CN114092529A - Calibration method of two-dimensional measuring instrument - Google Patents
Calibration method of two-dimensional measuring instrument Download PDFInfo
- Publication number
- CN114092529A CN114092529A CN202111336844.XA CN202111336844A CN114092529A CN 114092529 A CN114092529 A CN 114092529A CN 202111336844 A CN202111336844 A CN 202111336844A CN 114092529 A CN114092529 A CN 114092529A
- Authority
- CN
- China
- Prior art keywords
- image
- data
- workpiece
- measuring instrument
- acquiring
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
- G01B11/00—Measuring arrangements characterised by the use of optical techniques
- G01B11/24—Measuring arrangements characterised by the use of optical techniques for measuring contours or curvatures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Theoretical Computer Science (AREA)
- Length Measuring Devices By Optical Means (AREA)
Abstract
The invention relates to a calibration method of a two-dimensional measuring instrument, which solves the technical problems that the existing equipment possibly has device difference when leaving a factory, and the performance is unstable due to the non-universal internal parameters, and comprises the following steps: setting parameters of a transmitter and a receiver on the detection equipment for acquiring an image of an object to be measured; sequentially acquiring images of a workpiece to be measured at different positions, and acquiring an enough number of size image data sets; after edge processing is carried out on the acquired image data, size characteristics of the measured object are generated, and processing is carried out by applying a machine learning algorithm to obtain factory calibration parameters; verifying on a measuring platform, judging whether the precision requirement meets a set range, and if the precision requirement meets the set specification parameter, successfully calibrating; if not, turning to the step 2-3, re-acquiring the image data of the standard workpiece, and calibrating again; and updating the parameters meeting the precision requirement to the measurement model. The invention can be widely applied to factory detection of the detection consistency of the precision measuring instrument.
Description
Technical Field
The invention relates to the field of equipment calibration, in particular to a calibration method of a two-dimensional measuring instrument.
Background
In the high-quality product manufacturing and high-efficiency production environment, the importance of the measurement technology is increasing, and particularly, in the precision measurement instrument, the used electronic equipment is becoming more and more complex, and the used environmental conditions are becoming worse, so that higher requirements are put on electronic components, and not only good characteristics but also high-reliability operation is required.
However, the devices of the measuring equipment inevitably have consistency differences before leaving the factory, such as the consistency of the light source. The differences of the devices may cause different quality of the shot images and the difference of the gray values, which causes the condition that the subsequent image processing parameters are not universal and the performance is unstable, and has very important influence on the whole test result. In order to ensure the measurement accuracy, factory calibration is required in the process of mass production of the device.
Disclosure of Invention
The invention aims to solve the technical problems that the existing equipment is possibly different in devices when leaving a factory, and performance is unstable due to the fact that internal parameters are not universal, and the like, and provides the equipment which does not need to use a calibration plate and correct for many times, has good error correction precision and good detection precision. Meanwhile, the method has the advantages of simplicity in operation, high reliability and the like, and can solve the problem of difference in batch production of equipment.
The invention provides a calibration method of a two-dimensional measuring instrument, which comprises the following steps:
step 1, setting parameters of a transmitter and a receiver on detection equipment for acquiring an image of an object to be measured;
step 2, sequentially acquiring images of the workpiece to be measured at different positions, and acquiring an image data set with enough number and size;
step 3, carrying out edge detection processing on the acquired image data set to obtain profile data of the measured object, and processing the profile data by using a machine learning algorithm to obtain a fitting coefficient of the size of the measured workpiece and a mean value and standard deviation parameter which embody the image quality;
step 4, verifying on the measuring platform, judging whether the precision requirement meets the set range, and if the precision requirement meets the set description parameter, successfully calibrating; if not, turning to the step 2-3, re-acquiring the image data of the standard workpiece, and calibrating again;
and 5, updating the parameters meeting the precision requirement to the measurement model.
Preferably, the detection device in step 1 is a visual acquisition platform of a measuring instrument, the measuring instrument uses an optical transmission principle, a transmitter irradiates green two-dimensional parallel light, a CMOS of a receiver captures a shadow of a target object, and then an image of the measured object is acquired.
Preferably, the specific method of step 2 is: and acquiring image data of a workpiece to be measured in the next size after acquiring enough images of a certain workpiece to be measured in different positions respectively to form image data sets of workpieces to be measured in different sizes.
Preferably, the specific steps of step 3 include:
step (1), edge detection: carrying out pixel-level edge detection on the image by using a Canny edge detection operator to obtain coordinates of left and right edge points of the pixel of the image, and simultaneously removing coordinates of abnormal points of the image;
and (2) parameter fitting: respectively substituting left and right edge coordinates obtained by edge detection as features into a parameter fitting formula for fitting, circularly repeating for multiple times and correcting to obtain a group of optimal parameters, namely slope and intercept, and obtaining the predicted values of the left and right edge vertical coordinates by using the group of parameters; then comparing the obtained predicted value with the vertical coordinate data obtained by edge detection again, calculating loss, and eliminating abnormal data with large loss to obtain the processed workpiece edge data coordinate;
and (3) feature selection: on the basis of secondary data processing, selecting more than two lines of left and right edge detection data as basic characteristics of parameter calibration;
and (4) feature generation: performing mathematical operation processing on the selected basic feature data, circularly traversing each needle gauge image with each group of sizes, and performing splicing processing on features generated by each image to finally generate more than 10 groups of features;
and (5) performing parameter fitting again based on the finally generated more than 10 groups of characteristics, and calculating a group of fitting coefficients, the mean value and the standard deviation which are finally obtained.
Preferably, the specific step of obtaining the coordinates of the left and right edge points in step (1) is to perform pixel-level edge detection on the image by using a Canny edge detection operator to obtain the coordinate positions of the pixel edges of the image in the original image, then calculate the center position, and distinguish the detected left and right edge points, where a point smaller than the center position coordinate is a left edge point coordinate and a point larger than the center position coordinate is a right edge point coordinate.
Preferably, the step (2) of obtaining the optimal edge data coordinates of the workpiece includes: setting a parameter fitting equation of
Wherein x represents an argument, a represents a correlation coefficient, n represents a number,
representing dependent variables, wherein the independent variables and the dependent variables are selected edge point coordinates (x, y) in the fitting process; that is, the coordinates of the left and right edges obtained by edge detection are taken as features and respectively introduced into a parameter fitting formula for fitting to obtain a predicted value, and the left and right edges can respectively obtain a set of optimal parameters, that is, the predicted value
At the same time, the obtained prediction data needs to be judged again, and the loss function
And (4) integrally sequencing the loss, and performing secondary elimination on the left and right edge coordinates with the loss quantile value of more than 99% or less than 1% to obtain the optimal workpiece edge data coordinate.
Preferably, the operation method of the mathematical operation process in the step (4) includes: and performing difference calculation, square calculation, rounding-up calculation or normalization calculation on the data on the left side and the data on the right side respectively.
Preferably, the specific steps of step 4 include: setting a maximum error, placing a workpiece to be measured on a measuring platform for measurement, moving left and right to different positions for measurement, checking whether the accuracy requirement is met, if the error of the measurement result is smaller than the maximum error, considering that the accuracy requirement is met, and executing the step 5; and if the accuracy requirement is not met, turning to the step 2-3, re-acquiring the image data of the standard workpiece, and performing parameter calibration again.
The invention has the beneficial effects that:
according to the invention, through acquiring the image of the standard device and adopting the scheme of calibrating the internal parameters of the device by a fitting method, parameter calibration is completed on each measuring device before delivery, and meanwhile, through multiple abnormal point removal, accurate data is fitted to solve the parameters, so that the accuracy and precision can be ensured. The method has the advantages of complete parameters, strong universality and high efficiency, and can be applied to the factory parameter calibration of various industrial equipment.
In the parameter fitting and multiple abnormal point eliminating processes, whether the initialization of the parameters is successful or not is not required to be confirmed in each step, and whether the precision meets the requirement or not is only required to be tested for the last time, so that the parameters are more comprehensive, and the process is simpler.
Drawings
FIG. 1 is a schematic flow diagram of the present invention.
Detailed Description
The present invention is further described below with reference to the drawings and examples so that those skilled in the art can easily practice the present invention.
As shown in fig. 1, the present invention provides a two-dimensional measuring instrument calibration method, which comprises the following specific steps:
step 1, setting equipment camera parameters
And the user sets camera parameters on the detection equipment according to requirements.
Step 2, acquiring standard workpiece image data
Acquiring multiple groups of images of a workpiece to be measured at different positions respectively, acquiring the next size image data after acquiring enough quantity, and acquiring image data sets with different sizes continuously and repeatedly.
Step 3, image processing and parameter calibration
After edge processing is carried out on the acquired image data, a plurality of groups of features are generated, and processing is carried out by using machine learning algorithms such as parameter fitting and the like to obtain factory calibration parameters.
Step 4, parameter verification
Taking a workpiece to verify on a measuring platform, observing whether the workpiece can meet the precision requirement or not, namely the error is within a set range, and if the workpiece meets the specification parameter calibration success, carrying out the next step; if not, the step 2-3 is carried out, the image data of the standard workpiece is obtained again, and the calibration is carried out again.
Step 5, updating parameters
And updating the parameters meeting the precision requirement to the measurement model.
Example (b):
the embodiment provides a calibration method of a pin gauge, which is mainly used for measuring the size of a workpiece. The parameters to be calibrated are fitting coefficients of the measured diameter in the device and mean and standard deviation parameters reflecting the image quality. The present embodiment is not limited to the calibration with a pin gauge, and the method can be used for workpieces with the accuracy of about ± 1 micron. The method comprises the following specific steps:
step 1, setting equipment camera parameters
In this embodiment, the inspection equipment is a vision acquisition platform of the measuring instrument, the measuring instrument irradiates green two-dimensional parallel light by the emitter according to the optical transmission principle, and the shadow of the target object is captured by the CMOS of the receiver, so as to acquire the picture of the measured object. Firstly, a user can set camera parameters such as exposure time, light source brightness and the like according to requirements; if not, the initialization parameters built in the measurement model are adopted.
Step 2, acquiring standard workpiece image data
And acquiring image data on a visual acquisition platform of the measuring instrument. The method comprises the specific steps of placing a needle gauge in an effective measuring range of a camera, and moving to different positions to obtain a plurality of pictures. The needle gauges with different sizes respectively collect a group of pictures, and collected images of the needle gauges with different sizes are stored in the same directory to facilitate program reading so as to perform image processing and parameter calibration.
Step 3, image processing and parameter calibration
And in order to obtain the parameters to be calibrated, carrying out image processing on the collected multiple groups of needle gauge pictures. The method comprises the following specific steps:
step (1), edge detection: and carrying out pixel-level edge detection on the image by using a Canny edge detection operator to obtain the coordinate position of the pixel edge of the image in the original image. And then, calculating the central position, and distinguishing the detected left and right edge points, wherein the point smaller than the central position coordinate is the left edge point coordinate, and the point larger than the central position coordinate is the right edge point coordinate. And simultaneously removing abnormal points such as burrs and the like in the image. The number of the left and right edge points of the symmetrical pattern such as the needle gauge should be the same, i.e., num _ lefti＝num_rightiAnd eliminating abnormal points by judging whether the number of pixels in each line is symmetrical or not, wherein num _ leftiIndicates the number of ith row pixels on the left edge, num _ rightiAnd the number of ith rows of pixel points on the left edge is represented.
And (2) parameter fitting: setting a parameter fitting equation of
Wherein x represents an argument, a represents a correlation coefficient, n represents a number,
and representing dependent variables, wherein the independent variables and the dependent variables in the fitting process are selected edge point coordinates (x, y). That is, the coordinates of the left and right edges of the needle gauge obtained by edge detection are taken as features and respectively introduced into a parameter fitting formula for fitting, and finally a predicted value is obtained, and simultaneously, the left and right edges can respectively obtain a group of optimal parameters, namely [ a ]1,a2,a3,…an,b]. At the same time, the obtained prediction data needs to be judged again, and the loss function
And (4) integrally sequencing the loss, and performing secondary elimination on the left and right edge coordinates with the loss quantile value of more than 99% or less than 1% to obtain the optimal workpiece edge data coordinate.
And (3) feature selection: on the basis of secondary data processing, a plurality of rows of edge detection data are selected from the left and the right respectively and used as basic features of parameter calibration.
And (4) feature generation: and performing mathematical operation processing on the selected basic characteristic data, such as performing various operations of difference calculation, square calculation, upward rounding, normalization and the like on the data on the left side and the right side respectively to obtain multiple groups of data characteristics. And circularly traversing each needle gauge image with each group of sizes, and splicing the features generated by each image to finally generate 18 groups of features.
And (5) calculating again based on the finally generated 18 groups of characteristics, and finally obtaining a group of fitting coefficients, a mean value and a standard deviation for correcting the result in the test process.
Step 4, parameter verification
And setting a maximum error, taking a to-be-measured needle gauge, placing the to-be-measured needle gauge on a measuring platform for measurement, moving the to-be-measured needle gauge to different positions left and right to measure, and checking whether the accuracy requirement is met. If the accuracy requirement is not met, the step 2-3 is carried out if the accuracy requirement is not met, the image data of the standard workpiece is obtained again, and the parameter calibration is carried out again; if the error of the needle gauge measurement result is smaller than the maximum error, the accuracy requirement is considered to be met, and the step 5 can be executed.
Step 5, updating parameters
And updating the initialization parameters in the measurement model after the factory calibration parameters under the camera parameters set by the user are obtained and the accuracy requirements are met.
Before the calibration is carried out, the system initialization parameters are adopted to measure the needle gauges with different sizes, and the integral error is about 18 microns compared with the true value of the needle gauge.
By adopting the embodiment, the maximum error is set to be 4um during parameter verification, the measurement result of the 5mm needle gauge is taken to be 5.5mm, if the error is larger, the accuracy requirement is not met, if the error is larger, the step 2-3 is carried out, the image data of the standard workpiece is obtained again, and the parameter calibration is carried out again; if the measurement result of the 5mm needle gauge is 4.999mm and 5.001mm, and the error is less than 4um, the accuracy requirement is considered to be met, the parameter updating can be executed, and the measurement model can be modified. The test device in this verification process is not limited to a pin gauge, and can be any workpiece with a well-known dimension, for example; a cube block with a side length of 2 cm and the like can be used for judging whether the parameters meet the precision requirement.
The invention provides a method for fitting multiple groups of features to obtain calibration parameters aiming at the difference problem of a two-dimensional measuring instrument.
The above description is only for the purpose of illustrating preferred embodiments of the present invention and is not to be construed as limiting the present invention, and it is apparent to those skilled in the art that various modifications and variations can be made in the present invention. All changes, equivalents, modifications and the like which come within the scope of the invention as defined by the appended claims are intended to be embraced therein.
Claims (8)
1. A two-dimensional measuring instrument calibration method is characterized by comprising the following steps:
step 1, setting parameters of a transmitter and a receiver on detection equipment for acquiring an image of an object to be measured;
step 2, sequentially acquiring images of the workpiece to be measured at different positions, and acquiring an image data set with enough number and size;
step 3, carrying out edge detection processing on the acquired image data set to obtain profile data of the measured object, and processing the profile data by using a machine learning algorithm to obtain a fitting coefficient of the size of the measured workpiece and a mean value and standard deviation parameter which embody the image quality;
step 4, verifying on the measuring platform, judging whether the precision requirement meets the set range, and if the precision requirement meets the set description parameter, successfully calibrating; if not, turning to the step 2-3, re-acquiring the image data of the standard workpiece, and calibrating again;
and 5, updating the parameters meeting the precision requirement to the measurement model.
2. The calibration method of the two-dimensional measuring instrument according to claim 1, wherein the detection device in step 1 is a visual acquisition platform of the measuring instrument, the measuring instrument uses an optical transmission principle, a green two-dimensional parallel light is irradiated by the emitter, a CMOS of the receiver captures a shadow of the target object, and an image of the measured object is acquired.
3. The two-dimensional measuring instrument calibration method according to claim 1, wherein the specific method of the step 2 is as follows: and acquiring image data of a workpiece to be measured in the next size after acquiring enough images of a certain workpiece to be measured in different positions respectively to form image data sets of workpieces to be measured in different sizes.
4. The two-dimensional measuring instrument calibration method according to claim 1, wherein the specific steps of the step 3 comprise:
step (1), edge detection: carrying out pixel-level edge detection on the image by using a Canny edge detection operator to obtain coordinates of left and right edge points of the pixel of the image, and simultaneously removing coordinates of abnormal points of the image;
and (2) parameter fitting: respectively substituting left and right edge coordinates obtained by edge detection as features into a parameter fitting formula for fitting to respectively obtain a group of optimal parameters, namely slope and intercept, and obtaining predicted values of left and right edge vertical coordinates by using the group of parameters; then comparing the obtained predicted value with the vertical coordinate data obtained by edge detection again, calculating loss, and eliminating abnormal data with large loss to obtain the processed workpiece edge data coordinate;
and (3) feature selection: on the basis of secondary data processing, selecting more than two lines of left and right edge detection data as basic characteristics of parameter calibration;
and (4) feature generation: performing mathematical operation processing on the selected basic feature data, circularly traversing each needle gauge image with each group of sizes, and performing splicing processing on features generated by each image to generate more than 10 final groups of features;
and (5) performing parameter fitting again based on the finally generated more than 10 groups of characteristics, and calculating a group of fitting coefficients, the mean value and the standard deviation which are finally obtained.
5. The two-dimensional measuring instrument calibration method according to claim 4, wherein the step (1) of obtaining the coordinates of the left and right edge points comprises performing pixel-level edge detection on the image by using a Canny edge detection operator to obtain the coordinate position of the pixel edge of the image in the original image, then calculating the center position, and distinguishing the detected left and right edge points, wherein the point smaller than the center position coordinate is the left edge point coordinate, and the point larger than the center position coordinate is the right edge point coordinate.
6. The two-dimensional measuring instrument calibration method according to claim 4, wherein the step (2) of obtaining the optimal edge data coordinates of the workpiece comprises the following steps: setting a parameter fitting equation of
Wherein x represents an argument, a represents a correlation coefficient, n represents a number,
representing dependent variables, wherein the independent variables and the dependent variables are selected edge point coordinates (x, y) in the fitting process; that is, the coordinates of the left and right edges obtained by edge detection are taken as features and respectively introduced into a parameter fitting formula for fitting, and finally a predicted value is obtained, and the left and right edges can respectively obtain a group of optimal parameters, namely [ a ]1,a2,a3,…an,b](ii) a At the same time, the obtained prediction data needs to be judged again, and the loss function
And (4) integrally sequencing the loss, and performing secondary elimination on the left and right edge coordinates with the loss quantile value of more than 99% or less than 1% to obtain the optimal workpiece edge data coordinate.
7. The two-dimensional measuring instrument calibration method according to claim 4, wherein the operation method of the mathematical operation process in the step (4) includes: and performing difference calculation, square calculation, rounding-up calculation or normalization calculation on the data on the left side and the data on the right side respectively.
8. The two-dimensional measuring instrument calibration method according to claim 1, wherein the specific steps of the step 4 comprise: setting a maximum error, placing a workpiece to be measured on a measuring platform for measurement, moving left and right to different positions for measurement, checking whether the accuracy requirement is met, if the error of the measurement result is smaller than the maximum error, considering that the accuracy requirement is met, and executing the step 5; and if the accuracy requirement is not met, turning to the step 2-3, re-acquiring the image data of the standard workpiece, and performing parameter calibration again.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111336844.XA CN114092529A (en) | 2021-11-12 | 2021-11-12 | Calibration method of two-dimensional measuring instrument |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111336844.XA CN114092529A (en) | 2021-11-12 | 2021-11-12 | Calibration method of two-dimensional measuring instrument |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114092529A true CN114092529A (en) | 2022-02-25 |
Family
ID=80300096
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111336844.XA Pending CN114092529A (en) | 2021-11-12 | 2021-11-12 | Calibration method of two-dimensional measuring instrument |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114092529A (en) |
-
2021
- 2021-11-12 CN CN202111336844.XA patent/CN114092529A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110689579B (en) | Rapid monocular vision pose measurement method and measurement system based on cooperative target | |
| CN108230397B (en) | Multi-view camera calibration and correction method and apparatus, device, program and medium | |
| JP2000241120A (en) | Measuring apparatus | |
| CN115810041B (en) | Cable size detection method and system based on multi-camera vision | |
| CN114420619B (en) | Wafer alignment device and method, and wafer testing method | |
| CN116433780B (en) | Automatic calibration method for laser structured light based on machine vision | |
| CN116503316A (en) | Chip defect measurement method and system based on image processing | |
| CN104103543B (en) | Wafer defect dimension correction method | |
| CN116499362B (en) | Steel plate size online measurement system | |
| CN114092529A (en) | Calibration method of two-dimensional measuring instrument | |
| JP5136108B2 (en) | 3D shape measuring method and 3D shape measuring apparatus | |
| CN116563391A (en) | Automatic laser structure calibration method based on machine vision | |
| US11445162B2 (en) | Method and device for calibrating binocular camera | |
| CN113989513A (en) | Method for recognizing reading of square pointer type instrument | |
| TWI742391B (en) | Three-dimensional image surface defect detection system | |
| US20070217675A1 (en) | Z-axis optical detection of mechanical feature height | |
| CN113012121A (en) | Method and device for processing bare chip scanning result, electronic equipment and storage medium | |
| TW202113303A (en) | A method for inspecting a skeleton wafer | |
| CN113063353B (en) | Coordinate system establishment method, detection device, detection equipment and storage medium | |
| US8027528B2 (en) | Method for calculating height of chuck top and program storage medium for storing same method | |
| CN109449096B (en) | Method for identifying and detecting wafer chip | |
| CN114782436B (en) | PCB defect detection method, circuit board metering measurement method and manufacturing method | |
| CN117853429A (en) | Calibration image quality evaluation method | |
| CN115760856B (en) | Image recognition-based part spacing measurement method, system and storage medium | |
| CN111601100B (en) | TV distortion calculation method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |