A method for evaluating the precision of a vision inspection system

Abstract

This invention discloses a method for evaluating the accuracy of a visual inspection system. A calibration plate with marked points is fixed within the field of view of a visual sensor. A photogrammetric system II establishes a three-dimensional control field II. A plane Q is fitted using each marked point. The transformation relationship between the visual sensor coordinate system and the three-dimensional control field II is determined using the photogrammetric system II. The visual sensor acquires a two-dimensional image of the calibration plate to obtain the pixel coordinates of the marked points. A spatial straight line is constructed using the origin of the camera coordinate system and the pixel coordinates of the marked points. The spatial straight line is transformed into the three-dimensional control field II using the transformation relationship, and its intersection with plane Q is calculated. The coordinates of the corresponding marked points in the three-dimensional control field II are found. The difference between the found coordinates and the intersection coordinates is calculated. If the difference is less than a threshold, the accuracy of the visual inspection system meets the requirements; otherwise, it does not. This method is applicable to various types of visual inspection systems and has high universality.

Description

Technical Field

[0001] This invention relates to the field of accuracy verification, and more specifically to a method for evaluating the accuracy of a visual inspection system. Background Technology

[0002] Currently, visual inspection systems have been widely used in the processing and manufacturing field. In order to accurately control product quality during the product inspection process, different levels of requirements are placed on the inspection accuracy of visual inspection systems, especially in the precision manufacturing industry, where the requirements for inspection accuracy are even more stringent.

[0003] Therefore, accurately verifying and evaluating the detection accuracy of a vision inspection system has become an urgent technical problem. Existing accuracy evaluation methods typically involve the vision inspection system acquiring geometric features on the object being measured, such as holes and corners, calculating the measured coordinates of these features, and then using the theoretical coordinates of these features in the digital model of the object as the true values ​​to evaluate the error of the measured coordinates. However, theoretical coordinates only exist at specific locations (holes, corners) on the digital model of the object. In locations without significant feature variations, such as any point on a plane, theoretical coordinates cannot be found. Therefore, when a vision inspection system is used to acquire data from such locations where theoretical coordinates cannot be found, its detection accuracy cannot be evaluated. Summary of the Invention

[0004] To address the aforementioned problems, this invention proposes a method for evaluating the accuracy of a visual inspection system, which is applicable to various types of visual inspection systems and features ingenious design, high accuracy, and high versatility.

[0005] A method for evaluating the accuracy of a visual inspection system, utilizing a photogrammetric system II, verifies the accuracy of the visual inspection system through the following steps:

[0006] 1) Fix a calibration plate within the field of view of the vision sensor in the vision inspection system. The calibration plate is a flat plate with at least 3 marking points on it. Fix coding points and a scale around the calibration plate.

[0007] Photogrammetric System II acquires images of coded points, scales, and marker points from different poses to establish a three-dimensional control field II;

[0008] The spatial plane is fitted using the coordinates of each marker point in the three-dimensional control field II, and is denoted as plane Q.

[0009] 2) The transformation relationship between the visual sensor coordinate system and the three-dimensional control field II was determined using the photogrammetry system II calibration.

[0010] 3) The vision sensor acquires a two-dimensional image of the calibration board and obtains the pixel coordinates of the marker points in the two-dimensional image; a spatial straight line is constructed using the origin of the camera coordinate system in the vision sensor and the pixel coordinates of the marker points;

[0011] 4) Using the transformation relationship between the visual sensor coordinate system and the three-dimensional control field II, the spatial line is transformed into the three-dimensional control field II, and the intersection point with the plane Q is calculated; the coordinates of the marker point corresponding to the spatial line in the three-dimensional control field II are found, and the difference between the found coordinates and the intersection point coordinates is calculated. If the difference is less than the threshold, the accuracy of the visual detection system meets the requirements; otherwise, it does not meet the requirements.

[0012] Further, in step 2), the method of calibrating the transformation relationship between the visual sensor coordinate system and the three-dimensional control field II using the photogrammetric system II includes the following two methods:

[0013] Method 1: The vision sensor is installed at the end of the robot; a photogrammetry system target ball is installed at the end of the robot, the robot pose is adjusted, and the photogrammetry system II collects the target ball coordinates under different poses; based on the coordinates of the photogrammetry system target ball in the robot base coordinate system and the coordinates in the three-dimensional control field II under different poses, the transformation relationship between the robot base coordinate system and the three-dimensional control field II is calculated, and then combined with the transformation relationship between the vision sensor coordinate system and the robot base coordinate system obtained from hand-eye calibration and robot pose, the transformation relationship between the vision sensor coordinate system and the three-dimensional control field II is obtained;

[0014] Method 2: The vision sensor is fixed in the detection station. The vision sensor captures an image of the calibration plate and obtains the coordinates of the marker points in the image. Based on the coordinates of the marker points in the image and their coordinates in the three-dimensional control field II, the transformation relationship between the vision sensor coordinate system and the three-dimensional control field II is calculated.

[0015] Furthermore, when calibrating the external parameters of the vision inspection system using a laser tracker before its use, step 2), the method for determining the transformation relationship between the vision sensor coordinate system and the three-dimensional control field II using the photogrammetry system II is as follows:

[0016] At least four ball mounts are fixed around the calibration plate, the ball mounts being capable of mounting laser tracker target balls and photogrammetry target balls;

[0017] A photogrammetric target sphere is mounted on a pedestal. Photogrammetric system II acquires the coordinates of the photogrammetric target sphere in the three-dimensional control field II and stores them in point set A.

[0018] Replace the photogrammetric target ball on the spherical stand with the laser tracker target ball. The laser tracker acquires the coordinates of each laser tracker target ball in the tracker coordinate system and stores them in point set B.

[0019] Using the coordinates of each point in point set A and point set B, the transformation relationship between the three-dimensional control field II and the laser tracker coordinate system is calculated.

[0020] Furthermore, this is combined with the transformation relationship between the visual sensor coordinate system and the laser tracker coordinate system, which was previously calibrated using a laser tracker;

[0021] The transformation relationship between the visual sensor coordinate system and the three-dimensional control field II was derived.

[0022] Furthermore, the method for pre-calibrating the transformation relationship between the vision sensor coordinate system and the laser tracker coordinate system using a laser tracker is as follows:

[0023] The vision sensor is installed at the end of the robot, and a laser tracker target ball is installed at the end of the robot. The robot's pose is adjusted, and the laser tracker collects the coordinates of the target ball under different poses. Based on the coordinates of the laser tracker target ball in the robot's base coordinate system and the laser tracker's coordinate system under different poses, the transformation relationship between the robot's base coordinate system and the laser tracker's coordinate system is calculated. Then, combined with hand-eye calibration and robot pose, the transformation relationship between the vision sensor's coordinate system and the robot's base coordinate system is obtained, and the transformation relationship between the sensor's coordinate system and the laser tracker's coordinate system is obtained.

[0024] Furthermore, when the external parameters are calibrated using photogrammetry system I before the visual inspection system is used, step 2) involves using photogrammetry system II to determine the transformation relationship between the visual sensor coordinate system and the three-dimensional control field II.

[0025] Photogrammetric System I and Photogrammetric System II respectively acquire images of coded points, scales, and marker points to establish three-dimensional control field I and three-dimensional control field II;

[0026] Store the coordinates of each marker point in the three-dimensional control field I into point set I;

[0027] Store the coordinates of each marker point in the three-dimensional control field II into point set II;

[0028] Using the coordinates of each point in point set I and point set II, the transformation relationship between three-dimensional control field I and three-dimensional control field II is calculated.

[0029] Furthermore, this is combined with the transformation relationship between the visual sensor coordinate system and the three-dimensional control field I, which was previously calibrated using the photogrammetry system I;

[0030] The transformation relationship between the visual sensor coordinate system and the three-dimensional control field II was derived.

[0031] Furthermore, the methods for pre-calibrating the transformation relationship between the visual sensor coordinate system and the three-dimensional control field I using the photogrammetric system I include the following two:

[0032] Method A: The vision sensor is installed at the end of the robot; a target ball of a photogrammetry system is installed at the end of the robot, the robot pose is adjusted, and the photogrammetry system I collects the coordinates of the target ball under different poses; based on the coordinates of the target ball of the photogrammetry system in the robot base coordinate system and the coordinates in the three-dimensional control field I under different poses, the transformation relationship between the robot base coordinate system and the three-dimensional control field I is calculated, and then combined with the transformation relationship between the vision sensor coordinate system and the robot base coordinate system obtained from hand-eye calibration and robot pose, the transformation relationship between the sensor coordinate system and the three-dimensional control field I is obtained.

[0033] Method B: The vision sensor is fixed in the detection station. The vision sensor captures an image of the calibration plate and obtains the coordinates of the marker points in the image. Based on the coordinates of the marker points in the image and their coordinates in the three-dimensional control field I, the transformation relationship between the vision sensor coordinate system and the three-dimensional control field I is calculated.

[0034] To more accurately assess the precision, preferably, in step 3), the pixel coordinates of multiple marker points are obtained in the two-dimensional image; spatial lines are constructed using the origin of the camera coordinate system in the vision sensor and the pixel coordinates of each marker point, resulting in multiple spatial lines.

[0035] In step 4), the transformation relationship between the visual sensor coordinate system and the three-dimensional control field II is used to transform each spatial line into the three-dimensional control field, and the intersection points of each transformed spatial line and plane Q are obtained.

[0036] Find the coordinates of the marker point corresponding to each spatial line in the three-dimensional control field, and use the difference between the found coordinates and the corresponding intersection point coordinates;

[0037] The difference is taken as the mean or standard deviation. If the resulting value is less than the threshold, the accuracy of the visual inspection system meets the requirements; otherwise, it does not.

[0038] Preferably, in step 3), in the two-dimensional image, the edges of the marker points are extracted and edge fitting is performed using the Canny method or the Sobel method, and the geometric center coordinates of the marker points are obtained as the pixel coordinates of the marker points.

[0039] Preferably, step 1) uses the RANSAC method or the least squares method to fit the plane Q.

[0040] Preferably, there are 50 to 300 markers scattered on the calibration plate; the threshold value is 0.5 mm to 5 mm.

[0041] This method has the following characteristics:

[0042] The method of this invention uses a photogrammetry system II to determine the true coordinates of the marked points on the calibration plate; a visual inspection system uses the intersection of light rays to obtain the measured coordinates of the marked points; and the accuracy of the visual inspection system is evaluated by combining the deviation between the measured coordinates and the true coordinates. The whole scheme is simple, effective, and highly implementable.

[0043] This method has no requirements on the type of vision inspection system. When the vision inspection system is used to inspect any point on the object under test (without theoretical numerical coordinates), this scheme can still effectively evaluate the accuracy of the vision inspection system and provide data support for the normal use of the vision inspection system.

[0044] This method can not only evaluate the detection accuracy of visual inspection sensors, but also evaluate the overall detection accuracy of visual inspection sensors and calibration devices (laser trackers / photogrammetric systems), taking calibration errors into account; it has the characteristics of wide applicability, strong universality and ingenious design. Attached Figure Description

[0045] Figure 1 This is a schematic diagram showing the positional relationship between the robot, laser tracker, photogrammetry system II, and calibration plate in Example 2. Detailed Implementation

[0046] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and embodiments.

[0047] Example 1

[0048] A method for evaluating the accuracy of a visual inspection system is provided. In this embodiment, only the detection accuracy of the visual inspection system is evaluated. The visual inspection system includes a visual sensor. The visual sensor is installed in the inspection station and is used to detect measurement point information on the object being measured.

[0049] The plan is as follows:

[0050] A method for evaluating the accuracy of a visual inspection system, utilizing a photogrammetric system II, verifies the accuracy of the visual inspection system through the following steps:

[0051] 1) Fix a calibration plate within the field of view of the vision sensor. The calibration plate is a flat plate with at least 3 marking points (preferably 50 to 300, which are scattered on the calibration plate); fix coding points and a scale around the calibration plate.

[0052] Photogrammetric System II acquires images of coded points, scales, and marker points from different poses to establish a three-dimensional control field II;

[0053] The process of establishing the three-dimensional control field of the photogrammetry system is based on the equipment instruction manual and methods already disclosed in the existing technology, and is constructed using existing solutions.

[0054] The spatial plane is fitted using the coordinates of each marker point in the three-dimensional control field II, and is denoted as plane Q.

[0055] Preferred implementation: The method for fitting the plane Q is either the RANSAC method or the least squares method.

[0056] 2) The transformation relationship between the visual sensor coordinate system and the three-dimensional control field II obtained by calibrating using the photogrammetric system II;

[0057] 3) The vision sensor acquires a two-dimensional image of the calibration board and obtains the pixel coordinates of the marked points in the two-dimensional image;

[0058] A spatial straight line is constructed using the origin (0,0,0) of the camera coordinate system in the visual sensor and the pixel coordinates of the marker point;

[0059] 4) Using the transformation relationship between the visual sensor coordinate system and the three-dimensional control field II, the spatial line is transformed into the three-dimensional control field II, and the intersection point with the plane Q is calculated; the coordinates of the marker point corresponding to the spatial line in the three-dimensional control field II are found, and the difference between the found coordinates and the intersection point coordinates is calculated. If the difference is less than the threshold, the accuracy of the visual detection system meets the requirements; otherwise, it does not meet the requirements.

[0060] If there are 10 marker points, number them 1-10 respectively; in step 3), construct a spatial straight line using the origin (0,0,0) of the camera coordinate system in the vision sensor and the pixel coordinates of marker point 3; in step 4), obtain the intersection coordinates, then find the coordinates of marker point 3 in the three-dimensional control field II, and use the difference between the found coordinates and the intersection coordinates.

[0061] The threshold is set according to the accuracy requirements of the actual detection scenario, and is generally 0.5mm to 5mm.

[0062] In detail, step 2), the method of calibrating the transformation relationship between the visual sensor coordinate system and the three-dimensional control field II using photogrammetry system II includes the following two methods:

[0063] Method 1: The vision sensor is installed at the end of the robot; a photogrammetry system target ball is installed at the end of the robot, the robot pose is adjusted, and the photogrammetry system II collects the target ball coordinates under different poses; based on the coordinates of the photogrammetry system target ball in the robot base coordinate system and the coordinates in the three-dimensional control field II under different poses, the transformation relationship between the robot base coordinate system and the three-dimensional control field II is calculated, and then combined with the transformation relationship between the vision sensor coordinate system and the robot base coordinate system obtained from hand-eye calibration and robot pose, the transformation relationship between the vision sensor coordinate system and the three-dimensional control field II is obtained;

[0064] Method 2: The vision sensor is fixed in the detection station. The vision sensor captures an image of the calibration plate and obtains the coordinates of the marker points in the image. Based on the coordinates of the marker points in the image and their coordinates in the three-dimensional control field II, the transformation relationship between the vision sensor coordinate system and the three-dimensional control field II is calculated.

[0065] Specifically, in step 3), in the two-dimensional image, the edges of the marker points are extracted and edge fitting is performed using the Canny method or the Sobel method, and the geometric center coordinates of the marker points are obtained as the pixel coordinates of the marker points.

[0066] To more accurately evaluate the precision of the visual inspection system, as a preferred implementation, in step 3), the pixel coordinates of multiple marker points are obtained in the two-dimensional image; spatial lines are constructed using the origin of the camera coordinate system in the visual sensor and the pixel coordinates of each marker point, resulting in multiple spatial lines.

[0067] In step 4), the transformation relationship between the visual sensor coordinate system and the three-dimensional control field II is used to transform each spatial line into the three-dimensional control field, and the intersection points of each transformed spatial line and plane Q are obtained.

[0068] Find the coordinates of the marker point corresponding to each spatial line in the three-dimensional control field, and use the difference between the found coordinates and the corresponding intersection point coordinates;

[0069] The difference is taken as the mean or standard deviation. If the resulting value is less than the threshold, the accuracy of the visual inspection system meets the requirements; otherwise, it does not.

[0070] Example 2

[0071] A method for evaluating the accuracy of a visual inspection system, wherein in this embodiment, the visual inspection system is calibrated with an external parameter using a laser tracker before use.

[0072] This embodiment is used to evaluate the detection error of the vision inspection system and the error introduced by the external parameter calibration process of the laser tracker.

[0073] The vision inspection system includes a vision sensor; the vision sensor is installed in the inspection station to detect measurement point information on the object being measured; a laser tracker is placed around the vision sensor to calibrate the transformation relationship between the vision sensor coordinate system and the object's coordinate system.

[0074] The plan is as follows:

[0075] A method for evaluating the accuracy of a visual inspection system, utilizing a photogrammetric system II, verifies the accuracy of the visual inspection system through the following steps:

[0076] 1) Fix a calibration plate within the field of view of the vision sensor in the vision inspection system, such as... Figure 1 As shown, the calibration plate 4 is a flat plate with at least 3 marking points 5 on it; coding points and a scale are fixed around the calibration plate;

[0077] Photogrammetry System II 3 acquires images of coded points, scales, and marker points from different poses to establish a three-dimensional control field II;

[0078] The spatial plane is fitted using the coordinates of each marker point in the three-dimensional control field II, and is denoted as plane Q.

[0079] 2) The transformation relationship between the visual sensor coordinate system and the three-dimensional control field II was determined using the photogrammetry system II calibration.

[0080] 3) The vision sensor acquires a two-dimensional image of the calibration board and obtains the pixel coordinates of the marker points in the two-dimensional image; a spatial straight line is constructed using the origin of the camera coordinate system in the vision sensor and the pixel coordinates of the marker points;

[0081] 4) Using the transformation relationship between the visual sensor coordinate system and the three-dimensional control field II, the spatial line is transformed into the three-dimensional control field II, and the intersection point with the plane Q is calculated; the coordinates of the marker point corresponding to the spatial line in the three-dimensional control field II are found, and the difference between the found coordinates and the intersection point coordinates is calculated. If the difference is less than the threshold, the accuracy of the visual detection system meets the requirements; otherwise, it does not meet the requirements.

[0082] In this embodiment, step 2), the method for determining the transformation relationship between the visual sensor coordinate system and the three-dimensional control field II using photogrammetry system II calibration, is as follows:

[0083] like Figure 1 As shown, at least four ball mounts 6 are fixed around the calibration plate. These ball mounts 6 are capable of mounting both laser tracker target balls and photogrammetric target balls; that is, the ball mounts are compatible with the mounting of both types of target balls. The mounting positions of the ball mounts 6 are not collinear.

[0084] Install a photogrammetric target ball on ball seat 6. Photogrammetric system II 3 acquires the coordinates of the photogrammetric target ball in the three-dimensional control field II and stores them in point set A.

[0085] Replace the photogrammetric target ball on spherical base 6 with the laser tracker target ball. The laser tracker 2 acquires the coordinates of each laser tracker target ball in the tracker coordinate system and stores them in point set B.

[0086] Using the coordinates of each point in point set A and point set B, the transformation relationship between the three-dimensional control field II and the laser tracker coordinate system is calculated. Figure 1 RT2 in the middle);

[0087] Furthermore, this is combined with the transformation relationship between the visual sensor coordinate system and the laser tracker coordinate system obtained in advance using laser tracker 2;

[0088] The transformation relationship between the visual sensor coordinate system and the three-dimensional control field II was derived.

[0089] Among these steps, the transformation relationship between the visual sensor coordinate system and the laser tracker coordinate system is pre-calibrated using a laser tracker. Figure 1 The method for RT1 in the middle is as follows:

[0090] The vision sensor is installed at the end of robot 1, and a laser tracker target ball is installed at the end of robot 1. The pose of robot 1 is adjusted, and the laser tracker 2 collects the coordinates of the target ball under different poses. Based on the coordinates of the laser tracker target ball in the robot's base coordinate system and the laser tracker's coordinate system under different poses, the transformation relationship between the robot's base coordinate system and the laser tracker's coordinate system is calculated. Then, combined with hand-eye calibration and robot pose, the transformation relationship between the vision sensor's coordinate system and the robot's base coordinate system is obtained, and the transformation relationship between the sensor's coordinate system and the laser tracker's coordinate system is obtained.

[0091] To more accurately assess the precision of the visual inspection system, this embodiment includes 300 marker points; the visual sensor is used to detect surface defects on the product.

[0092] In step 3), the pixel coordinates of 300 marker points are obtained in the two-dimensional image; spatial lines are constructed by using the origin of the camera coordinate system in the vision sensor and the pixel coordinates of the 300 marker points respectively, resulting in 300 spatial lines.

[0093] In step 4), the transformation relationship between the visual sensor coordinate system and the three-dimensional control field II is used to transform each spatial line into the three-dimensional control field, and the intersection points (300) of each transformed spatial line with plane Q are obtained.

[0094] Find the coordinates of the marker point corresponding to each spatial line in the three-dimensional control field, and use the difference between the found coordinates and the corresponding intersection point coordinates;

[0095] 300 differences will be obtained; the average of the differences is 0.9732 mm, and the standard deviation of the differences is 0.285 mm.

[0096] If the obtained value is less than the threshold of 2mm, the accuracy of the visual inspection system meets the requirements; otherwise, it does not.

[0097] Example 3

[0098] A method for evaluating the accuracy of a visual inspection system, wherein in this embodiment, the visual inspection system is calibrated using a photogrammetry system I before use.

[0099] This embodiment is used to evaluate the detection error of the visual inspection system and the error introduced by the external parameter calibration process of the photogrammetry system I.

[0100] The visual inspection system includes a visual sensor; the visual sensor is installed in the inspection station to detect measurement point information on the object being measured; the photogrammetry system I is placed around the visual sensor to calibrate the transformation relationship between the visual sensor coordinate system and the object's coordinate system.

[0101] The plan is as follows:

[0102] A method for evaluating the accuracy of a visual inspection system, utilizing a photogrammetric system II, verifies the accuracy of the visual inspection system through the following steps:

[0103] 1) Fix a calibration plate within the field of view of the vision sensor in the vision inspection system, such as... Figure 1 As shown, the calibration plate 4 is a flat plate with at least 3 marking points 5 on it; coding points and a scale are fixed around the calibration plate;

[0104] Photogrammetry System II 3 acquires images of coded points, scales, and marker points from different poses to establish a three-dimensional control field II;

[0105] The spatial plane is fitted using the coordinates of each marker point in the three-dimensional control field II, and is denoted as plane Q.

[0106] 2) The transformation relationship between the visual sensor coordinate system and the three-dimensional control field II was determined using the photogrammetry system II calibration.

[0107] 3) The vision sensor acquires a two-dimensional image of the calibration board and obtains the pixel coordinates of the marker points in the two-dimensional image; a spatial straight line is constructed using the origin of the camera coordinate system in the vision sensor and the pixel coordinates of the marker points;

[0108] 4) Using the transformation relationship between the visual sensor coordinate system and the three-dimensional control field II, the spatial line is transformed into the three-dimensional control field II, and the intersection point with the plane Q is calculated; the coordinates of the marker point corresponding to the spatial line in the three-dimensional control field II are found, and the difference between the found coordinates and the intersection point coordinates is calculated. If the difference is less than the threshold, the accuracy of the visual detection system meets the requirements; otherwise, it does not meet the requirements.

[0109] In this embodiment, step 2), the method for transforming the visual sensor coordinate system and the three-dimensional control field II obtained by photogrammetry system II calibration, is as follows:

[0110] Photogrammetric System I and Photogrammetric System II respectively acquire images of coded points, scales, and marker points to establish three-dimensional control field I and three-dimensional control field II;

[0111] Store the coordinates of each marker point in the three-dimensional control field I into point set I;

[0112] Store the coordinates of each marker point in the three-dimensional control field II into point set II;

[0113] Using the coordinates of each point in point set I and point set II, the transformation relationship between three-dimensional control field I and three-dimensional control field II is calculated.

[0114] Furthermore, this is combined with the transformation relationship between the visual sensor coordinate system and the three-dimensional control field I, which was previously calibrated using the photogrammetry system I;

[0115] The transformation relationship between the visual sensor coordinate system and the three-dimensional control field II was derived.

[0116] Among them, the methods for pre-calibrating the transformation relationship between the visual sensor coordinate system and the three-dimensional control field I using the photogrammetric system I include the following two:

[0117] Method A: The vision sensor is installed at the end of the robot; a target ball of a photogrammetry system is installed at the end of the robot, the robot pose is adjusted, and the photogrammetry system I collects the coordinates of the target ball under different poses; based on the coordinates of the target ball of the photogrammetry system in the robot base coordinate system and the coordinates in the three-dimensional control field I under different poses, the transformation relationship between the robot base coordinate system and the three-dimensional control field I is calculated, and then combined with the transformation relationship between the vision sensor coordinate system and the robot base coordinate system obtained from hand-eye calibration and robot pose, the transformation relationship between the sensor coordinate system and the three-dimensional control field I is obtained.

[0118] Method B: The vision sensor is fixed in the detection station. The vision sensor captures an image of the calibration plate and obtains the coordinates of the marker points in the image. Based on the coordinates of the marker points in the image and their coordinates in the three-dimensional control field I, the transformation relationship between the vision sensor coordinate system and the three-dimensional control field I is calculated.

[0119] This invention's method is not limited by the type of visual inspection system; single-lens, binocular, and structured light sensors can all be used to evaluate accuracy. When the visual sensor contains multiple cameras, each camera acquires images separately, and the errors are calculated separately. The accuracy of the visual inspection system is then comprehensively evaluated based on the error results. This method can be used for performance testing of visual inspection systems before they leave the factory, providing data support for the normal use of the visual inspection system.

[0120] The foregoing description of specific exemplary embodiments of the present invention is for illustrative and descriptive purposes. It is not intended to be exhaustive, nor to limit the invention to the precise forms disclosed; obviously, many changes and variations are possible in accordance with the foregoing teachings. The exemplary embodiments were chosen and described to explain the specific principles of the invention and its practical application, thereby enabling others skilled in the art to implement and utilize various exemplary embodiments of the invention, as well as their different alternatives and modifications. The scope of the invention is intended to be defined by the appended claims and their equivalents.

Claims

1. A method for evaluating the accuracy of a visual inspection system, characterized in that: Using photogrammetry system II, the accuracy of the visual inspection system is verified through the following steps: 1) Fix a calibration plate within the field of view of the vision sensor in the vision inspection system. The calibration plate is a flat plate with at least 3 marking points on it. Fix coding points and a scale around the calibration plate. Photogrammetric System II acquires images of coded points, scales, and marker points from different poses to establish a three-dimensional control field II; The spatial plane is fitted using the coordinates of each marker point in the three-dimensional control field II, and is denoted as plane Q. 2) Using either Method 1 or Method 2, the transformation relationship between the visual sensor coordinate system and the three-dimensional control field II is determined by calibrating the photogrammetry system II. Method 1: The vision sensor is installed at the end of the robot; a photogrammetry system target ball is installed at the end of the robot, the robot pose is adjusted, and the photogrammetry system II collects the target ball coordinates under different poses; based on the coordinates of the photogrammetry system target ball in the robot base coordinate system and the coordinates in the three-dimensional control field II under different poses, the transformation relationship between the robot base coordinate system and the three-dimensional control field II is calculated, and then combined with the transformation relationship between the vision sensor coordinate system and the robot base coordinate system obtained from hand-eye calibration and robot pose, the transformation relationship between the vision sensor coordinate system and the three-dimensional control field II is obtained; Method 2: The vision sensor is fixed in the detection station. The vision sensor captures an image of the calibration plate and obtains the coordinates of the marker points in the image. Based on the coordinates of the marker points in the image and their coordinates in the three-dimensional control field II, the transformation relationship between the vision sensor coordinate system and the three-dimensional control field II is calculated. 3) The vision sensor acquires a two-dimensional image of the calibration board and obtains the pixel coordinates of the marker points in the two-dimensional image; a spatial straight line is constructed using the origin of the camera coordinate system in the vision sensor and the pixel coordinates of the marker points; 4) Using the transformation relationship between the visual sensor coordinate system and the three-dimensional control field II, the spatial straight line is transformed into the three-dimensional control field II, and its intersection with plane Q is calculated; Find the coordinates of the marker point corresponding to the spatial straight line in the three-dimensional control field II. Subtract the found coordinates from the coordinates of the intersection point. If the difference is less than the threshold, the accuracy of the visual inspection system meets the requirements; otherwise, it does not.

2. The accuracy evaluation method for the visual inspection system as described in claim 1, characterized in that: When calibrating the external parameters of the vision inspection system using a laser tracker before its use, step 2) involves using the photogrammetry system II to determine the transformation relationship between the vision sensor coordinate system and the three-dimensional control field II. At least four ball mounts are fixed around the calibration plate, the ball mounts being capable of mounting laser tracker target balls and photogrammetry target balls; A photogrammetric target sphere is mounted on a pedestal. Photogrammetric system II acquires the coordinates of the photogrammetric target sphere in the three-dimensional control field II and stores them in point set A. Replace the photogrammetric target ball on the spherical stand with the laser tracker target ball. The laser tracker acquires the coordinates of each laser tracker target ball in the tracker coordinate system and stores them in point set B. Using the coordinates of each point in point set A and point set B, the transformation relationship between the three-dimensional control field II and the laser tracker coordinate system is calculated. Furthermore, this is combined with the transformation relationship between the visual sensor coordinate system and the laser tracker coordinate system, which was previously calibrated using a laser tracker; The transformation relationship between the visual sensor coordinate system and the three-dimensional control field II was derived.

3. The accuracy evaluation method for the visual inspection system as described in claim 2, characterized in that: The method for pre-calibrating the transformation relationship between the vision sensor coordinate system and the laser tracker coordinate system using a laser tracker is as follows: The vision sensor is installed at the end of the robot, and a laser tracker target ball is installed at the end of the robot. The robot's pose is adjusted, and the laser tracker collects the coordinates of the target ball under different poses. Based on the coordinates of the laser tracker target ball in the robot's base coordinate system and the laser tracker's coordinate system under different poses, the transformation relationship between the robot's base coordinate system and the laser tracker's coordinate system is calculated. Then, combined with hand-eye calibration and robot pose, the transformation relationship between the vision sensor's coordinate system and the robot's base coordinate system is obtained, and the transformation relationship between the sensor's coordinate system and the laser tracker's coordinate system is obtained.

4. The accuracy evaluation method for the visual inspection system as described in claim 1, characterized in that: When the external parameters of the vision inspection system are calibrated using photogrammetry system I before use, step 2) involves using the transformation relationship between the vision sensor coordinate system and the three-dimensional control field II obtained from the photogrammetry system II calibration as follows: Photogrammetric System I and Photogrammetric System II respectively acquire images of coded points, scales, and marker points to establish three-dimensional control field I and three-dimensional control field II; Store the coordinates of each marker point in the three-dimensional control field I into point set I; Store the coordinates of each marker point in the three-dimensional control field II into point set II; Using the coordinates of each point in point set I and point set II, the transformation relationship between three-dimensional control field I and three-dimensional control field II is calculated. Furthermore, this is combined with the transformation relationship between the visual sensor coordinate system and the three-dimensional control field I, which was previously calibrated using the photogrammetry system I; The transformation relationship between the visual sensor coordinate system and the three-dimensional control field II was derived.

5. The accuracy evaluation method for the visual inspection system as described in claim 4, characterized in that: The methods for pre-calibrating the transformation relationship between the visual sensor coordinate system and the three-dimensional control field I using the photogrammetric system I include the following two: Method A: The vision sensor is installed at the end of the robot; a target ball of a photogrammetry system is installed at the end of the robot, the robot pose is adjusted, and the photogrammetry system I collects the coordinates of the target ball under different poses; based on the coordinates of the target ball of the photogrammetry system in the robot base coordinate system and in the three-dimensional control field I under different poses, the transformation relationship between the robot base coordinate system and the three-dimensional control field I is calculated, and then combined with the transformation relationship between the vision sensor coordinate system and the robot base coordinate system obtained from hand-eye calibration and robot pose, the transformation relationship between the sensor coordinate system and the three-dimensional control field I is obtained; Method B: The vision sensor is fixed in the detection station. The vision sensor captures an image of the calibration plate and obtains the coordinates of the marker points in the image. Based on the coordinates of the marker points in the image and their coordinates in the three-dimensional control field I, the transformation relationship between the vision sensor coordinate system and the three-dimensional control field I is calculated.

6. The accuracy evaluation method for the visual inspection system as described in any one of claims 1 to 5, characterized in that: In step 3), the pixel coordinates of multiple marker points are obtained in the two-dimensional image; spatial lines are constructed by using the origin of the camera coordinate system in the vision sensor and the pixel coordinates of each marker point to obtain multiple spatial lines. In step 4), the transformation relationship between the visual sensor coordinate system and the three-dimensional control field II is used to transform each spatial line into the three-dimensional control field, and the intersection points of each transformed spatial line and plane Q are obtained. Find the coordinates of the marker point corresponding to each spatial line in the three-dimensional control field, and use the difference between the found coordinates and the corresponding intersection point coordinates; The difference is taken as the mean or standard deviation. If the resulting value is less than the threshold, the accuracy of the visual inspection system meets the requirements; otherwise, it does not.

7. The accuracy evaluation method for the visual inspection system as described in any one of claims 1 to 5, characterized in that: Step 3): In the two-dimensional image, use the Canny method or the Sobel method to extract the edges of the marker points and perform edge fitting. The geometric center coordinates of the obtained marker points are then used as the pixel coordinates of the marker points.

8. The accuracy evaluation method for the visual inspection system as described in any one of claims 1 to 5, characterized in that: Step 1) The method for fitting the plane Q is either the RANSAC method or the least squares method.

9. The accuracy evaluation method for the visual inspection system according to any one of claims 1 to 5, characterized in that: There are 50 to 300 markers scattered on the calibration plate; the threshold value is 0.5 mm to 5 mm.

Images