error, analyzes the influence degree of the static residual stability on the system error, obtains the weight coefficients of the dynamic residual and the static residual, and obtains the residual weight coefficient table;

[0185] The compensation and fusion module calls the global error gradient field, the residual trend map, and the residual weight coefficient table, extracts the error compensation value, the direction correction value, and the application intensity coefficient of each coordinate point, fuses the direction correction value and the dimension compensation value, constructs the calibration matrix parameters, calibrates the vision measurement system, and obtains the system calibration result.

[0186] The above embodiments can be implemented in whole or in part by software, hardware (such as circuits), firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on the computer, the processes or functions described in the embodiments of the present invention are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by wire (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that the computer can access or a data storage device such as a server or data center that contains one or more collections of available media. The available medium can be a magnetic medium (such as a floppy disk, hard disk, magnetic tape), an optical medium (such as a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state drive.

[0187] It should be understood that the term "and / or" in this article is only a description of the association relationship between associated objects, indicating that three relationships can exist. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. Among them, A and B can be singular or plural. In addition, the character " / " in this article generally represents an "or" relationship between the associated objects before and after, but it may also represent an "and / or" relationship, which can be specifically understood with reference to the context.

[0188] In the present invention, "at least one" means one or more, and "a plurality" means two or more. "At least one of the following" or a similar expression means any combination of these items, including any combination of single item(s) or plural item(s). For example, at least one of a, b, or c can represent: a, b, c, a - b, a - c, b - c, or a - b - c, where a, b, and c can be single or plural.

[0189] It should be understood that in various embodiments of the present invention, the magnitudes of the serial numbers of the above - mentioned processes do not imply the order of execution. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.

[0190] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods for each specific application to implement the described functions, but such implementation should not be considered to exceed the scope of the present invention.

[0191] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the devices, apparatuses, and units described above can refer to the corresponding processes in the foregoing method embodiments, and will not be elaborated herein.

[0192] In several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division. In actual implementation, there can be other division methods. For example, multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the couplings or direct couplings or communication connections shown or discussed with each other can be through some interfaces. The indirect couplings or communication connections of the devices or units can be in electrical, mechanical, or other forms.

[0193] The units described as separate components may or may not be physically separated. The components shown as units may or may not be physical units, that is, they can be located in one place, or can be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0194] In addition, in each embodiment of the present invention, each functional unit can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit.

[0195] If the above-mentioned functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs.

[0196] The above is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. A calibration method for an industrial machine vision measurement system, characterized in that, The method includes: S1: Using a glass linear scale, collecting multiple size error measurement values at multiple calibration positions and calculating the average value. Using a circular target image, synchronously collecting the residual extreme difference values of multiple measurement points in a dynamic scene and the residual mean values in a static scene to obtain a standardized calibration data set; The steps of using a glass linear scale to collect multiple size error measurement values at multiple calibration positions and calculating the average value, and using a circular target image to synchronously collect the residual extreme difference values of multiple measurement points in a dynamic scene and the residual mean values in a static scene to obtain a standardized calibration data set are specifically as follows: S101: Using a glass linear scale, performing multiple size measurements at each calibration position, calling the multiple size error measurement values corresponding to each position, calculating the average value between each group of size error measurement values, and combining the position numbers to establish a set of size error average values; S102: Calling the set of size error average values, using a circular target image, collecting the residual extreme difference values of each measurement point in a dynamic scene, obtaining the residual mean values of each group of measurement points in a static scene, and generating a multi-scene measurement point residual data set; S103: According to the multi-scene measurement point residual data set, combining the measurement position information, performing standardized processing on the size error measurement values, dynamic residual extreme difference values, and static residual mean values, and respectively constructing measurement data sets, including a size error measurement value data set, a dynamic residual extreme difference value data set, and a static residual mean value data set, to obtain a standardized calibration data set; S2: Calling the standardized calibration data set, using the size error measurement value data set, calculating the average error value of each calibration position, analyzing the gradient change and calculating the smoothing correction parameter, and using discrete points to construct a continuous error field of the entire workbench to generate a global error gradient field; S3: Calling the standardized calibration data set, using the dynamic residual extreme difference value data set, taking the fitting center as the origin, dividing the residual distribution area into multiple sector units according to a fixed angle, calculating the sector residual extreme difference by aggregating the residual values in each sector, and analyzing the change trend of the radial distance of the measurement points to generate a residual trend map; S4: Calling the standardized calibration data set, calling the dynamic residual extreme difference value data set and the static residual mean value data set, calculating the weight coefficients of the dynamic residual and the static residual by analyzing the influence degree of the dynamic residual fluctuation data and the static residual stability on the system error, and generating a residual weight coefficient table; S5: Calling the global error gradient field, the residual trend map, and the residual weight coefficient table, using the gradient field as the size error compensation reference value, using the trend map as the directional error correction value, combining the residual weight coefficients, dynamically adjusting the application intensity of the directional correction value, and according to the coordinate data, fusing the adjusted correction value and the size error compensation value to construct a calibration matrix to calibrate the vision measurement system and obtain the system calibration result.

2. The calibration method of the industrial machine vision measurement system according to claim 1, characterized in that The global error gradient field specifically refers to the error change direction layer, the smoothing correction parameter layer, and the path continuity fitting layer. The residual trend spectrum specifically refers to the radial expansion trend layer, the sector range difference distribution layer, and the angle mapping layer. The residual weight coefficient table includes the dynamic error response weight, the static error stability weight, and the direction sensitivity coefficient. The system calibration result specifically is the coordinate mapping compensation value, the directional error adjustment coefficient, and the fusion correction output value.

3. The calibration method of the industrial machine vision measurement system according to claim 1, characterized in that The steps of calling the standardized calibration data set, using the dimensional error measurement value data set to calculate the average error value at each calibration position, analyzing the gradient change and calculating the smoothing correction parameter, and using discrete points to construct the continuous error field of the entire workbench to generate the global error gradient field are specifically as follows: S201: Call the standardized calibration data set, use the dimensional error measurement value data set to extract each set of measurement coordinate values and error values, calculate the average error value of each set of coordinate points, establish the corresponding relationship between the error value and the coordinate number in combination with the position information, and generate a coordinate error mean value list. S202: According to the coordinate error mean value list, obtain the gradient change rate of the error data at each position according to the error change trend of adjacent coordinate points, and calculate the smoothing correction parameter. S203: Call the smoothing correction parameter, combine spatial interpolation to construct the continuous error field of the entire workbench, and obtain the global error gradient field.

4. The calibration method of the industrial machine vision measurement system according to claim 3, characterized in that The specific formula for constructing the continuous error field of the entire workbench is: ； Calculate the coordinate interpolation prediction value. Among them, represents the error interpolation prediction value of the coordinate point , represents the smoothing correction parameter of the th measurement point, represents the residual weight factor of the th measurement point, represents the Euclidean distance between the th measurement point and the target point , represents the smoothing control factor, represents the total number of measurement points participating in the interpolation calculation, represents the measurement point number, represents the X-axis coordinate value of the target interpolation point, represents the Y-axis coordinate value of the target interpolation point.

5. The calibration method of the industrial machine vision measurement system according to claim 3, characterized in that, The steps of calling the standardized calibration data set, using the dynamic residual range difference data set, taking the fitting center as the origin, dividing the residual distribution area into multiple sector units according to a fixed angle, calculating the sector residual range difference by aggregating the residual values in each sector, and analyzing the change trend of the radial distance of the measurement points to generate the residual trend spectrum are specifically as follows: S301: Call the standardized calibration data set, use the dynamic residual range difference data set to extract the coordinate position data of each measurement point relative to the fitting center, and divide the residual distribution area into multiple sector units in sequence according to the coordinate angle of each measurement point to generate a sector division coordinate group. S302: Based on the sector division coordinate group, aggregate all the residual measurement point values in each sector, record the residual range difference corresponding to each sector, analyze and mark the residual fluctuation direction, and obtain a sector residual fluctuation parameter set. S303: According to the sector residual fluctuation parameter set, use the residual range difference and fluctuation direction of each sector, combine the radial distance data from each measurement point to the center of the circle, extract the change amount of the radial distance of the measurement points in each sector, analyze the change trend of the radial distance of the measurement points, and establish a residual trend spectrum.

6. The calibration method of the industrial machine vision measurement system according to claim 5, characterized in that, The steps of calling the standardized calibration data set, calling the dynamic residual range difference data set and the static residual mean value data set, calculating the weight coefficients of the dynamic residual and the static residual by analyzing the influence degree of the dynamic residual fluctuation data and the static residual stability on the system error, and generating the residual weight coefficient table are specifically as follows: S401: Call the standardized calibration data set, call the dynamic residual extreme value data set and the static residual mean data set, analyze the dynamic residual fluctuation intensity by calculating the extreme values of multiple measurement points in the dynamic category, identify the static residual stability by calculating the mean square deviation values of multiple measurement points in the static category, and generate a residual characteristic data group; S402: Based on the residual characteristic data group, perform normalization processing on the fluctuation intensity index and the stability index, and establish a residual influence analysis result by analyzing the influence degree of the fluctuation intensity index and the stability index on the visual measurement system error; S403: According to the residual influence analysis result, set the dynamic residual coefficient and the static residual coefficient according to the influence degree, and obtain the residual weight coefficient table.

7. The calibration method of the industrial machine vision measurement system according to claim 6, characterized in that The steps of calling the global error gradient field, the residual trend map and the residual weight coefficient table, using the gradient field as the size error compensation reference value, using the trend map as the directional error correction value, combining the residual weight coefficient, dynamically adjusting the application intensity of the directional correction value, and fusing the adjusted correction value and the size error compensation value according to the coordinate data to construct a calibration matrix to calibrate the visual measurement system and obtain the system calibration result are specifically as follows: S501: Call the global error gradient field, the residual trend map and the residual weight coefficient table, and calculate the basic compensation value and the directional correction value of multiple coordinate points according to the size error compensation value and the directional error correction value to generate a coordinate mapping error value group; S502: According to the coordinate mapping error value group, adjust the directional correction value according to the dynamic and static residual weight values corresponding to each coordinate point to obtain a corrected adjustment application value set; S503: Based on the corrected adjustment application value set, fuse the adjusted correction value and the size error compensation value according to the coordinate data, calculate the error correction parameters of multiple positions, construct a calibration matrix and calibrate the visual measurement system to obtain the system calibration result; The specific formula for calculating the error correction parameters of multiple positions is: ； Calculate the standardized error correction parameter after fusion; Among them, represents the fusion standard error correction parameter value of the th coordinate position, represents the fusion weight coefficient, represents the directional correction adjustment value of the th coordinate position, represents the dimensional error compensation value of the th coordinate position, is the total number of coordinates of the calibration point to be fused, is the number of the coordinate point currently calculated.

8. An industrial machine vision measurement system calibration device, characterized in that, According to the industrial machine vision measurement system calibration method according to any one of claims 1-7, the device includes: The data acquisition module, based on the glass linear scale, collects multiple size error measurement values at multiple calibration positions and calculates the average value, uses the circular target image to synchronously obtain the residual measurement data of multiple measurement points in the dynamic and static scenarios, extracts the residual extreme value of the dynamic measurement points and the residual variance of the static measurement points, and generates a standardized calibration data set; The error field construction module, based on the standardized calibration data set, uses the size error measurement value data set to extract the coordinates of multiple calibration points and the corresponding error average value, performs gradient analysis according to the error change amplitude and the spatial distance between adjacent points, calculates the smoothing correction parameter, and fills the error data in the area by combining spatial interpolation to establish a global error gradient field; Based on the standardized calibration dataset, the residual analysis module uses the dynamic residual extreme value dataset, calls the fitted center position as the origin of angle division, divides the residual distribution region into multiple equal-angle sector units, aggregates the residual measurement points in multiple sectors and calculates the range, combines the radial distance information to extract the change trends in multiple sectors, and establishes a residual trend map; Based on the standardized calibration dataset, the weight calculation module uses the dynamic residual extreme value dataset and the static residual mean dataset to normalize the two groups of residual data, analyzes the influence degree of dynamic residual fluctuation data on the system error, analyzes the influence degree of static residual stability on the system error, obtains the weight coefficients of dynamic residuals and static residuals, and obtains a residual weight coefficient table; The compensation fusion module calls the global error gradient field, the residual trend map and the residual weight coefficient table, extracts the error compensation value, direction correction value and application intensity coefficient of each coordinate point, fuses the direction correction value and the dimension compensation value, constructs the calibration matrix parameters, calibrates the vision measurement system, and obtains the system calibration result.

Images