A multi-class circle-shaped target against heart deflection and a binocular vision positioning method thereof

Abstract

The application discloses a multi-class circle special-shaped target resisting heart drift and a binocular vision positioning method thereof, and belongs to the technical field of industrial vision measurement and precision positioning. The target comprises a white background substrate and seven symmetrical black circles, wherein a large circle is arranged at the center, a standard normal small circle is arranged above and below the large circle respectively, and four specially-made class circles are arranged at four corners respectively. The specially-made class circles are designed according to a cylindrical surface nonlinear projection formula. The positioning method comprises the following steps: a binocular camera collects a target image and extracts two-dimensional circle center coordinates; three-dimensional point clouds are solved through stereo matching; a cylindrical space axial vector is extracted and a normal plane is constructed, and three-dimensional feature points are orthogonally projected to the normal plane for two-dimensional circle fitting; and the coordinates of an internal anchor cylinder center and a surface highest axial reference point are solved through coordinate inverse transformation. The application effectively overcomes positioning errors caused by cylindrical curvature deformation, space inclination and rotation around an axis, and significantly improves the positioning robustness and universality under complex working conditions.

Description

Technical Field

[0001] This invention relates to the field of industrial vision measurement and precision positioning technology, and in particular to a multi-type circular irregular target with anti-centroid offset and its binocular vision positioning method. Background Technology

[0002] Cylindrical or near-cylindrical industrial tools are widely used in applications such as industrial automation assembly, precision machining, end effector guidance, and measurement system accuracy calibration. Examples include tubular tooling, assembly actuators, and various cylindrical end components. To achieve spatial position detection, motion trajectory tracking, and assembly accuracy control of these tools during operation, a stereo vision measurement scheme based on a "visual target + binocular camera" is typically employed in industrial settings. This involves attaching a visual target to the surface of the cylindrical or near-cylindrical tool, acquiring target images, and calculating their three-dimensional spatial coordinates, thereby achieving tool position measurement and system calibration.

[0003] However, due to the structural characteristics of the object being measured, the complexity of imaging conditions, and the interference from multi-degree-of-freedom pose changes in actual operations, existing visual measurement solutions still face several key issues affecting measurement accuracy and stability in practical applications, mainly in the following aspects:

[0004] 1. Target projection deformation caused by cylindrical curved surface attachment

[0005] Existing visual targets mostly adopt planar circular or checkerboard structures. When they are attached to a cylindrical or near-cylindrical curved surface, the curvature causes the target to undergo significant nonlinear projection deformation in the camera's imaging plane. The originally regular geometric structure appears as an "ellipse-like" or irregular contour in the image, resulting in the degradation of the target's geometric features and thus reducing the accuracy of extracting two-dimensional feature points or the center of the circle.

[0006] 2. The problem of centroidal systemic shift caused by minute changes in tool orientation

[0007] In actual industrial operations or system calibration, cylindrical or quasi-cylindrical tools inevitably experience small-angle rotations or attitude disturbances, typically within ±5°. When the tool rotates, the grayscale distribution and contour shape of the visual target attached to its surface change in the image, causing a systematic shift in the target centroid calculated based on grayscale or geometric features towards the direction of rotation. This shift increases with the rotation angle, resulting in a significant error between the extracted two-dimensional centroid and the target's true geometric center.

[0008] 3. The problem of positioning reference point slippage caused by rotation around the axis.

[0009] During tool movement or continuous machining operations, a slight rotation (Roll) around its own central axis is often present. This rotation causes the visual target attached to the tool surface to physically slip in space. If the feature points on the surface target are directly used as the absolute reference points for continuous positioning or displacement measurement, the surface slip caused by the rotation will be seriously misjudged by the system as a spatial translation of the entire tool, resulting in the loss of distance measurement and positioning reference. Existing algorithms generally lack a mechanism to extract "absolutely fixed points" that are not affected by the rotation.

[0010] 4. Error superposition caused by lens distortion and field-of-view edge effects

[0011] In binocular vision measurement systems with large working distances and fields of view, the radial and tangential distortion effects of the lens are significantly enhanced when the target imaging position is close to the edge of the camera's field of view. Under the combined effects of cylindrical surface attachment and attitude changes, the distortion effect further amplifies the irregularity of the target contour, leading to a significant increase in the error of extracting the two-dimensional center or feature points, thus adversely affecting the subsequent three-dimensional spatial coordinate calculation and system calibration accuracy.

[0012] 5. Existing target structures are insufficient to simultaneously meet the requirements of high-precision measurement and calibration.

[0013] As industrial assembly precision and measurement system calibration requirements continue to increase, the permissible range of 3D spatial positioning errors for cylindrical or quasi-cylindrical tools is constantly decreasing, necessitating sufficient assembly and safety redundancy. However, due to factors such as surface deformation, centroid offset, and imaging distortion, existing planar target structures struggle to simultaneously meet the requirements of measurement accuracy, stability, and robustness in complex industrial environments. This results in 3D positioning results that cannot effectively support applications such as high-precision positioning, assembly guidance, and system calibration.

[0014] In summary, with the continuous improvement of industrial assembly precision and measurement system calibration requirements, more stringent "all-attitude and anti-interference" requirements have been placed on the three-dimensional spatial positioning of cylindrical or quasi-cylindrical tools. However, due to multiple factors such as target surface deformation, centroid offset, fitting failure caused by spatial tilt, and reference slippage caused by rotation around the axis, the existing "planar target design + traditional dimensionality reduction algorithm" is difficult to meet the application requirements of high precision and high robustness in complex industrial environments, becoming an industry technical bottleneck restricting the high-precision spatial positioning of cylindrical tools. Summary of the Invention

[0015] The purpose of this invention is to overcome the technical problems existing in the prior art and to provide a multi-type circular irregular target with anti-centroid offset and its binocular vision positioning method, so as to realize the high-precision three-dimensional positioning, measurement and system calibration of key spatial feature points of cylindrical or quasi-cylindrical tools.

[0016] The objective of this invention is achieved through the following technical solution:

[0017] A first aspect of the present invention provides a multi-type circular irregular target with centroidal displacement resistance, comprising a white background substrate and seven black circles disposed on the white background substrate; the seven black circles are symmetrically distributed about the center position of the white background substrate, wherein:

[0018] A large central circle is set at the center of a white background substrate to serve as a reference for various types of irregularly shaped circular targets;

[0019] Two standard small circles are set at the top and bottom of a white background substrate;

[0020] A specially designed small circle is set in the upper left, lower left, upper right, and lower right directions of the white background substrate;

[0021] The outline of the specially designed small circle is determined by the following nonlinear projection formula:

[0022]

[0023] in, Let v be the horizontal coordinate of any point on the unfolded plane, v be the vertical coordinate of any point on the unfolded plane, r be the radius of the standard circle, α be the offset angle of the target relative to the cylindrical or cylindrical tool surface, and R be the cylindrical radius of the cylindrical or cylindrical tool.

[0024] In some embodiments, the white background substrate is made of polyethylene terephthalate with a thickness of 0.1mm-0.2mm, a size of 75-100mm×90-120mm, and a grayscale value ≥230; the seven black circles are all UV printed with a grayscale value ≤20 and a grayscale difference ≥210 with the white background substrate.

[0025] In some embodiments, the diameter of the central large circle is 12mm-18mm, and the diameters of the standard small circle and the specially made small circle are both 10mm-15mm; the offset angle α ranges from 0° to ±20°, and the cylinder radius R ranges from 50mm to 100mm.

[0026] A second aspect of the present invention provides a binocular vision positioning method for cylindrical or quasi-cylindrical tools based on any of the multiple types of circular irregular targets described in the first aspect, comprising the following steps:

[0027] S1. Simultaneously acquire left and right views of various circular irregular targets on the cylindrical surface of a cylindrical or cylindrical tool using a binocular camera;

[0028] S2. Perform adaptive threshold segmentation and circular outline filtering on the left and right views respectively to extract the two-dimensional center coordinates of the seven black circles;

[0029] S3. Number the seven black circles based on the central large circle, match the circles with the same number in the left and right views, and calculate the three-dimensional spatial coordinates of each circle through binocular vision epipolar constraints and triangulation principles.

[0030] S4. Based on the three-dimensional spatial coordinates of the central great circle and the standard small circles in the upward and downward directions, calculate the initial axial direction vector of the cylinder in space, and construct a spatial normal plane perpendicular to the initial axial direction vector.

[0031] S5. Project the three-dimensional spatial coordinates of the seven black circles onto the spatial normal plane, and perform two-dimensional circular curve fitting or vector difference operation on the projection points in the spatial normal plane to obtain the two-dimensional coordinates of the center of the cylinder section and the actual fitting radius of the cylinder.

[0032] S6. Using the two-dimensional coordinates of the center of the cylindrical cross section, combined with the normal vector of the spatial normal plane and the coordinates of the reference point, the three-dimensional anchored cylinder center coordinates of the corresponding cross section position on the real central axis inside the cylinder are calculated through inverse coordinate transformation.

[0033] S7. Based on the three-dimensional anchored cylinder center coordinates and the set binocular camera observation direction, extend the normal direction to the cylinder surface with radius R, calculate the three-dimensional spatial coordinates of the highest axis relative to the fixed reference point that is not affected by the cylinder's rotation, and use these three-dimensional spatial coordinates as the benchmark for continuous positioning and movement distance measurement.

[0034] In some embodiments, in step S2, the threshold range of the adaptive threshold segmentation is 120-180, and the roundness threshold of the circular contour filtering is ≥0.85.

[0035] In some embodiments, in step S3, the triangulation principle is based on the intrinsic and extrinsic parameters of the binocular camera. The intrinsic parameters include focal length and principal point coordinates, and the extrinsic parameters include baseline distance, optical axis parallelism, rotation matrix, and translation vector. The resolution of the binocular camera is ≥2048×1536, and the baseline distance is 100mm-120mm.

[0036] In some embodiments, in step S5, the least squares method is used when fitting the two-dimensional circular curve of the projection points in the spatial normal plane. By minimizing the sum of the squares of the distances from the projection points corresponding to the seven black circles to the center of the fitted circle, the two-dimensional coordinates of the center of the cylindrical section and the actual fitted radius of the cylinder in the spatial normal plane are calculated.

[0037] In some embodiments, within a depth range of 1.2m-3.2m, the repeatability of the three-dimensional spatial coordinates of the highest axis of the cylindrical or cylindrical tool relative to a fixed reference point is ≤1mm, and the movement accuracy is ≤3mm.

[0038] It should be further noted that the technical features corresponding to the above-mentioned options and embodiments can be combined or substituted with each other to form new technical solutions without conflict.

[0039] Compared with the prior art, the beneficial effects of the present invention are:

[0040] 1. Significantly improved resistance to centroid shift: The specially designed circular profile adapts to the curvature of the cylinder through a nonlinear projection formula, and the symmetrical layout of the seven black circles can offset the centroid shift caused by small-angle rotation. The two-dimensional circle center extraction accuracy reaches 0.05mm, which is 10 times higher than that of existing targets.

[0041] 2. The three-dimensional positioning accuracy meets industrial requirements: Within a depth range of 1.2m-3.2m, the repeatability of the three-dimensional coordinates of the highest point of the cylinder is ≤1mm, and the movement accuracy is ≤3mm, which can cover the safety redundancy requirements of most industrial scenarios.

[0042] 3. Strong recognition robustness: The high contrast (grayscale difference ≥210) between the white background substrate and the black circle can resist interference from complex industrial backgrounds, and the circular outline screening (circularity ≥0.85) can avoid false detections, with a recognition success rate of ≥99.5%.

[0043] 4. Wide adaptability: The parameters (α, R) of the specially designed small circular target can be adjusted according to cylinders of different diameters, without the need to redesign the target structure, and the adaptability range covers cylinder structures with a radius of 50-100mm.

[0044] 5. Strong adaptability to spatial attitude: Through spatial axial vector extraction and normal plane projection fitting algorithms, the limitation that the measured tool must maintain a strictly vertical state is broken. Even if the tool is pitched or tilted in three-dimensional space, the coordinates of feature points on the cylindrical surface can still be calculated with high accuracy, greatly enhancing the universality of the algorithm in complex industrial scenarios. Attached Figure Description

[0045] Figure 1 This is a planar layout diagram of various types of irregularly shaped circular targets shown in an embodiment of the present invention;

[0046] Figure 2 The above are schematic diagrams of various types of circular irregular targets in planar view and cylindrical surface texture, as shown in the embodiments of the present invention.

[0047] Figure 3 This is a flowchart illustrating a binocular vision positioning method for cylindrical or cylindrical tools according to an embodiment of the present invention.

[0048] Figure 4 This is a schematic diagram showing the highest point of the cylinder, the anti-spin incenter, and the axial direction vector in an embodiment of the present invention. Detailed Implementation

[0049] The technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0050] It should be noted that the defects in the solutions in the prior art are all the results of the inventors' practice and careful research. Therefore, the discovery process of the above problems and the solutions proposed by the embodiments of this application in the following text should be the inventors' contributions to this application in the process of invention and creation, and should not be understood as technical content known to those skilled in the art.

[0051] In view of the technical problems pointed out in the background art, the present invention provides the following embodiments:

[0052] In one exemplary embodiment, a variety of circular irregular targets resistant to centroidal shift are provided, such as... Figure 1 As shown, it includes a white background substrate and seven black circles disposed on the white background substrate; the seven black circles are generally symmetrically distributed about the center of the white background substrate (symmetrically distributed along both the x-axis and y-axis), wherein:

[0053] A central large circle (number 1) is set at the center of a white background substrate as a reference for various types of irregularly shaped circular targets;

[0054] Two standard small circles (numbered 2 and 3) are set at the top and bottom of the white background substrate, with a radius of r1 and a diameter of D2. They are perfect circles and are suitable for scenes where the cylindrical tool axis does not rotate.

[0055] On a white background substrate, a specially designed small circular object (numbered 4, 5, 6, 7) with a diameter of D3 is set in the upper left, lower left, upper right, and lower right directions. Its geometry is designed based on the radius of the cylinder (let's call it R) to compensate for the centroid shift caused by the curvature and rotation of the cylindrical surface.

[0056] The outline of the specially designed circular miniature is determined by the following nonlinear projection formula, ensuring that after it is pasted onto the cylindrical or near-cylindrical surface, it approximates a perfect circle on the camera's imaging plane:

[0057] (1)

[0058] Alternatively, its upper bound expression can be written in display form as:

[0059] (2)

[0060] The lower bound expression is:

[0061] (3)

[0062] in, Let u = R be the horizontal coordinate (mm) of any point on the unfolded plane, corresponding to the circumferential coordinates of the cylinder. v is the vertical coordinate (mm) of any point on the unfolded plane, corresponding to the axial coordinate of the cylinder; r is the radius (mm) of the small circle of the standard perfect circle; α is the offset angle (° / rad) of the target relative to the surface of the cylinder or cylindrical tool; a positive value indicates rightward offset, usually within ±15°; R is the cylinder radius (mm) of the cylinder or cylindrical tool, determined according to the actual cylinder diameter. This refers to the cylindrical facet angle parameter (rad). The textures for these various types of irregularly shaped circular targets are as follows: Figure 2 As shown, This represents the horizontal coordinates of the origin of the unfolded plane. This represents the vertical coordinates of the origin of the unfolded plane.

[0063] Preferably, the white background substrate is made of polyethylene terephthalate (PET), with a thickness of 0.1mm-0.2mm, a size of 75-100mm×90-120mm, and a grayscale value ≥230; the seven black circles are all UV printed, with a grayscale value ≤20, and a grayscale difference ≥210 with the white background substrate.

[0064] Preferably, the diameter of the central large circle is 12mm-18mm, and the diameters of the standard small circle and the specially made small circle are both 10mm-15mm; the offset angle α ranges from 0° to ±20°, and the cylinder radius R ranges from 50mm to 100mm.

[0065] In another exemplary embodiment, based on the aforementioned various types of irregularly shaped circular targets, a binocular vision positioning method for cylindrical or quasi-cylindrical tools is provided. This method primarily achieves high-precision positioning of the cylindrical surface through "image extraction → stereo matching → cylinder fitting → highest point calculation," such as... Figure 3 As shown, the specific steps include:

[0066] S1. Simultaneously acquire left and right views of various circular irregular targets on the cylindrical surface of a cylindrical or cylindrical tool using a binocular camera;

[0067] S2. Perform adaptive threshold segmentation and circular outline filtering on the left and right views respectively to extract the two-dimensional center coordinates of the seven black circles;

[0068] S3. Number the seven black circles based on the central large circle, match the circles with the same number in the left and right views, and calculate the three-dimensional spatial coordinates of each circle through binocular vision epipolar constraints and triangulation principles.

[0069] S4. Based on the three-dimensional spatial coordinates of the central great circle and the standard small circles in the upward and downward directions, calculate the initial axial direction vector of the cylinder in space, and construct a spatial normal plane perpendicular to the initial axial direction vector.

[0070] S5. Project the three-dimensional spatial coordinates of the seven black circles onto the spatial normal plane, and perform two-dimensional circular curve fitting or vector difference operation on the projection points in the spatial normal plane to obtain the two-dimensional coordinates of the center of the cylinder section and the actual fitting radius of the cylinder.

[0071] S6. Using the two-dimensional coordinates of the center of the cylindrical cross section, combined with the normal vector of the spatial normal plane and the coordinates of the reference point, the three-dimensional anchored cylinder center coordinates of the corresponding cross section position on the real central axis inside the cylinder are calculated through inverse coordinate transformation.

[0072] S7. Based on the three-dimensional anchored cylinder center coordinates and the set binocular camera observation direction, extend the normal direction to the cylinder surface with radius R, calculate the three-dimensional spatial coordinates of the highest axis relative to the fixed reference point that is not affected by the cylinder's rotation, and use these three-dimensional spatial coordinates as the benchmark for continuous positioning and movement distance measurement.

[0073] Specifically, in step S1, target image acquisition is performed at a frequency of 10 fps. In step S2, the target center is extracted and denoted as the coordinates of the left view (…). ) to( ), right view coordinates ( )to( The adaptive threshold segmentation threshold range is 120-180, and the circularity threshold for circular contour filtering is ≥0.85.

[0074] In step S3, using the large central circle numbered 1 in the left view as a reference, the seven black circles are numbered in the order of "top → bottom → top right → bottom right → top left → bottom left". Based on the numbering, the black circles with the same number in the left and right views are matched. Using the binocular visual epipolar constraint and triangulation principles, the three-dimensional spatial coordinates of each black circle are calculated (denoted as...). ) to( (where the X-axis is the horizontal direction, the Z-axis is the depth direction, and the Y-axis is the vertical direction). Preferably, the triangulation principle is based on the intrinsic and extrinsic parameters of a binocular camera, wherein the intrinsic parameters include focal length, principal point coordinates, and lens distortion coefficient; the physical structural parameters of the binocular camera include baseline distance and optical axis parallelism, and the extrinsic parameters include rotation matrix and translation vector; the resolution of the binocular camera is ≥2048×1536, and the baseline distance is 100mm-120mm.

[0075] Furthermore, considering the potential spatial tilt, attitude changes, and rotation around an axis that cylindrical or quasi-cylindrical tools may experience in industrial settings, this method abandons traditional dimensionality reduction simplification and employs full-space three-dimensional coordinate calculation and cross-section anchoring mechanisms. The specific calculation process is illustrated in steps S4-S7.

[0076] In step S4, spatial axis vector extraction is performed. Since the target's numbered circles 2, 1, and 3 are located on the same axial generatrix on the cylinder surface, the line connecting them is parallel to the actual internal central axis of the cylinder in space. The algorithm first calculates the three-dimensional coordinates of these three points (let's call them...). Perform spatial line fitting or vector difference operations to extract the unit direction vector that can equivalently represent the direction of the true central axis of the cylinder in space. This allows us to obtain the tool's precise spatial tilt attitude in its current pose.

[0077] In step S5, the normal plane is constructed and the point cloud is orthogonally projected, using the three-dimensional spatial coordinates of the central great circle numbered 1. Using the origin as the reference point, the extracted axial direction vector As the normal vector, establish the local spatial normal plane. The three-dimensional spatial coordinates of the seven black circles calculated by binocular vision are then along... The directions are orthogonally projected onto the local spatial normal plane. Since the projection plane is strictly perpendicular to the cylinder axis, the longitudinal positional differences of the seven feature points are eliminated after projection, and they present a standard two-dimensional circular arc distribution within the local spatial normal plane.

[0078] Step S5 also involves fitting a two-dimensional circle to the normal plane and obtaining the local center of the cross-section. Within the aforementioned local spatial normal plane, the least squares method is used to fit a circular curve to the two-dimensional plane coordinates of the seven projection points. By minimizing the sum of the squares of the distances from each projection point to the fitted circle center, the two-dimensional coordinates of the local cross-section center of this specific cross-section are accurately obtained. and the actual fitted radius of the cylinder .

[0079] Step S6 involves inverse spatial coordinate transformation and three-dimensional anchoring column center calculation, using the two-dimensional coordinates of the local cross-section center obtained by fitting within the local space method plane. By performing an inverse transformation using the homogeneous transformation matrix of rotation and translation formed when constructing the normal plane, the coordinates of the three-dimensional anchored cylinder center in the global visual coordinate system are calculated. This point lies on the true central axis inside the cylinder, and its longitudinal position along the axial direction is uniquely locked by the central great circle numbered 1. Therefore, even if the measured cylinder rotates around its own axis, causing the surface target to slip, this internally anchored cylinder center remains intact. Its physical spatial position remains absolutely unchanged, possessing extremely strong geometric anti-interference capabilities.

[0080] In step S7, the highest axis fixed reference point is precisely mapped, such as... Figure 4 As shown, to obtain a relatively fixed observation reference point on the cylindrical surface that is unaffected by rotation, let the optical axis observation direction vector of the binocular camera be... The algorithm calculates the surface normal vector perpendicular to the cylinder axis and pointing towards the camera. :

[0081] 1. First, calculate the axial vector of the cylinder. With the observation direction vector outer product: ;

[0082] 2. Then calculate and The outer product is used to obtain the surface normal reference: and to Unitize the data;

[0083] 3. Finally, the internal three-dimensional anchor point of the column is used. Starting from the surface normal vector In the direction of expanding outward from the surface of the cylinder by a cylinder radius The formula for calculating the three-dimensional coordinates of its final reference point is:

[0084]

[0085] The solution This refers to a clearly defined and fixed reference point on the highest axis of the cylindrical surface. This point only undergoes actual displacement with the spatial translation and pitch tilt of the measured tool, completely shielding the surface feature slippage error caused by the tool's rotation, thereby achieving high-precision and robust continuous positioning and movement distance measurement of the target tool.

[0086] Based on the above target design and positioning method, the following provides a specific method for target fabrication and positioning, including:

[0087] I. Target Preparation and Pasting

[0088] Parameter determination: Based on the physical parameters of the cylindrical end effector in the industrial field, the radius of the actuator rod R is determined to be 50mm, the offset angle of the special circle is set to 15°, and the contour coordinates of the special circle are calculated by substituting into formula (1) to obtain the range of values ​​of u and v.

[0089] Target fabrication: Using a PET substrate (75mm×90mm, white, grayscale value ≥230), seven black circles are fabricated using UV printing technology. Circle 1 has a diameter of 18mm, circles 2 and 3 have a diameter of 10mm (perfect circle), and circles 4-7 are printed according to the outline of circle 2 based on formula (1).

[0090] Target attachment: Attach the target to the upper middle part of the cylinder, ensuring that the target coincides with the axial generatrix of the cylindrical tool. After attachment, calibrate using a binocular camera and record the initial Y coordinate of the target.

[0091] II. Binocular Camera Calibration and Parameter Settings

[0092] Camera calibration: The binocular camera is calibrated using a checkerboard calibration board to obtain intrinsic and extrinsic parameters, and the lens distortion coefficient is corrected.

[0093] Acquisition parameter settings: Camera resolution is set to 3088×2064, exposure time is set to 10ms, and gain is set to 1.0 to avoid overexposure or underexposure of images caused by excessively strong or weak ambient light.

[0094] III. Positioning Test and Result Verification

[0095] Test environment: Movement step size 20mm, range 0-600mm, detection depth range 1.2m-3.2m.

[0096] Test steps:

[0097] 1. Triaxial datum calibration:

[0098] First, the spatial correspondence between the coordinate system of the measurement system and the physical coordinate system of the guide rail is established through a three-axis linear guide rail. The reference height value of the cylindrical or quasi-cylindrical tool in the Y direction is determined, and the initial calibration of the system is completed.

[0099] 2. Single-axis planar movement test (X / Y direction):

[0100] While keeping the depth in the Z direction constant, the test tool is controlled to move along the X and Y axes in single-axis steps of 20 mm. Target images at corresponding positions are acquired point by point within the range of 0-600 mm to evaluate the system's positioning repeatability and measurement error in the planar directions (X, Y).

[0101] 3. Multi-depth condition testing (Z direction):

[0102] Based on the above planar movement test, the test tool was adjusted to three different depth positions of 1.2 m, 2.2 m and 3.2 m respectively. The single-axis movement test in the X and Y directions was repeated under each depth condition to evaluate the Z-axis measurement stability and overall error variation of the system under different working depth conditions.

[0103] 4. Three-dimensional coordinate calculation:

[0104] The acquired binocular image data is processed sequentially as follows: “Circle center extraction → Stereo matching → Spatial method plane projection fitting → Feature point calculation” to obtain the three-dimensional spatial coordinates (X,Y,Z) of the target point corresponding to each test position.

[0105] 5. Accuracy Comparison and Error Calculation:

[0106] The three-dimensional coordinate results obtained by binocular vision measurement are compared with the measurement results of laser tracker (accuracy ±0.01 mm). The absolute error and repeatability error in the X, Y and Z directions are calculated respectively to evaluate the overall three-dimensional positioning accuracy of the measurement system.

[0107] Test results:

[0108] Error in 2D circle center extraction:

[0109] Under different positions and depths, the error range for extracting the two-dimensional center of the target is 0.03 mm to 0.05 mm, with an average value of 0.04 mm.

[0110] 3D coordinate repeatability accuracy:

[0111] The repeatability errors in the X, Y, and Z directions are all kept within the range of 0.7 mm to 0.9 mm, and the average repeatability is about 0.8 mm, indicating that the system has good measurement stability.

[0112] Overall accuracy of the 3D measurement system:

[0113] Within a working depth range of 1.2 m to 3.2 m, the system's overall measurement error in the X, Y, and Z directions is distributed between 1.2 mm and 2.8 mm, with an average three-dimensional measurement accuracy of 2.4 mm, all of which meet the design requirements and industrial application needs.

[0114] The above detailed embodiments are a description of the present invention. It should not be considered that the specific embodiments of the present invention are limited to these descriptions. For those skilled in the art, several simple deductions and substitutions can be made without departing from the concept of the present invention, and all of these should be considered to fall within the protection scope of the present invention.

Claims

1. A multi-type circular irregular target with resistance to centroid displacement, characterized in that, It includes a white background substrate and seven black circles disposed on the white background substrate; the seven black circles are symmetrically distributed about the center of the white background substrate, wherein: A large central circle is set at the center of a white background substrate to serve as a reference for various types of irregularly shaped circular targets; Two standard small circles are set at the top and bottom of a white background substrate; A specially designed small circle is set in the upper left, lower left, upper right, and lower right directions of the white background substrate; The outline of the specially designed small circle is determined by the following nonlinear projection formula: Where u is the horizontal coordinate of any point on the cylindrical unfolded plane, corresponding to the circumferential unfolded coordinate of the cylinder; v is the vertical coordinate of any point on the cylindrical unfolded plane, corresponding to the axial coordinate of the cylinder; r is the radius of the standard circle; α is the offset angle of the target relative to the surface of the cylinder or cylindrical tool, α is within ±15°; and R is the cylindrical radius of the cylinder or cylindrical tool. The target is pasted on the surface of the cylinder or cylindrical tool, and the target coincides with the axial generatrix of the cylinder or cylindrical tool.

2. The multi-type circular irregular target with anti-centroid displacement according to claim 1, characterized in that, The white background substrate is made of polyethylene terephthalate, with a thickness of 0.1mm-0.2mm, a size of 75-100mm×90-120mm, and a grayscale value ≥230; the seven black circles are all UV printed, with a grayscale value ≤20, and a grayscale difference ≥210 with the white background substrate.

3. A multi-type circular irregular target with anti-centroid displacement according to claim 1, characterized in that, The diameter of the central large circle is 12mm-18mm, and the diameters of the standard small circle and the specially made small circle are both 10mm-15mm; the value range of the offset angle α is 0°-±20°, and the value range of the cylinder radius R is 50mm-100mm.

4. A binocular vision positioning method for cylindrical or quasi-cylindrical tools based on any of the multiple types of circular irregular targets described in claims 1-3, characterized in that, Includes the following steps: S1. Simultaneously acquire left and right views of various circular irregular targets on the cylindrical surface of a cylindrical or cylindrical tool using a binocular camera; S2. Perform adaptive threshold segmentation and circular outline filtering on the left and right views respectively to extract the two-dimensional center coordinates of the seven black circles; S3. Number the seven black circles based on the central large circle, match the circles with the same number in the left and right views, and calculate the three-dimensional spatial coordinates of each circle through binocular vision epipolar constraints and triangulation principles. S4. Based on the three-dimensional spatial coordinates of the central great circle and the standard small circles in the upward and downward directions, calculate the initial axial direction vector of the cylinder in space, and construct a spatial normal plane perpendicular to the initial axial direction vector. S5. Project the three-dimensional spatial coordinates of the seven black circles onto the spatial normal plane, and perform two-dimensional circular curve fitting or vector difference operation on the projection points in the spatial normal plane to obtain the two-dimensional coordinates of the center of the cylinder section and the actual fitting radius of the cylinder. S6. Using the two-dimensional coordinates of the center of the cylindrical cross section, combined with the normal vector of the spatial normal plane and the coordinates of the reference point, the three-dimensional anchored cylinder center coordinates of the corresponding cross section position on the real central axis inside the cylinder are calculated through inverse coordinate transformation. S7. Based on the three-dimensional anchored cylinder center coordinates and the set binocular camera observation direction, extend the normal direction to the cylinder surface with radius R, calculate the three-dimensional spatial coordinates of the highest axis relative to the fixed reference point that is not affected by the cylinder's rotation, and use these three-dimensional spatial coordinates as the benchmark for continuous positioning and movement distance measurement.

5. The binocular vision positioning method for cylindrical or quasi-cylindrical tools according to claim 4, characterized in that, In step S2, the threshold range of the adaptive threshold segmentation is 120-180, and the roundness threshold of the circular contour filtering is ≥0.

85.

6. The binocular vision positioning method for cylindrical or quasi-cylindrical tools according to claim 4, characterized in that, In step S3, the triangulation principle is based on the intrinsic and extrinsic parameters of the binocular camera. The intrinsic parameters include focal length and principal point coordinates, and the extrinsic parameters include baseline distance, optical axis parallelism, rotation matrix, and translation vector. The resolution of the binocular camera is ≥2048×1536, and the baseline distance is 100mm-120mm.

7. The binocular vision positioning method for cylindrical or quasi-cylindrical tools according to claim 4, characterized in that, In step S5, the least squares method is used when fitting the two-dimensional circular curve of the projection points in the spatial normal plane. By minimizing the sum of the squares of the distances from the projection points corresponding to the seven black circles to the center of the fitted circle, the two-dimensional coordinates of the center of the cylindrical section and the actual fitted radius of the cylinder in the spatial normal plane are calculated.

8. The binocular vision positioning method for cylindrical or quasi-cylindrical tools according to claim 4, characterized in that, Within a depth range of 1.2m-3.2m, the repeatability of the three-dimensional spatial coordinates of the highest axis of the cylindrical or cylindrical tool relative to a fixed reference point is ≤1mm, and the movement accuracy is ≤3mm.

Images