Inverted coplanarity detection method based on 3D vision and refraction compensation

Abstract

The invention discloses an inverted coplanarity detection method based on 3D vision and refraction compensation, and the method comprises the following steps: 1, selecting an optical measuring head of which the measuring range can cover the whole bottom surface of a measured object according to the size of the bottom surface of the measured object, installing the optical measuring head below glass, and enabling light to penetrate through the glass from bottom to top; 2, establishing a glass refraction compensation model; 3, correcting the glass refraction compensation model; 4, collecting three-dimensional point cloud data of the bottom surface of the measured object; 5, point cloud extraction and processing; step 6, calculating and evaluating the coplanarity; and 7, repeating the steps 4-6 to realize coplanarity analysis under different conditions. Through fusion of a theoretical model and experimental calibration, nonlinear error measurement caused by factors such as glass refraction and non-ideal installation is solved, the measurement precision is improved, the reference plane is established by adopting a minimum area method, the coplanarity evaluation reasonability is improved, and the method can be expanded and applied to all occasions needing to measure the coplanarity of the bottom structure from bottom to top.

Description

Technical Field

[0001] The present invention belongs to the technical field of three-dimensional measurement, and particularly relates to an inverted coplanarity detection method based on 3D vision and refraction compensation. Background Art

[0002] In the current industrial inspection field, the three-dimensional topography measurement of precision components still faces significant technical bottlenecks. The traditional inspection system shows a transitional feature from two-dimensional to three-dimensional evolution: in the two-dimensional inspection level, the planar dimension and contour inspection technology based on machine vision has been relatively mature; while in the three-dimensional inspection dimension, especially when it comes to complex geometric tolerance indicators such as flatness and coplanarity, the existing inspection methods have systematic defects in terms of inspection accuracy, efficiency, and adaptability.

[0003] Taking the coplanarity inspection of the support feet of a laptop as an example, the current industrial standard still uses the contact measurement method, and uses contact tools such as feeler gauges to measure the gap between the feet and the platform, which has the following technical defects:

[0004] First, the measurement pressure causes elastic deformation of the feet: the contact pressure generated when the measurement tool contacts the flexible feet causes non-negligible elastic deformation of the feet, and the measurement error introduced by this deformation can reach the order of ±0.05 mm;

[0005] Second, limited by the manual operation mode, the single measurement takes more than 30 s, and it is necessary to repeat the positioning of the measurement points, which cannot meet the inspection requirements of the modern production line with a beat of more than 60 pieces / minute;

[0006] Third, the current method can only obtain the local height information of discrete points, lacking the ability to continuously represent the full three-dimensional topography of the feet, resulting in the inability to obtain the full-area topography of the feet, thus leading to the risk of missed detection of potential quality defects.

[0007] In the non-contact three-dimensional vision inspection field, the existing technical solutions face multiple technical challenges:

[0008] First, the current three-dimensional vision inspection method has poor adaptability to non-planar feet (inclined plane / curved surface). When the inspection object shows inclined plane or curved surface characteristics (such as special-shaped foot structure), the point cloud registration error increases significantly, and the typical deviation is 0.1 - 0.3 mm;

[0009] Second, there is refraction interference of the transparent medium: when there is a glass protection layer between the measured part and the reference plane, the light refraction effect causes systematic distortion of the three-dimensional point cloud, and currently there is still a lack of an effective refraction path compensation algorithm, resulting in a large measurement error;

[0010] Third, the current method cannot analyze the true three-dimensional spatial relationship between the contact area of the bottom surface of the feet and the reference plane, resulting in a matching degree between the true contact state and the theoretical model of less than 85%.

[0011] Based on this, the present invention discloses an inverted coplanarity detection method based on 3D vision and refraction compensation. Summary of the Invention

[0012] To solve the technical problems existing in the prior art, the object of the present invention is to provide an inverted coplanarity detection method based on 3D vision and refraction compensation, which improves the coplanarity measurement accuracy and evaluation reliability by solving the measurement error problems caused by glass refraction and reflection differences and the problem of establishing a reference plane.

[0013] To achieve the above object and reach the above technical effects, the technical solution adopted by the present invention is as follows:

[0014] An inverted coplanarity detection method based on 3D vision and refraction compensation, comprising the following steps:

[0015] Step 1: Hardware configuration

[0016] According to the bottom surface size of the object to be measured, select an optical probe whose measurement range can cover the entire bottom surface of the object to be measured, install the optical probe below the glass, and the light can penetrate the glass from bottom to top;

[0017] Step 2: Establish a glass refraction compensation model to correct the path offset caused by refraction and / or reflection when the light passes through the glass;

[0018] Step 3: Correct the glass refraction compensation model obtained in Step 2;

[0019] Step 4: Collect three-dimensional point cloud data of the bottom surface of the object to be measured;

[0020] Step 5: Extract and process the point cloud;

[0021] Step 6: Calculate and evaluate the coplanarity;

[0022] Step 7: Repeat Steps 4-6 to achieve coplanarity analysis under different conditions.

[0023] Further, in Step 1, select a suitable optical probe according to the material of the object to be measured. For the measurement of a black light-absorbing surface, use a high-power pulsed laser and a long-exposure mode laser line-scanning 3D optical probe. For the measurement of a reflective surface, suppress the reflection noise by adding a polarization filter in front of the optical probe. The optical axis of the optical probe is perpendicular to the glass plane;

[0024] Further, in Step 1, the glass is a plane glass, and an antireflection film is coated on the glass to reduce the influence of the glass surface reflection on the measurement.

[0025] Further, in Step 2, the steps for establishing the glass refraction compensation model are as follows:

[0026] Based on the selected glass in Step 1, establish a theoretical correction formula based on the law of optical refraction, directly calculate the error compensation value through the glass refractive index, glass thickness, and air refractive index, and convert the measured value Z of the optical probe measured to the true height Z corrected :

[0027]

[0028] In the formula, n glass is the glass refractive index, n air is the air refractive index, and t is the glass thickness.

[0029] Furthermore, in Step 3, the steps for correcting the glass refraction compensation model obtained in Step 2 include:

[0030] According to the glass refraction compensation model established in Step 2, through experimental calibration data, fit the compensation coefficients (a, b) of the actual error, generate a dynamic lookup table to cover the non-ideal factors not considered in the theoretical glass refraction compensation model, and correct the residual error of the theoretical glass refraction compensation model through linear regression:

[0031]

[0032] In the formula, Z real is the true value of the step gauge, and Z meas is the height value of the step gauge measured by the optical probe through the glass.

[0033] Furthermore, in Step 3, the experimental calibration data includes the known height of the step gauge.

[0034] Furthermore, in Step 4, the steps for collecting the three-dimensional point cloud data of the bottom surface of the object to be measured include:

[0035] Use the optical probe to perform a full-field measurement on the bottom surface of the object to be measured, calculate the error compensation value according to the glass refraction compensation model in Step 2, and perform model correction according to Step 3 to obtain high-precision three-dimensional point cloud data of the bottom surface of the object to be measured.

[0036] Furthermore, in Step 5, according to the obtained three-dimensional point cloud data of the bottom surface of the object to be measured, through template matching with the CAD drawing, or using a pre-trained U-Net model to perform semantic segmentation on the point cloud, extract the surface point cloud, and perform a filtering operation to eliminate high-frequency noise.

[0037] Furthermore, in Step 6, perform coplanarity calculation and evaluation to obtain the coplanarity evaluation index, including the following steps:

[0038] Based on the surface point cloud data obtained in step 5, identify the surface type of the object under test, and define the contact area of the object under test as the set of points in the point cloud where the z value is less than the threshold according to different surface types;

[0039] Determine an optimal reference plane so that the deviation of all the point clouds in the contact area of the object under test with respect to the optimal reference plane satisfies the minimum zone condition, that is, the difference between the maximum positive deviation and the maximum negative deviation is the smallest. Then, use the absolute height difference between the highest point and the lowest point of the point cloud in the contact area of the object under test to the optimal reference plane as the coplanarity evaluation index.

[0040] Furthermore, the calculation steps of the coplanarity evaluation index are as follows:

[0041] (1) Obtain the coordinate set {P i} of the three-dimensional point cloud of the bottom surface of the object under test from step 4;

[0042] (2) Use the RANSAC algorithm to perform plane fitting on all the point clouds. The plane equation is αx + βy + γz + δ = 0, and obtain the initial plane parameters (α0, β0, γ0, δ0);

[0043] (3) Define the objective function f(α, β, γ, δ), and use the particle swarm optimization algorithm to search for the optimal reference plane parameters. The constraint condition is that the plane normal vector is unitized, α 2 + β 2 + γ 2 = 1, and obtain the optimal reference plane parameters (α’, β’, γ’, δ’):

[0044] f(α, β, γ, δ) = max(d i ) - min(d i )

[0045]

[0046] In the formula, d i is the algebraic distance from the i-th surface point cloud of the object under test to the optimal reference plane, and (x i , y i , z i ) are the coordinates of the surface point cloud of the object under test;

[0047] (4) Define the coplanarity evaluation index as the absolute height difference Δh i between the highest point max(z i ) and the lowest point min(z max ) of the point cloud in the contact area of the object under test to the optimal reference plane:

[0048] Δh max = max(z i ) - min(z i ).

[0049] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0050] 1) By fusing the theoretical model (glass refraction compensation model) with experimental calibration, the non-linear error measurement caused by factors such as glass refraction and reflection and non-ideal installation is solved, and the measurement accuracy is improved;

[0051] 2) The minimum zone method is used to establish a reference plane, which improves the rationality of coplanarity evaluation;

[0052] 3) By solving the measurement error problem caused by the differences in glass refraction and reflection and the problem of establishing a reference plane, the present invention improves the coplanarity measurement accuracy and evaluation reliability. When detecting the coplanarity of laptop foot pads, the detection efficiency reaches 5 seconds per unit (supporting on-line detection on the production line), the matching degree between the actual contact state and the theoretical model is as high as over 90%, and it can be compatible with the detection of different types of foot pads of all laptop models. It can be extended and applied to all occasions where the coplanarity of the bottom structure needs to be measured from bottom to top. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 It is a flowchart of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0054] The present invention will be described in detail below so that the advantages and features of the present invention can be more easily understood by those skilled in the art, thereby making a clearer and more definite definition of the protection scope of the present invention.

[0055] The following gives a brief overview of one or more aspects to provide a basic understanding of these aspects. This overview is not an exhaustive survey of all contemplated aspects, and is neither intended to identify key or decisive elements of all aspects nor to define the scope of any or all aspects. Its sole purpose is to present some concepts of one or more aspects in a simplified form as a prelude to the more detailed description that follows.

[0056] As Figure 1 shown, the present invention discloses an inverted coplanarity detection method based on 3D vision and refraction compensation, including the following steps:

[0057] Step 1: Hardware configuration

[0058] According to the bottom surface size of the object to be measured, an optical probe and a glass with a measurement range that can cover the entire bottom surface of the object to be measured are selected. The optical probe is installed below the glass, and the optical path direction penetrates the glass from bottom to top, and the optical axis of the optical probe is perpendicular to the glass plane.

[0059] The optical probe used in the present invention is preferably a 3D optical probe. It is necessary to select a suitable 3D optical probe according to the material of the object to be measured. For example, for the measurement of a black light-absorbing surface, a 3D optical probe with a high-power pulsed laser and a long-exposure mode laser line scan can be used. For the measurement of a reflective surface, it is necessary to add a polarization filter in front of the 3D optical probe to suppress the reflection noise.

[0060] The glass used in the present invention is preferably flat glass, and it can be selected according to actual needs whether to reduce the influence of the reflection on the glass surface on the measurement by coating an anti-reflection film.

[0061] Step 2: Establishment of the glass refraction compensation model

[0062] Based on the glass selected in Step 1, establish a theoretical correction formula based on the optical refraction law, directly calculate the error compensation value through physical parameters (glass refractive index, glass thickness, air refractive index), correct the path offset caused by refraction when the light passes through the flat glass, and convert the measured value Z of the optical probe measured into the true height Z corrected :

[0063]

[0064] In the formula, n glass is the glass refractive index, n air is the air refractive index, and t is the glass thickness.

[0065] Step 3: System calibration

[0066] According to the glass refraction compensation model established in Step 2, fit the compensation coefficients (a, b) of the actual error through experimental calibration data, generate a dynamic look-up table to cover the non-ideal factors not considered in the theoretical model, and correct the residual error of the theoretical model through linear regression (such as uneven local thickness of the glass, installation tilt, refractive index fluctuation, etc.):

[0067]

[0068] In the formula, Z real is the true value of the step gauge, and Z meas is the height value of the step gauge measured by the optical probe through the glass.

[0069] Step 4: Acquisition of the 3D point cloud data of the bottom surface of the object to be measured

[0070] Use an optical probe (which can be a point-scanning type, line-scanning type or surface-structured optical probe) to perform a full-field measurement on the bottom surface of the object to be measured, calculate the error compensation value according to the glass refraction compensation model in Step 2, perform model correction according to Step 3, compensate for the residual error, and obtain high-precision 3D point cloud data of the bottom surface of the object to be measured.

[0071] Step 5: Point cloud extraction and processing of the foot pad area

[0072] According to the obtained three-dimensional point cloud data of the bottom surface of the object to be measured, through template matching with the CAD drawing or semantic segmentation of the point cloud using a pre-trained U-Net model, the point clouds on the surfaces of each foot pad are extracted, and further filtering operations (such as Gaussian filtering, bilateral filtering, etc.) are performed on the segmented foot pad point clouds to eliminate high-frequency noise.

[0073] Step 6: Coplanarity calculation and evaluation

[0074] According to the point cloud data of the foot pad surface obtained in Step 5, curvature analysis and model fitting algorithms are used to automatically identify the surface type of the foot pad (plane, inclined plane, arc surface), and the contact area of the foot pad is defined as the point set with z value less than the threshold in the point cloud according to different surface types;

[0075] Determine an optimal reference plane so that the deviation of all point clouds in the foot pad contact area relative to this plane meets the minimum area condition, that is, the difference between the maximum positive deviation and the maximum negative deviation is the smallest, and then use the absolute height difference between the highest point and the lowest point of the point cloud in the foot pad contact area to this plane as the coplanarity evaluation index. The specific steps are as follows:

[0076] (1) Obtain the coordinate set {P i} of each foot pad point cloud from Step 4;

[0077] (2) Perform plane fitting on all foot pad point clouds. The plane equation is αx + βy + γz + δ = 0, and the initial plane parameters (α0, β0, γ0, δ0) are obtained;

[0078] (3) Define the objective function f(α, β, γ, δ), search for the optimal plane parameters, and set the constraint condition as the unit normalization of the plane normal vector (α 2 +β 2 +γ 2 = 1), and obtain the optimal plane parameters (α’, β’, γ’, δ’):

[0079] f(α, β, γ, δ) = max(d i ) - min(d i )

[0080]

[0081] In the formula, d i is the algebraic distance from the i-th foot pad surface point cloud to the optimal reference plane, and (x i , y i , z i ) are the coordinates of the foot pad surface point cloud;

[0082] (4) Define the coplanarity evaluation index as the absolute height difference Δh between the highest point max(z i ) and the lowest point min(z i ) of the point cloud in the foot pad contact area from the optimal reference plane, which can be obtained by the following formula: max , and it can be obtained from the following formula:

[0083] Δh max = max(z i ) - min(z i ).

[0084] Step 7: Repeat steps 4 - 6 to achieve coplanarity analysis under different conditions.

[0085] Embodiment 1

[0086] As Figure 1 shown, an inverted coplanarity detection method based on 3D vision and refraction compensation, which is applied to the coplanarity detection of laptop foot pads, includes the following steps:

[0087] Step 1: Hardware configuration

[0088] According to the bottom surface size of the object to be measured (laptop foot pad), select a 3D optical probe and glass whose measurement range can cover the entire bottom surface of the object to be measured. Install the 3D optical probe under the glass, and the optical path direction penetrates the glass from bottom to top. The optical axis of the 3D optical probe is perpendicular to the glass plane.

[0089] In this embodiment, a suitable 3D optical probe can be selected according to the material of the object to be measured. For example, for the measurement of a black light-absorbing surface, a laser line-scanning 3D optical probe with high-power pulsed laser and long exposure mode can be used. For the measurement of a reflective surface, the reflection noise can be suppressed by adding a polarization filter in front of the 3D optical probe.

[0090] The glass used in this embodiment is a plane glass, and the reflection on the glass surface can be reduced by coating an anti-reflection film to reduce the influence on the measurement.

[0091] Step 2: Establishment of glass refraction compensation model

[0092] According to the plane glass selected in step 1, establish a theoretical correction formula based on the optical refraction law, directly calculate the error compensation value through physical parameters (glass refractive index, glass thickness, air refractive index), and convert the measurement value Z measured of the 3D optical probe into the true height Z corrected to correct the path offset caused by refraction when the light passes through the plane glass:

[0093]

[0094] In the formula, n glass is the glass refractive index, nair where \(n\) is the refractive index of air and \(t\) is the thickness of the glass.

[0095] Step 3: System calibration

[0096] According to the glass refraction compensation model established in Step 2, the compensation coefficients \((a, b)\) of the actual error are fitted through experimental calibration data (the known height of the step gauge), and a dynamic look-up table is generated to cover non-ideal factors not considered in the theoretical model. The residual error of the theoretical model (such as uneven local thickness of the glass, installation tilt, refractive index fluctuation, etc.) is corrected through linear regression:

[0097]

[0098] In the formula, \(Z\) real is the true value of the step gauge, and \(Z\) meas is the height value of the step gauge measured by the optical probe through the glass.

[0099] Step 4: Acquisition of three-dimensional point cloud data of the bottom surface of the measured object

[0100] The full-field measurement of the bottom surface of the measured object is carried out by using a point-scanning 3D optical probe. The error compensation value is calculated according to the glass refraction compensation model in Step 2, and the model is corrected according to Step 3 to compensate for the residual error, so as to obtain high-precision three-dimensional point cloud data of the bottom surface of the measured object.

[0101] Step 5: Point cloud extraction and point cloud processing of the foot pad area

[0102] According to the obtained three-dimensional point cloud data of the bottom surface of the measured object, the point clouds on the surfaces of each foot pad are extracted by template matching with the CAD drawing, and the foot pad point clouds are filtered by the conventional Gaussian filtering method to eliminate high-frequency noise.

[0103] Step 6: Coplanarity calculation and evaluation

[0104] According to the point cloud data of the foot pad surface obtained in Step 5, the curvature analysis and model fitting algorithm are used to automatically identify the foot pad surface type (plane, inclined plane, arc surface), and the contact area of the foot pad is defined as the point set with the \(z\) value less than the threshold in the point cloud according to different surface types, \(z < z\) min + 0.05 mm;

[0105] Furthermore, an optimal reference plane is determined so that the deviation of the point cloud in all foot pad contact areas with respect to this plane satisfies the minimum zone condition, that is, the difference between the maximum positive deviation and the maximum negative deviation is the smallest. Then, the absolute height difference between the highest point and the lowest point of the foot pad contact area point cloud to this plane is used as the coplanarity evaluation index. The specific steps are as follows:

[0106] (1) The coordinate set \(\{P\) i \} of each foot pad point cloud is obtained from Step 4;

[0107] (2) Use the RANSAC algorithm to perform plane fitting on all foot pad point clouds. The plane equation is αx + βy + γz + δ = 0, and the initial plane parameters (α0, β0, γ0, δ0) are obtained.

[0108] (3) Define the objective function f(α, β, γ, δ), and use the particle swarm optimization algorithm to search for the optimal plane parameters. The constraint condition is that the plane normal vector is unitized (α 2 + β 2 + γ 2 = 1), and the optimal plane parameters (α’, β’, γ’, δ’) are obtained:

[0109] f(α, β, γ, δ) = max(d i ) - min(d i )

[0110]

[0111] In the formula, d i is the algebraic distance from the i-th foot pad surface point cloud to the optimal reference plane, and (x i , y i , z i ) are the coordinates of the foot pad surface point cloud.

[0112] (4) Define the coplanarity evaluation index as the absolute height difference Δh i between the highest point max(z i ) and the lowest point min(z max ) of the foot pad contact area point cloud to the optimal reference plane, which can be obtained by the following formula:

[0113] Δh max = max(z i ) - min(z i ).

[0114] Step 7: Coplanarity evaluation at different opening angles of the laptop screen

[0115] Control the screen opening angle to 0° / 90° / 120°, repeat steps 4 - 6, compare the coplanarity data at different angles, and generate a deformation trend report.

[0116] For the parts or structures not specifically described in the present invention, existing technologies or existing products can be adopted, and no further elaboration will be made here.

[0117] The above are only the embodiments of the present invention, and do not limit the patent scope of the present invention accordingly. Any equivalent structure or equivalent process transformation made by using the content of the specification of the present invention, or directly or indirectly applied in other related technical fields, shall be included in the patent protection scope of the present invention by the same token.

Claims

1. An inverted coplanarity detection method based on 3D vision and refraction compensation, characterized in that, It includes the following steps: Step 1: Hardware configuration According to the bottom surface size of the object to be measured, select an optical probe whose measurement range can cover the entire bottom surface of the object to be measured, install the optical probe below the glass, and the light can penetrate the glass from bottom to top; Step 2: Establish a glass refraction compensation model to correct the path offset caused by refraction and / or reflection when the light passes through the glass; Step 3: Correct the glass refraction compensation model obtained in Step 2; Step 4: Collect three-dimensional point cloud data of the bottom surface of the object to be measured; Step 5: Extract and process the point cloud; Step 6: Calculate and evaluate the coplanarity; Step 7: Repeat Steps 4 - 6 to achieve coplanarity analysis under different conditions.

2. The inverted coplanarity detection method based on 3D vision and refraction compensation according to claim 1, wherein In Step 1, select a suitable optical probe according to the material of the object to be measured. For the measurement of a black light-absorbing surface, use a high-power pulsed laser and a long-exposure mode laser line-scanning 3D optical probe. For the measurement of a reflective surface, suppress the reflection noise by adding a polarization filter in front of the optical probe. The optical axis of the optical probe is perpendicular to the glass plane.

3. A method for detecting the inverted coplanarity based on 3D vision and refraction compensation according to claim 1, characterized in that, In Step 1, the glass is a plane glass, and an anti-reflection film is coated on the glass to reduce the influence of the glass surface reflection on the measurement.

4. A method for detecting the inverted coplanarity based on 3D vision and refraction compensation according to claim 1, characterized in that In Step 2, the steps for establishing the glass refraction compensation model are as follows: According to the glass selected in step 1, a theoretical correction formula is established based on the law of optical refraction. The error compensation value is calculated directly through the glass refractive index, glass thickness, and air refractive index. The measured value Z of the optical probe is converted to measured Convert to real height Z corrected : where n glass is the refractive index of the glass, and n air is the refractive index of air, and t is the thickness of the glass.

5. A method for detecting the inverted coplanarity based on 3D vision and refraction compensation according to claim 1, characterized in that In Step 3, the steps for correcting the glass refraction compensation model obtained in Step 2 include: According to the glass refraction compensation model established in Step 2, through experimental calibration data, fit the compensation coefficients (a, b) of the actual error, generate a dynamic lookup table to cover the non-ideal factors not considered in the theoretical glass refraction compensation model, and correct the residual error of the theoretical glass refraction compensation model through linear regression: where Z real is the true value of the step gauge, and Z meas is the height value of the step gauge measured by the optical probe through the glass.

6. The inverted coplanarity detection method based on 3D vision and refraction compensation according to claim 1, characterized in that, In Step 3, the experimental calibration data includes the known height of the step gauge.

7. A method for detecting the inverted coplanarity based on 3D vision and refraction compensation according to claim 1, characterized in that, In Step 4, the steps for collecting three-dimensional point cloud data of the bottom surface of the object to be measured include: Use the optical probe to perform a full-field measurement on the bottom surface of the object to be measured, calculate the error compensation value according to the glass refraction compensation model in Step 2, and perform model correction according to Step 3 to obtain high-precision three-dimensional point cloud data of the bottom surface of the object to be measured.

8. A method for detecting the inverted coplanarity based on 3D vision and refraction compensation according to claim 1, characterized in that, In Step 5, according to the obtained three-dimensional point cloud data of the bottom surface of the object to be measured, by performing template matching with the CAD drawing or using a pre-trained U-Net model to perform semantic segmentation on the point cloud, extract the surface point cloud and perform a filtering operation to eliminate high-frequency noise.

9. A method for detecting the inverted coplanarity based on 3D vision and refraction compensation according to claim 1, characterized in that, In Step 6, calculate and evaluate the coplanarity to obtain a coplanarity evaluation index, including the following steps: According to the surface point cloud data obtained in Step 5, identify the surface type of the object to be measured, and define the contact area of the object to be measured as the point set with a z value less than the threshold in the point cloud according to different surface types; Determine an optimal reference plane so that the deviation of all the point clouds in the contact area of the object to be measured with respect to the optimal reference plane satisfies the minimum zone condition, that is, the difference between the maximum positive deviation and the maximum negative deviation is the smallest, and then use the absolute height difference between the highest point and the lowest point of the point cloud in the contact area of the object to be measured to the optimal reference plane as the coplanarity evaluation index.

10. A method for detecting the inverted coplanarity based on 3D vision and refraction compensation according to claim 9, characterized in that The calculation steps of the coplanarity evaluation index are as follows: (1) The coordinate set {P of the three-dimensional point cloud of the bottom surface of the object to be measured obtained from step 4 i}; (2) Use the RANSAC algorithm to perform plane fitting on all the point clouds, and the plane equation is αx + βy + γz + δ = 0 to obtain the initial plane parameters (α0, β0, γ0, δ0); (3) Define the objective function f(α, β, γ, δ), and use the particle swarm optimization algorithm to search for the optimal reference plane parameters. The constraint condition is that the unit vector of the plane normal is α 2 + β 2 + γ 2 = 1, and obtain the optimal reference plane parameters (α’, β’, γ’, δ’): f(α,β,γ,δ) = max(d i ) - min(d i ) where d i is the algebraic distance from the surface point cloud of the i-th measured object to the optimal reference plane, and (x i , y i , z i ) are the coordinates of the surface point cloud of the measured object; (4) Define the coplanarity evaluation index as the absolute height difference Δh between the highest point max(z i ) and the lowest point min(z i ) of the point cloud in the contact area of the measured object to the optimal reference plane max : Δh max = max(z i ) - min(z i )。

Images