A method and device for detecting the profile of a surface coating on a flat panel

Abstract

The application provides a kind of plane panel surface coating profile size detection method, comprising the following steps: arranging positioning target in the uncoated area of the plane panel to be measured and collecting the image of the plane panel to be measured, extracting the panel outer contour and coating profile from the image;High-precision three-dimensional reconstruction is carried out on the positioning target to obtain three-dimensional point cloud data, and the spatial plane expression of the plane panel is fitted;The coating profile and the panel outer contour are calculated by inverse projection to the plane panel surface, and the three-dimensional point cloud data of the coating profile and the panel outer contour is obtained, and the two-dimensional point cloud data is obtained by dimension reduction on the three-dimensional point cloud data;Read the design profile of the plane panel, calculate the coordinate conversion relationship with the outer contour as the registration object;The two-dimensional point cloud data of the coating profile is registered with the design profile by the coordinate conversion relationship, and the detection value of the coating profile size error is obtained.The application can be suitable for surface coating size detection of plane panels with different sizes, sizes and coating shapes.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of industrial visual detection, and particularly relates to a surface coating profile size detection method and device for a planar panel. BACKGROUND

[0002] Visual detection technology is a non-contact optical measurement method, which uses a camera to collect image data of a measured object from different angles, and then extracts three-dimensional coordinate information of specific points or the whole field on the surface of the measured object by analyzing image information, and has the advantages of high precision and high speed, and is widely used in the detection of geometric quantities in the fields of aviation, aerospace and automobiles.

[0003] The honeycomb sandwich structure panel used on a spacecraft is usually made of aluminum alloy panel. In order to enhance the surface radiation heat exchange, a thermal control coating with high emissivity is coated on the surface of the aluminum alloy panel. In order to ensure the accuracy of the equipment installation and assembly parts reserved in the coating production process and avoid the delay caused by the interference problem in the subsequent assembly, the consistency of the thermal control coating profile size of the structure panel and the size marked on the design drawing needs to be checked.

[0004] The existing visual detection technology can measure the profile in pattern detection, but it is difficult to accurately align the measurement results with the design drawing and realize size inspection for planar panels with "multiple types and small batches" characteristics which lack fixed features. Therefore, a detection scheme capable of stably realizing profile extraction, reconstruction and model registration is needed. SUMMARY

[0005] The application solves the technical problem of overcoming the shortcomings of the prior art and providing a surface coating profile size detection method and device for a planar panel, which realizes high-precision detection of the surface coating profile size of the planar panel.

[0006] The technical solution of the application is as follows:

[0007] A surface coating profile size detection method for a planar panel, the surface of the planar panel being coated with a coating in part of the area, comprising the following steps:

[0008] S1, a positioning target with random texture is placed on the area without coating on the surface of the planar panel to be measured, and a binocular camera is used to collect images of the planar panel to be measured;

[0009] S2, the coating profile and the outer profile of the planar panel to be measured are extracted from the collected images;

[0010] S3, three-dimensional reconstruction is performed on the target area based on the collected images to obtain three-dimensional point cloud data of the surface of the target, and three-dimensional plane fitting is performed on the three-dimensional point cloud data to obtain a fitted space plane of the surface of the planar panel to be measured.

[0011] S4 takes the fitting space plane as a projection plane, and performs inverse projection calculation on the extracted coating profile and the outer profile of the to-be-measured planar part to obtain three-dimensional point cloud data of the coating profile and the outer profile of the to-be-measured planar part;

[0012] S5 performs dimension reduction processing on the three-dimensional point cloud data of the coating profile and the outer profile of the to-be-measured planar part to obtain two-dimensional point cloud data of the coating profile and the outer profile of the to-be-measured planar part;

[0013] S6 reads design data of the outer profile of the to-be-measured planar part from a design drawing, and performs registration on the two-dimensional point cloud data of the outer profile of the to-be-measured planar part and the design data of the outer profile of the to-be-measured planar part to obtain a coordinate conversion relationship that minimizes registration error;

[0014] S7 converts the two-dimensional point cloud data of the coating profile by using the coordinate conversion relationship obtained in step S6 to obtain measured data of the coating profile, reads design data of the coating profile from the design drawing, and obtains size error of the coating profile according to the measured data and the design data of the coating profile.

[0015] Preferably, in step S2, the coating profile and the outer profile of the to-be-measured planar part are extracted from the collected image, and specifically comprising:

[0016] S21 separates the to-be-measured planar part in the collected image from the background to extract the outer profile of the to-be-measured planar part with integral pixel precision;

[0017] S22 extracts the coating profile with integral pixel precision by using a Canny algorithm;

[0018] S23 calculates the gray gradient direction of the profile point with integral pixel precision by using a Hessian operator, and approximates it to an integral multiple of 45°, extracts two neighborhood points of the profile point with integral pixel precision in the approximated gray gradient direction, and obtains the profile point with sub-pixel precision by one-dimensional cubic interpolation, and the above method is used to process the coating profile and the outer profile of the to-be-measured planar part with integral pixel precision to obtain the coating profile and the outer profile of the to-be-measured planar part with sub-pixel precision.

[0019] Preferably, in step S21, the to-be-measured planar part in the collected image is separated from the background to extract the outer profile of the to-be-measured planar part with integral pixel precision, and specifically comprising: an Otsu algorithm is used to perform binaryzation processing on the collected image to obtain a binary image with only 0 and 255 gray scales, four vertices of the binary image are taken as seed points, the gray scale of the seed points is set to 1, the seed points are diffused to four neighborhood pixel points with 0 gray scale, and when the diffusion cannot continue, the gray scale of the remaining pixel points is set to 255, thereby obtaining a background segmentation image, and the boundary of the background segmentation image is extracted to obtain the outer profile of the to-be-measured planar part with integral pixel precision.

[0020] Preferably, in the step S3, the target region is reconstructed based on the collected images to obtain three-dimensional point cloud data of the target surface, specifically comprising:

[0021] S31 performs grid division on the target region of one of the two images collected by the binocular camera simultaneously, and selects a grid point as a matching point;

[0022] S32 takes the ZNSSD correlation coefficient as the matching criterion, and calculates the matching point in the other image by using the epipolar constraint to obtain the matching point pair with integer pixel accuracy;

[0023] S33 takes the matching point pair with integer pixel accuracy as the initial value, and calculates the matching point pair with sub-pixel accuracy by using the digital image correlation algorithm;

[0024] S34 calculates the three-dimensional point cloud data of the target surface by using the triangular method on the matching point pair with sub-pixel accuracy.

[0025] Preferably, in the step S5, the three-dimensional point cloud data of the coating profile and the outer profile of the to-be-measured planar panel is processed by dimension reduction, specifically comprising: constructing a dimension reduction conversion coordinate system, taking the point corresponding to the upper left corner of the outer profile of the to-be-measured planar panel in the three-dimensional point cloud as the origin, taking the normal of the fitted space plane as the z-axis direction, taking the point corresponding to the upper right corner of the outer profile of the to-be-measured planar panel in the three-dimensional point cloud as the point on the x-axis to determine the x-axis direction, and the y-axis and the x-axis and the z-axis form a right-handed system; converting the three-dimensional point cloud data of the coating profile and the outer profile of the to-be-measured planar panel from the original coordinate system to the dimension reduction conversion coordinate system, discarding the z-axis coordinate value, and obtaining the two-dimensional point cloud data of the coating profile and the outer profile of the to-be-measured planar panel.

[0026] Preferably, in the step S6, the two-dimensional point cloud data of the outer profile of the to-be-measured planar panel is registered with the design data of the outer profile of the to-be-measured planar panel to obtain the coordinate conversion relationship that minimizes the registration error, specifically comprising:

[0027] S61 defines the coordinate conversion variable p = [a, u, v] ', wherein u is the translation along the x-axis, v is the translation along the y-axis, and a is the angle of counterclockwise rotation with the origin as the center;

[0028] S62 constructs the coordinate conversion relationship based on the coordinate conversion variable:

[0029]

[0030] Wherein, (x, y) is the two-dimensional coordinate before conversion, and (x', y') is the two-dimensional coordinate after conversion;

[0031] S63 uses the coordinate conversion relationship constructed in step S62 to convert the two-dimensional point cloud data of the outer contour of the to-be-measured planar panel, and calculates the shortest distance from each contour point in the two-dimensional point cloud of the outer contour of the to-be-measured planar panel to the design data of the outer contour of the to-be-measured planar panel after coordinate conversion;

[0032] S64 constructs a registration error function:

[0033]

[0034] wherein dist(p i ) is the shortest distance from the i-th contour point in the two-dimensional point cloud data of the outer contour of the to-be-measured planar panel to the design data of the outer contour of the to-be-measured planar panel after coordinate conversion, i = 1, 2, …, N, N is the number of contour points in the two-dimensional point cloud data of the outer contour of the to-be-measured planar panel,

[0035] S65 optimizes and solves with the minimum value of the registration error function as the optimization target, to obtain the coordinate conversion variable that minimizes the registration error function, thereby obtaining the coordinate conversion relationship that minimizes the registration error.

[0036] Preferably, in step S65, Newton's method is used to optimize and solve the registration error function, and the coordinate conversion variable that meets the convergence condition is output.

[0037] Preferably, in step S5, the two-dimensional point cloud data of the coating contour and the outer contour of the to-be-measured planar panel extracted have the same unit as the coating contour and the design data of the outer contour of the to-be-measured planar panel in the design drawing.

[0038] A device for detecting the size of a coating contour on a surface of a planar panel, comprising a binocular industrial camera, a positioning target, and a detection host, wherein the positioning target has random texture features and is arranged in a non-coated area on the surface of the planar panel for feature positioning; the binocular industrial camera is arranged above the to-be-measured planar panel to collect images of the to-be-measured planar panel and transmit the collected images to the coating contour detection host; the detection host processes the collected images and detects the size error of the coating contour on the surface of the to-be-measured planar panel in combination with a design drawing of the to-be-measured planar panel, and the detection process specifically comprises:

[0039] extracting the coating contour and the outer contour of the to-be-measured planar panel from the collected images;

[0040] reconstructing the target area in three dimensions based on the collected images, obtaining three-dimensional point cloud data of the surface of the target, and fitting a three-dimensional plane to the three-dimensional point cloud data to obtain a fitted spatial plane on the surface of the to-be-measured planar panel;

[0041] The extracted coating profile and the outer profile of the to-be-measured planar panel are inversely projected by taking the fitted space plane as a projection plane to obtain three-dimensional point cloud data of the coating profile and the outer profile of the to-be-measured planar panel;

[0042] The three-dimensional point cloud data of the coating profile and the outer profile of the to-be-measured planar panel are processed in a dimension reduction manner to obtain two-dimensional point cloud data of the coating profile and the outer profile of the to-be-measured planar panel;

[0043] Design data of the outer profile of the to-be-measured planar panel is read from a design drawing, and the two-dimensional point cloud data of the outer profile of the to-be-measured planar panel is registered with the design data to obtain a coordinate conversion relationship minimizing registration error;

[0044] The two-dimensional point cloud data of the coating profile is converted by using the obtained coordinate conversion relationship to obtain measured data of the coating profile, design data of the coating profile is read from the design drawing, and size error of the coating profile is obtained according to the measured data and the design data.

[0045] Preferably, the method further comprises a measurement platform, a planar light source array and a light shield; the measurement platform has a dark surface to provide a background for the to-be-measured planar panel; the planar light source array comprises a series of planar light sources arranged around the measurement platform to provide uniform illumination for the to-be-measured planar panel; and the light shield completely covers the measurement platform, the binocular industrial camera, the planar light source array, the to-be-measured planar panel and the positioning target to avoid environmental light interference.

[0046] Compared with the prior art, the method has the following advantages:

[0047] (1) The method uses the outer profile to perform profile registration, and can realize registration of a measured profile of the to-be-measured part without fixed visual features and a design profile;

[0048] (2) The method extracts coating surface point cloud by using the positioning target in combination with the digital image correlation algorithm, can extract a large amount of high-precision point cloud data from a small number of targets, has a simple target pasting step, and has an accurate coating plane expression obtained by fitting; meanwhile, the method has good robustness, a very low mismatching rate and good result repeatability;

[0049] (3) The method has simple structure and low cost, and mainly comprises a measurement platform, a pair of industrial cameras, a planar light source array, a light shield, a positioning target and a host computer, all of which have no customization requirement, have large market supply and are easy to obtain. BRIEF DESCRIPTION OF DRAWINGS

[0050] Figure 1 The figure is a flowchart of a planar panel surface coating profile size detection method of the application;

[0051] Figure 2The schematic diagram of the device for detecting the profile size of the surface coating of the flat panel is shown in the figure. DETAILED DESCRIPTION

[0052] The features and advantages of the present application will become more apparent from the following detailed description, taken in conjunction with the accompanying drawings.

[0053] The present application provides a method for detecting the profile size of the surface coating of a flat panel, as shown in the figure, comprising the following steps: Figure 1

[0054] S1: Place a positioning target with random texture on the flat panel to be measured, collect the image of the flat panel to be measured, extract the coating profile with sub-pixel accuracy from the image, separate the measured object from the background, and obtain the outer contour of the measured object.

[0055] Separating the measured object from the background in the image and obtaining the outer contour of the measured object specifically comprises:

[0056] Otsu algorithm is used to perform binaryzation processing on the camera image to obtain a binary image with only 0 and 255 gray scales; the points at the four corners of the binary image are taken as seed points, the gray scale is set to 1, and the four-neighbor pixel points with 0 gray scale are diffused until no further diffusion is possible, then the gray scale of the remaining pixels is set to 255 to obtain a background segmentation graph. The boundary of the background segmentation graph is extracted to obtain the whole-pixel outer contour. Due to the influence of background noise and other factors, multiple contours may be extracted, and the one with the highest number of contour point sets is taken as the outer contour.

[0057] Extracting the coating profile with sub-pixel accuracy from the image specifically comprises:

[0058] Canny algorithm is used to extract the whole-pixel contour of the coating; Hessian operator is used to calculate the gray scale gradient direction of the whole-pixel contour points, and the direction is approximated to an integer multiple of 45°; two neighbor points of the contour points are extracted in this direction, and the sub-pixel contour point coordinates are obtained by one-dimensional cubic interpolation, thereby obtaining the sub-pixel coating contour and the outer contour.

[0059] The above profile extraction uses any one of the two images collected by the binocular camera.

[0060] S2: Extract the matching point pairs in the region of the positioning target in the image, perform high-precision three-dimensional reconstruction, obtain the three-dimensional point cloud data of the target surface, then perform three-dimensional plane fitting on the three-dimensional point cloud data, and obtain the spatial plane expression of the surface of the flat panel.

[0061] Extracting the matching point pairs in the region of the positioning target in the image and performing high-precision three-dimensional reconstruction to obtain the three-dimensional point cloud data of the target surface specifically comprises:

[0062] ​The selected target region in one of the two images collected by the binocular camera is meshed, and the mesh points are selected as the matching points, and the matching is performed according to the ZNSSD correlation coefficient;

[0063] The epipolar line in the target image after distortion is calculated according to the epipolar constraint, the correlation coefficient of each point on the epipolar line is calculated, and the pixel with the highest value is taken as the matching point of the integer pixel;

[0064] The matching point pair of the integer pixel is taken as the initial value, and the digital image correlation (DIC) algorithm is used to optimize and calculate the matching point pair of the sub-pixel accuracy;

[0065] The matching point pair is de-distorted, and the three-dimensional point cloud corresponding to the mesh point is calculated by the triangular method.

[0066] S3, inverse projection calculation is performed on the coating profile and the outer profile obtained in step S1, the projection plane is the spatial plane expression obtained in step S2; the spatial three-dimensional point cloud of the coating profile and the outer profile is calculated; the spatial three-dimensional point cloud is processed to obtain the actual coating profile and the outer profile two-dimensional point cloud with the same unit as the profile in the design drawing;

[0067] The spatial three-dimensional point cloud processing includes:

[0068] The point corresponding to the upper left corner of the outer profile in the three-dimensional point cloud is taken as the origin, the normal of the spatial plane is taken as the z axis, the point corresponding to the upper right corner of the outer profile in the point cloud is selected as the x axis point to determine the direction of the x axis, and the y axis is perpendicular to the x axis; the rotation matrix and the translation vector of the current point cloud coordinate system to the newly established coordinate are calculated, the point cloud coordinates are converted to the new coordinate system, at this time the z coordinate of the new point cloud is 0, and the coordinate direction is also close to the design drawing, and the z coordinate data is removed to complete the dimension reduction.

[0069] S4, reading the design profile in the design drawing, taking the outer profile two-dimensional point cloud obtained in step S3 as the registration object, calculating the coordinate conversion relationship that makes the outer profile reach the minimum registration error;

[0070] The read design drawing is a CAD file, and the obtained design profile is a set of profile elements, including: line segment, circle, circular arc, and polyline; since the algorithm uses the outer profile registration, the profile elements contained in the outer profile in the design profile need to be extracted for registration;

[0071] Step S4 specifically includes:

[0072] S401, defining the coordinate conversion variable p=[α,u,v]' and the coordinate conversion relationship:

[0073]

[0074] where u is the translation along x-axis, v is the translation along y-axis, and a is the angle of counter-clockwise rotation with the origin as the center; [a, u, v]' represents the transpose of [a, u, v]; (x, y) is the two-dimensional coordinate before transformation, and (x', y') is the two-dimensional coordinate after transformation.

[0075] Definition is the sum of the minimum distances from all the coordinate-transformed coating profile points to the design profile, as the registration error function, where dist(p i ) is the shortest distance calculated by the i-th coordinate-transformed coating profile point traversing all the profile elements in the design profile;

[0076] S402, based on Newton method, the iterative increment is calculated as follows:

[0077]

[0078] According to Newton method, the third-order Hessian matrix and the one-dimensional Jacobian vector are composed of the second-order partial derivative and the first-order partial derivative of the error registration function with respect to the coordinate transformation variables.

[0079] As e' u represents the first-order partial derivative of e(p) with respect to u, e" uv represents the second-order partial derivative of e(p) with respect to u and v.

[0080] Since it is difficult to derive, the elements in the above formula can be replaced as follows, taking the first-order partial derivative of e(p) with respect to u and the second-order partial derivative of e(p) with respect to u and v as examples:

[0081]

[0082]

[0083] where Δu and Δv are the difference steps, which are taken as very small positive values, and when Δu and Δv tend to 0, the difference is equivalent to the derivative.

[0084] S403, starting from a certain initial value, iteratively update p, until the convergence condition (e(p n -e(p n-1 ) / e(p n-1 < 10 -6 is satisfied, and the coordinate transformation relationship is obtained; where e(p n is the registration error function value of the n-th iteration, and e(p n-1 is the registration error function value of the n-1-th iteration.

[0085] Wherein the initial value of the translation variable can be obtained by averaging the x, y coordinates of the point cloud, and the angle variable is set to 0; it is necessary to place the measured flat panel in the direction of the design drawing according to the foregoing steps: by placing the measured flat panel in the direction of the design drawing; when reducing the dimension, the x-axis direction is reasonably selected to ensure that the measured profile is close enough to the design profile in terms of turning (the difference is less than 45°);

[0086] S5, using the coordinate transformation relationship obtained in step S4, performing coordinate transformation on the actual coating profile point cloud in step S3 to register with the design profile in the design drawing, and calculating the profile detection error.

[0087] The application also provides a flat panel surface coating profile size detection device, as shown in the drawings, the detection device comprises: Figure 2

[0088] The measurement platform 1 has a dark surface to provide a high-contrast background;

[0089] The industrial camera 2 is a binocular system composed of two cameras and is arranged laterally above the measurement platform;

[0090] The planar light source array 3 comprises a series of planar light sources arranged laterally around the measurement platform to provide uniform illumination conditions and avoid the problem of light reflection;

[0091] The positioning target 4 has random texture and features for positioning the coating surface by the software algorithm;

[0092] The light shield 5 completely covers the measured part and the camera to avoid environmental light interference;

[0093] The detection host 6 comprises algorithms such as profile extraction, three-dimensional reconstruction, and model registration to extract and detect the size of the coating profile from the camera image.

[0094] Specifically, a pair of industrial cameras 2 is arranged laterally above the measurement platform 1 for imaging. The measured part is smaller than the measurement platform 1, and the measurement platform 1 has a dark color to increase the contrast with the measured part. The planar light source array 3 is arranged laterally around the measurement platform 1 to avoid the interference of the mirror reflection of the measured object with the imaging quality; the light shield 5 covers the on-site facilities including the measurement platform 1, the pair of industrial cameras 2, and the planar light source array 3 to block the interference of the environmental light in the test site. The host 6 is located outside the light shield and is connected with the industrial cameras 2 to control the image acquisition of the industrial cameras 2 and comprises algorithms such as profile extraction, three-dimensional reconstruction, and model registration to extract and detect the size information of the coating profile from the camera image.

[0095] ​When measuring, the measured piece is placed on the measuring platform 1 in the direction shown in the design drawing, and the surface is placed away from the coating profile to position the target 4, which is used for the software algorithm to position the spatial position of the surface of the measured piece; then, the host 6 controls the industrial camera 2 to collect the image of the measured piece, the image is an 8-bit gray image, and the content only contains the measured piece and the background, and cannot be collected outside the background provided by the measuring platform 1, and then the image is transmitted back to the host 6 for calculation.

[0096] The contents not described in detail in the specification of the present application are the known technology of those skilled in the art.

Claims

1. A method for detecting the contour dimensions of a coating on the surface of a planar plate, wherein a coating is applied to a portion of the surface of the planar plate, characterized in that... Includes the following steps: S1 places a positioning target with random texture in an uncoated area on the surface of the flat plate to be tested, and uses a binocular camera to acquire images of the flat plate to be tested. S2 extracts the coating contour and the outer contour of the flat plate to be tested from the acquired image; S3 performs three-dimensional reconstruction of the target area based on the acquired image, obtains three-dimensional point cloud data of the target surface, performs three-dimensional plane fitting on the three-dimensional point cloud data, and obtains the fitting space plane of the surface of the plane plate to be tested. S4 uses the fitted spatial plane as the projection plane to perform inverse projection calculations on the extracted coating contour and the outer contour of the planar plate under test, thereby obtaining the three-dimensional point cloud data of the coating contour and the outer contour of the planar plate under test. S5 performs dimensionality reduction processing on the three-dimensional point cloud data of the coating profile and the outer contour of the planar plate under test to obtain two-dimensional point cloud data of the coating profile and the outer contour of the planar plate under test. S6 reads the design data of the outer contour of the planar plate to be measured from the design drawings, registers the two-dimensional point cloud data of the outer contour of the planar plate to be measured with the design data of the outer contour of the planar plate to be measured, and obtains the coordinate transformation relationship that minimizes the registration error. S7 uses the coordinate transformation relationship obtained in step S6 to transform the two-dimensional point cloud data of the coating profile to obtain the measured data of the coating profile. The design data of the coating profile is read from the design drawings, and the dimensional error of the coating profile is obtained based on the measured data and design data of the coating profile.

2. The method for detecting the contour dimensions of a coating on a planar plate according to claim 1, characterized in that, In step S2, the coating contour and the outer contour of the planar plate to be tested are extracted from the acquired image, specifically including: S21 separates the test planar plate from the background in the acquired image and extracts the outer contour of the test planar plate with integer pixel precision. S22 uses the Canny algorithm to extract the coating contour with integer pixel precision; S23 uses the Hessian operator to calculate the gray-level gradient direction of the contour points with integer pixel precision, and approximates it as an integer multiple of 45°. It then extracts two neighboring points of the contour points with integer pixel precision from the approximate gray-level gradient direction. Sub-pixel precision contour points are obtained through one-dimensional cubic interpolation. The above methods are used to process the coating contour with integer pixel precision and the outer contour of the planar plate under test, respectively, to obtain the coating contour with sub-pixel precision and the outer contour of the planar plate under test.

3. The method for detecting the contour dimensions of a coating on a planar plate according to claim 2, characterized in that, In step S21, the planar component to be tested in the acquired image is separated from the background, and the outer contour of the planar component to be tested with integer pixel precision is extracted. Specifically, the acquired image is binarized using the Otsu algorithm to obtain a binary image with gray levels of only 0 and 255. The four vertices of the binary image are used as seed points. The gray level of the seed points is set to 1 and diffused to the four neighboring pixels with gray levels of 0 until it can no longer diffuse. Then, the gray level of the remaining pixels is set to 255 to obtain a background segmentation image. The boundary of the background segmentation image is extracted to obtain the outer contour of the planar component to be tested with integer pixel precision.

4. The method for detecting the contour dimensions of a coating on a planar plate according to claim 1, characterized in that, In step S3, the target region is reconstructed in three dimensions based on the acquired image to obtain three-dimensional point cloud data of the target surface, specifically including: S31 divides the target area of ​​one of the two images simultaneously acquired by the binocular camera into a grid and selects grid points as matching points. S32 uses the ZNSSD correlation coefficient as the matching criterion and employs epipolar constraints to calculate the matching points in another image, obtaining matching point pairs with integer pixel precision. S33 uses matching point pairs with integer pixel precision as initial values ​​and employs a digital image correlation algorithm to optimize and calculate matching point pairs with sub-pixel precision. S34 uses triangulation to calculate the three-dimensional point cloud data of the target surface for sub-pixel precision matching point pairs.

5. The method for detecting the contour dimensions of a coating on a planar plate according to claim 1, characterized in that, In step S5, the 3D point cloud data of the coating contour and the outer contour of the planar plate to be tested are subjected to dimensionality reduction processing. Specifically, a dimensionality reduction transformation coordinate system is constructed. The dimensionality reduction transformation coordinate system takes the point corresponding to the upper left corner of the outer contour of the planar plate to be tested in the 3D point cloud as the origin, the normal of the fitted spatial plane as the z-axis direction, and the point corresponding to the upper right corner of the outer contour of the planar plate to be tested in the 3D point cloud as the point on the x-axis to determine the x-axis direction. The y-axis forms a right-handed system with the x-axis and z-axis. The 3D point cloud data of the coating contour and the outer contour of the planar plate to be tested are transformed from the original coordinate system to the dimensionality reduction transformation coordinate system, and the z-axis coordinate value is discarded to obtain the 2D point cloud data of the coating contour and the outer contour of the planar plate to be tested.

6. The method for detecting the contour dimensions of a coating on a planar plate according to claim 1, characterized in that, In step S6, the two-dimensional point cloud data of the outer contour of the planar plate to be measured is registered with the design data of the outer contour of the planar plate to be measured to obtain the coordinate transformation relationship that minimizes the registration error. Specifically, this includes: S61 defines the coordinate transformation variable p = [α, u, v]', where u is the translation along the x-axis, v is the translation along the y-axis, and α is the angle of counterclockwise rotation around the origin; S62 constructs coordinate transformation relationships based on coordinate transformation variables: Where (x,y) are the two-dimensional coordinates before transformation, and (x',y') are the two-dimensional coordinates after transformation; S63 uses the coordinate transformation relationship constructed in step S62 to transform the two-dimensional point cloud data of the outer contour of the planar plate to be measured, and calculates the shortest distance from each contour point in the two-dimensional point cloud of the outer contour of the planar plate to be measured to the design data of the outer contour of the planar plate after coordinate transformation. S64 constructs the registration error function: Among them, dist(p i Let Σ(i=1,2...N) represent the shortest distance from the i-th contour point in the 2D point cloud data of the outer contour of the planar plate to the design data of the outer contour of the planar plate after coordinate transformation, where i = 1,2...N, and N is the number of contour points in the 2D point cloud data of the outer contour of the planar plate. S65 optimizes the solution by minimizing the registration error function value, thereby obtaining the coordinate transformation variable that minimizes the registration error function value, and thus finding the coordinate transformation relationship that minimizes the registration error.

7. The method for detecting the contour dimensions of a coating on a planar plate according to claim 6, characterized in that, In step S65, the registration error function is optimized using Newton's method, and the coordinate transformation variables that satisfy the convergence condition are output.

8. A method for detecting the contour dimensions of a coating on a planar plate according to any one of claims 1 to 7, characterized in that, In step S5, the extracted two-dimensional point cloud data of the coating contour and the outer contour of the planar plate to be measured have the same units as the design data of the coating contour and the outer contour of the planar plate to be measured in the design drawings.

9. A device for detecting the contour dimensions of a coating on a planar plate, characterized in that, The system includes a binocular industrial camera (2), a positioning target (4), and a detection host (6). The positioning target (4) has random texture features and is set in an uncoated area on the surface of the planar plate for feature localization. The binocular industrial camera (2) is positioned above the planar plate to be tested, acquires images of the planar plate, and transmits the acquired images to the coating contour detection host (6). The detection host (6) processes the acquired images and, in conjunction with the design drawings of the planar plate, detects the dimensional error of the coating contour on the surface of the planar plate. The detection process specifically includes: Extract the coating contour and the outer contour of the planar plate to be tested from the acquired images; Based on the acquired images, the target area is reconstructed in three dimensions to obtain the three-dimensional point cloud data of the target surface. The three-dimensional point cloud data is then fitted in three dimensions to obtain the fitting space plane of the surface of the planar plate to be tested. Using the fitted spatial plane as the projection plane, the extracted coating contour and the outer contour of the planar plate under test are back-projected to obtain the three-dimensional point cloud data of the coating contour and the outer contour of the planar plate under test. The 3D point cloud data of the coating profile and the outer contour of the planar plate under test are reduced in dimension to obtain the 2D point cloud data of the coating profile and the outer contour of the planar plate under test. The design data of the outer contour of the planar plate to be measured is read from the design drawings. The two-dimensional point cloud data of the outer contour of the planar plate to be measured is registered with the design data of the outer contour of the planar plate to be measured, so as to obtain the coordinate transformation relationship that minimizes the registration error. The obtained coordinate transformation relationship is used to transform the two-dimensional point cloud data of the coating profile to obtain the measured data of the coating profile. The design data of the coating profile is read from the design drawings. The dimensional error of the coating profile is obtained based on the measured data and the design data.

10. The device for detecting the contour dimensions of a surface coating on a planar plate according to claim 9, characterized in that, It also includes a measurement platform (1), a planar light source array (3), and a light shield (5); the measurement platform (1) has a dark surface to provide a background for the planar plate to be measured; the planar light source array (3) includes a series of planar light sources arranged around the measurement platform (1) to provide uniform illumination for the planar plate to be measured; the light shield (5) completely covers the measurement platform (1), the binocular industrial camera (2), the planar light source array (3), the planar plate to be measured, and the positioning target (4) to avoid ambient light interference.

Images