Hyperboloid plate finished product acceptance inspection method and system

Abstract

The invention discloses a hyperboloid plate finished product acceptance inspection detection method and system, and belongs to the technical field of plate detection. Ideal required hyperboloid plates and actual hyperboloid plates are respectively divided into a plurality of ideal single curved plates and a plurality of actual single curved plates, and corresponding ideal single-plate three-dimensional point cloud models and actual single-plate three-dimensional point cloud models are obtained; after multiple matching point pairs are obtained and pose alignment is carried out, the qualification condition of the actual hyperboloid plate is analyzed according to the curvature deviation of each matching point pair, overall detection is converted into local single plate analysis, a detection blind area caused by hyperboloid geometric complexity is effectively solved, the pose deviation between the actual model and the ideal model is eliminated through pose alignment, and the detection accuracy is improved. Consistency of comparison references is ensured, and detection accuracy is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of sheet material detection, and particularly to a method and system for acceptance inspection of hyperbolic panel finished products. Background Art

[0002] A hyperbolic panel is a structural member with a complex geometric shape, and its surface has hyperbolic characteristics. A hyperbolic surface is a three-dimensional surface generated by two independent curves (usually parabolas, ellipses or hyperbolas), so the hyperbolic panel presents a non-planar curved shape in space. This special geometric shape makes it have wide application value in the fields of aerospace, shipbuilding, building structures, etc.

[0003] Therefore, the acceptance inspection of hyperbolic panels is extremely important. Because the core characteristics of hyperbolic panels lie in their complex geometric shapes and surface characteristics, curvature detection plays a crucial role in the acceptance inspection of hyperbolic panels. Through curvature detection, it can be ensured that the geometric accuracy, mechanical properties and function realization of hyperbolic panels meet the design requirements; Therefore, how to improve the accuracy of curvature detection of hyperbolic panels has become an urgent problem to be solved. Summary of the Invention

[0004] In order to solve the technical problems existing in the prior art, the present invention provides a method for acceptance inspection of hyperbolic panel finished products, including the following steps: Obtain an ideal three-dimensional point cloud model and an actual three-dimensional point cloud model respectively; the ideal three-dimensional point cloud model and the actual three-dimensional point cloud model are the three-dimensional point cloud models of the ideal required hyperbolic panel and the actual hyperbolic panel respectively; Divide the ideal required hyperbolic panel and the actual hyperbolic panel into multiple ideal single-curved plates and multiple actual single-curved plates respectively, and obtain the corresponding ideal single-plate three-dimensional point cloud models and actual single-plate three-dimensional point cloud models; the ideal single-curved plates and the actual single-curved plates correspond one by one; Map the actual single-plate three-dimensional point cloud model and the corresponding ideal single-plate three-dimensional point cloud model to the same coordinate system, obtain multiple matching point pairs, and perform pose alignment; Analyze the curvature deviation of each matching point pair between the actual single-plate three-dimensional point cloud model and the corresponding ideal single-plate three-dimensional point cloud model after pose alignment; If (z1 / Z)≥YZ, and / or JZ is greater than or equal to YJZ, then the corresponding actual single-curved plate is unqualified. If any actual single-curved plate is unqualified, then the corresponding actual hyperbolic panel is unqualified; z1 represents the number of matching point pairs with a curvature deviation greater than or equal to the preset deviation threshold, Z represents the number of matching point pairs, YZ represents the preset ratio, JZ represents the average value of the sum of the curvature deviations of each matching point pair, and YJZ represents the preset deviation average value.

[0005] Further, the ideal three-dimensional point cloud model is constructed and obtained according to the design parameters of the ideal required hyperbolic panel; The acquisition of the actual three-dimensional point cloud model is specifically as follows: Generate multiple grid regions on the surface of the actual hyperbolic panel; Perform three-dimensional scanning on the actual hyperbolic panel to obtain the initial point cloud of each grid region and analyze the complex regions; Adaptively adjust the scanning step size and perform point cloud encryption scanning on the complex regions.

[0006] Further, the analysis of the complex regions is specifically as follows: Calculate the curvature entropy value of each grid region, and regard the grid regions with curvature entropy values greater than or equal to the preset threshold as complex regions; The curvature entropy value: ; where h represents the curvature entropy value, represents the curvature value at the i-th initial point position in the corresponding grid region, n represents the number of initial points of the initial point cloud in the corresponding grid region, and i is an index symbol.

[0007] Further, the adaptive adjustment of the scanning step size and the point cloud encryption scanning of the complex regions are specifically as follows: ; where, represents the scanning step size used when scanning from the i-th initial point position to other initial point positions in the corresponding grid region, represents the curvature gradient at the i-th initial point position in the corresponding grid region, ||·|| represents the modulus calculation, and i is an index symbol.

[0008] Further, map the actual single-board three-dimensional point cloud model and the corresponding ideal single-board three-dimensional point cloud model to the same coordinate system, obtain multiple matching point pairs, and perform pose alignment, which specifically includes the following steps: Extract feature points from the actual single-board three-dimensional point cloud model and the corresponding ideal single-board three-dimensional point cloud model, and perform feature description on the feature points; Match the feature points of the actual single-board three-dimensional point cloud model with the feature points of the corresponding ideal single-board three-dimensional point cloud model through a feature matching algorithm to obtain multiple matching point pairs; According to the matching point pairs, use the SVD algorithm to analyze the rotation matrix R and the translation vector t, and transform the actual single-board three-dimensional point cloud model according to the rotation matrix and the translation vector to complete the pose alignment with the corresponding ideal single-board three-dimensional point cloud model.

[0009] Further, after obtaining multiple matching point pairs, it also includes using the random sample consensus algorithm to remove the mismatched point pairs.

[0010] Further, the rotation matrix R and the translation vector t are analyzed using the SVD algorithm, specifically as follows: Obtain the centroid coordinate positions of the actual single-board three-dimensional point cloud model and the corresponding ideal single-board three-dimensional point cloud model, denoted as P1 and P2 respectively; Analyze the covariance matrix H based on P1 and P2; Perform SVD decomposition on the covariance matrix to obtain the left singular vector matrix U and the right singular vector matrix V; Analyze the rotation matrix R and the translation vector t based on the left singular vector matrix U and the right singular vector matrix V: ; T represents matrix transpose.

[0011] Further, the transformation of the actual single-board three-dimensional point cloud model according to the rotation matrix and the translation vector is specifically as follows: ; represents the coordinate position of the i-th point after the transformation of the actual single-board three-dimensional point cloud model, where i is an index symbol.

[0012] Further, the curvature deviation of each matching point pair between the actual single-board three-dimensional point cloud model after pose alignment and the corresponding ideal single-board three-dimensional point cloud model is analyzed, specifically as follows: In the actual three-dimensional point cloud model after pose alignment and the corresponding ideal single-board three-dimensional point cloud model respectively, obtain the maximum principal curvature and the minimum principal curvature of each point belonging to the matching point pair; Calculate the Gaussian curvature of each point belonging to the matching point pair: ; Calculate the curvature deviation of each matching point pair respectively: ; Among them, represents the Gaussian curvature of the i-th point belonging to the matching point pair in the actual three-dimensional point cloud model after pose alignment, represents the maximum principal curvature of the i-th point belonging to the matching point pair in the actual three-dimensional point cloud model after pose alignment, represents the minimum principal curvature of the i-th point belonging to the matching point pair in the actual three-dimensional point cloud model after pose alignment, represents the Gaussian curvature of the i-th point belonging to the matching point pair in the corresponding ideal single-board three-dimensional point cloud model, represents the maximum principal curvature of the i-th point belonging to the matching point pair in the corresponding ideal single-board three-dimensional point cloud model, represents the minimum principal curvature of the i-th point belonging to the matching point pair in the corresponding ideal single-board three-dimensional point cloud model, represents the curvature deviation of the i-th matching point pair, where i is an index symbol.

[0013] The present invention also provides a finished product acceptance detection system for a hyperbolic panel, including: A point cloud model acquisition module for respectively acquiring an ideal three-dimensional point cloud model and an actual three-dimensional point cloud model; the ideal three-dimensional point cloud model and the actual three-dimensional point cloud model are respectively the three-dimensional point cloud models of the ideal required hyperbolic panel and the actual hyperbolic panel; A single-curved plate division module for respectively dividing the ideal required hyperbolic panel and the actual hyperbolic panel into a plurality of ideal single-curved plates and a plurality of actual single-curved plates, and acquiring the corresponding ideal single-plate three-dimensional point cloud models and actual single-plate three-dimensional point cloud models; the ideal single-curved plates and the actual single-curved plates correspond one by one; A matching point acquisition and pose alignment module for mapping the actual single-plate three-dimensional point cloud model and the corresponding ideal single-plate three-dimensional point cloud model to the same coordinate system, acquiring a plurality of matching point pairs, and performing pose alignment; A calculation and judgment module for analyzing the curvature deviation of each matching point pair between the actual single-plate three-dimensional point cloud model and the corresponding ideal single-plate three-dimensional point cloud model after pose alignment; if (z1 / Z)≥YZ, and / or JZ is greater than or equal to YJZ, then the corresponding actual single-curved plate is unqualified, and if any actual single-curved plate is unqualified, then the corresponding actual hyperbolic panel is unqualified; z1 represents the number of matching point pairs with a curvature deviation greater than or equal to the preset deviation threshold, Z represents the number of matching point pairs, YZ represents the preset ratio, JZ represents the mean value of the sum of the curvature deviations of each matching point pair, and YJZ represents the preset deviation mean Compared with the prior art, the beneficial effects of the present invention are as follows: By respectively dividing the ideal required hyperbolic panel and the actual hyperbolic panel into a plurality of ideal single-curved plates and a plurality of actual single-curved plates, and acquiring the corresponding ideal single-plate three-dimensional point cloud models and actual single-plate three-dimensional point cloud models, then acquiring a plurality of matching point pairs and performing pose alignment, and analyzing the qualification of the actual hyperbolic panel according to the curvature deviation of each matching point pair, the overall detection is transformed into local single-plate analysis, effectively solving the detection blind area caused by the geometric complexity of the hyperbolic surface, and eliminating the pose deviation between the actual and ideal models through pose alignment to ensure the consistency of the comparison benchmark and improve the detection accuracy; Regarding the region with a curvature entropy value greater than or equal to the preset threshold as a complex region, that is, in the high-entropy region, further encrypted scanning is performed, which can ensure that the region with a large curvature change is fully detected, and the scanning step size is adaptively adjusted according to each curvature gradient, reducing the scanning step size in the region with a large curvature change to increase the scanning density, and increasing the scanning step size in the region with a gentle curvature change to reduce the scanning density, thereby adapting to the curvature change near the current point, improving the scanning accuracy and efficiency, and thus obtaining an accurate three-dimensional point cloud model for the actual hyperbolic panel to ensure the accuracy of the finished product detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The accompanying drawings here are incorporated into the specification and form a part of this specification, showing the embodiments in line with the present invention and, together with the specification, used to explain the principles of the present invention.

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings required for use in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0016] Figure 1 is a flowchart of a method for accepting and inspecting the finished product of a hyperbolic panel of the present invention; Figure 2 is a flowchart of a method for obtaining an actual three-dimensional point cloud model in a method for accepting and inspecting the finished product of a hyperbolic panel of the present invention; Figure 3 is a flowchart of step S3 in a method for accepting and inspecting the finished product of a hyperbolic panel of the present invention. Detailed implementation manners

[0017] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention.

[0018] It should be noted that all directional indications (such as up, down, left, right, front, back...) in the embodiments of the present invention are only used to explain the relative positional relationship and movement conditions between components in a specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indications will also change accordingly.

[0019] In addition, the descriptions involving "first", "second", etc. in the present invention are only for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include at least one of such features. In addition, the technical solutions between various embodiments can be combined with each other, but it must be based on the fact that those of ordinary skill in the art can implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0020] Embodiment 1 Refer to Figure 1 As shown, a method for accepting and inspecting the finished product of a hyperbolic panel provided by the present invention specifically includes the following steps: S1. Obtain the ideal three-dimensional point cloud model and the actual three-dimensional point cloud model respectively; The ideal three-dimensional point cloud model and the actual three-dimensional point cloud model are the three-dimensional point cloud models of the ideal required hyperbolic panel and the actual hyperbolic panel respectively; S2. Divide the ideal required hyperbolic panel and the actual hyperbolic panel into multiple ideal single-curved plates and multiple actual single-curved plates respectively, and obtain the corresponding ideal single-plate three-dimensional point cloud model and actual single-plate three-dimensional point cloud model; The ideal single-curved plates and the actual single-curved plates correspond one by one; S3. Map the actual single-plate three-dimensional point cloud model and the corresponding ideal single-plate three-dimensional point cloud model to the same coordinate system, obtain multiple matching point pairs, and perform pose alignment; S4. Analyze the curvature deviation of each matching point pair between the actual single-plate three-dimensional point cloud model and the corresponding ideal single-plate three-dimensional point cloud model after pose alignment; S5. If (z1 / Z)≥YZ, and / or JZ is greater than or equal to YJZ, then the corresponding actual single-curved plate is unqualified. If any actual single-curved plate is unqualified, then the corresponding actual hyperbolic panel is unqualified; z1 represents the number of matching point pairs with curvature deviation greater than or equal to the preset deviation threshold, Z represents the number of matching point pairs, YZ represents the preset ratio, JZ represents the mean value of the sum of the curvature deviations of each matching point pair, and YJZ represents the preset deviation mean value.

[0021] In step S1, the ideal three-dimensional point cloud model is specifically obtained according to the design parameters of the ideal required hyperbolic panel; Refer to Figure 2 As shown, the acquisition method of the actual three-dimensional point cloud model specifically includes the following steps: S11. Generate multiple grid regions on the surface of the actual hyperbolic panel; S12. Perform three-dimensional scanning on the actual hyperbolic panel to obtain the initial point cloud of each grid region and analyze the complex regions; S13. Adaptively adjust the scanning step size and perform point cloud encryption scanning on the complex regions.

[0022] In step S11, generating multiple grid regions on the surface of the actual hyperbolic panel is specifically generated by an existing grid generation algorithm.

[0023] In step S12, the analysis of the complex regions is specifically: Calculate the curvature entropy value of each grid region, and regard the grid region with curvature entropy value greater than or equal to the preset threshold as the complex region; The acquisition method of the curvature entropy value is: where h represents the curvature entropy value, represents the curvature value of the \(i\)-th initial point position in the corresponding grid region, \(n\) represents the number of initial points of the initial point cloud in the corresponding grid region, and \(i\) is an index symbol.

[0024] In step S13, the adaptive adjustment of the scanning step size is performed to perform point cloud encryption scanning on complex regions, specifically as follows: Among them, represents the scanning step size used when scanning from the \(i\)-th initial point position to other initial point positions in the corresponding grid region, represents the curvature gradient of the \(i\)-th initial point position in the corresponding grid region, and \(\|\cdot\|\) represents the modulus calculation, and \(i\) is an index symbol.

[0025] Regions with curvature entropy values greater than or equal to the preset threshold are regarded as complex regions, that is, in high-entropy regions, further encryption scanning is performed, which can ensure that regions with large curvature changes are fully detected, and the scanning step size is adaptively adjusted according to each curvature gradient, reducing the scanning step size in regions with drastic curvature changes to increase the scanning density, and increasing the scanning step size in regions with gentle curvature changes to reduce the scanning density, thereby adapting to the curvature changes near the current point, improving the scanning accuracy and efficiency, so as to obtain an accurate three-dimensional point cloud model for the actual hyperbolic panel and ensure the accuracy of finished product detection.

[0026] In step S2, the ideal single-board three-dimensional point cloud model and the actual single-board three-dimensional point cloud model are obtained by dividing the ideal three-dimensional point cloud model and the actual three-dimensional point cloud model while dividing the ideal required hyperbolic panel and the actual hyperbolic panel into multiple ideal single-curved plates and multiple actual single-curved plates respectively.

[0027] Refer to Figure 3 As shown, in step S3, the actual single-board three-dimensional point cloud model and the corresponding ideal single-board three-dimensional point cloud model are mapped to the same coordinate system to obtain multiple matching point pairs and perform pose alignment, which specifically includes the following steps: S31. Extract feature points from the actual single-board three-dimensional point cloud model and the corresponding ideal single-board three-dimensional point cloud model, and perform feature description on the feature points; S32. Match the feature points of the actual single-board three-dimensional point cloud model with the feature points of the corresponding ideal single-board three-dimensional point cloud model through a feature matching algorithm to obtain multiple matching point pairs; S33. According to the matching point pairs, use the SVD algorithm to analyze the rotation matrix \(R\) and the translation vector \(t\), and transform the actual single-board three-dimensional point cloud model according to the rotation matrix and the translation vector to complete the pose alignment with the corresponding ideal single-board three-dimensional point cloud model.

[0028] When mapping the actual single-board three-dimensional point cloud model constructed by scanning and the ideal single-board three-dimensional point cloud model constructed by data into a coordinate system, there may be differences in position and attitude. The position difference will introduce additional errors, resulting in inaccurate subsequent curvature calculations, while the attitude difference will affect the local characteristics of the curvature and cannot truly reflect the geometric differences between the actual single-curved board and the ideal single-curved board. In this solution, the actual single-board three-dimensional point cloud model is aligned with the corresponding ideal single-board three-dimensional point cloud model in terms of position and attitude, thereby ensuring the basic consistency of curvature analysis, further improving the detection accuracy, and guaranteeing the effectiveness of the acceptance inspection of the double-curved panel finished product.

[0029] In step S31, for the actual single-board three-dimensional point cloud model and the corresponding ideal single-board three-dimensional point cloud model, feature point extraction is specifically performed through a key point detection algorithm. The key point detection algorithm can be selected from the Harris3D algorithm, ISS algorithm, etc. Feature points are usually areas with large curvature changes or inflection points of geometric shapes.

[0030] For the feature description of the feature points, it is specifically performed through the FPFH descriptor.

[0031] In step S32, the feature matching algorithm can be selected from the nearest neighbor search algorithm (such as KD-tree), the matching algorithm based on the Euclidean distance, that is, points with a Euclidean distance less than the set distance threshold are matched.

[0032] In some embodiments, step S32 further includes: After obtaining multiple pairs of matching points, the random sample consensus algorithm (RANdom SAmple Consensus, RANSAC) is used to eliminate the mismatched point pairs. The specific implementation is the prior art and will not be elaborated here.

[0033] In step S33, the rotation matrix R and the translation vector t are analyzed using the SVD algorithm, specifically as follows: S331. Obtain the centroid coordinate positions of the actual single-board three-dimensional point cloud model and the corresponding ideal single-board three-dimensional point cloud model, denoted as P1 and P2 respectively: Among them, N1 represents the number of points in the actual single-board three-dimensional point cloud model, N2 represents the number of points in the corresponding ideal single-board three-dimensional point cloud model, represents the coordinate position of the i-th point in the actual single-board three-dimensional point cloud model, represents the coordinate position of the i-th point in the corresponding ideal single-board three-dimensional point cloud model, and i is an index symbol; S332. Analyze the covariance matrix H according to P1 and P2: Among them, N3 represents the number of matching point pairs between the actual single-board three-dimensional point cloud model and the corresponding ideal single-board three-dimensional point cloud model. represents the position coordinates of the i-th point belonging to the actual single-board three-dimensional point cloud model in the matching point pair. represents the position coordinates of the i-th point belonging to the corresponding ideal single-board three-dimensional point cloud model in the matching point pair. T represents matrix transpose, and i is an index symbol. S334. Perform SVD decomposition on the covariance matrix to obtain the left singular vector matrix U and the right singular vector matrix V. The above SVD decomposition (Singular Value Decomposition) is an important matrix decomposition method in linear algebra. The specific implementation is prior art and will not be elaborated here. S335. Analyze the rotation matrix R and the translation vector t based on the left singular vector matrix U and the right singular vector matrix V: Among them, T represents matrix transpose.

[0034] In step S33, the transformation of the actual single-board three-dimensional point cloud model according to the rotation matrix and the translation vector is specifically as follows: represents the coordinate position of the i-th point after the transformation of the actual single-board three-dimensional point cloud model, and i is an index symbol.

[0035] In step S4, the analysis of the curvature deviation of each matching point pair between the actual single-board three-dimensional point cloud model after pose alignment and the corresponding ideal single-board three-dimensional point cloud model is specifically as follows: S41. Respectively, in the actual three-dimensional point cloud model after pose alignment and the corresponding ideal single-board three-dimensional point cloud model, obtain the maximum principal curvature and the minimum principal curvature of each point belonging to the matching point pair. S42. Calculate the Gaussian curvature of each point belonging to the matching point pair: Among them, represents the Gaussian curvature of the i-th point belonging to the matching point pair in the actual three-dimensional point cloud model after pose alignment. represents the maximum principal curvature of the i-th point belonging to the matching point pair in the actual three-dimensional point cloud model after pose alignment, and represents represents the minimum principal curvature of the i-th point belonging to the matching point pair in the actual three-dimensional point cloud model after pose alignment. represents the Gaussian curvature of the i-th point belonging to the matching point pair in the corresponding ideal single-board three-dimensional point cloud model. denote the maximum principal curvature of the i-th point belonging to the matching point pairs in the corresponding ideal single-panel three-dimensional point cloud model, denote the minimum principal curvature of the i-th point belonging to the matching point pairs in the corresponding ideal single-panel three-dimensional point cloud model, where i is an index symbol; S43. Calculate the curvature deviation of each matching point pair respectively: wherein, denote the curvature deviation of the i-th matching point pair.

[0036] The calculation of the principal curvature of each point in the three-dimensional point cloud usually depends on the local neighborhood information of the point. After pose alignment, the coordinate position information of the local neighborhood points changes. Therefore, the principal curvature of each point is calculated using the actual single-panel three-dimensional point cloud model after pose alignment to improve the accuracy of calculating the curvature deviation.

[0037] The specific implementation of obtaining the principal curvature of each point according to the three-dimensional point cloud is the prior art and will not be elaborated here.

[0038] Embodiment 2 The present invention also provides a hyperbolic panel finished product acceptance detection system, specifically including: A point cloud model acquisition module, used to respectively acquire an ideal three-dimensional point cloud model and an actual three-dimensional point cloud model; the ideal three-dimensional point cloud model and the actual three-dimensional point cloud model are respectively the three-dimensional point cloud models of the ideal required hyperbolic panel and the actual hyperbolic panel; A single-panel division module, used to divide the ideal required hyperbolic panel and the actual hyperbolic panel into multiple ideal single panels and multiple actual single panels respectively, and acquire the corresponding ideal single-panel three-dimensional point cloud model and actual single-panel three-dimensional point cloud model; the ideal single panels and the actual single panels correspond one by one; A matching point acquisition and pose alignment module, used to map the actual single-panel three-dimensional point cloud model and the corresponding ideal single-panel three-dimensional point cloud model to the same coordinate system, acquire multiple matching point pairs, and perform pose alignment; A calculation and judgment module, used to analyze the curvature deviation of each matching point pair between the actual single-panel three-dimensional point cloud model and the corresponding ideal single-panel three-dimensional point cloud model after pose alignment; if (z1 / Z)≥YZ, and / or JZ is greater than or equal to YJZ, then the corresponding actual single panel is unqualified. If any actual single panel is unqualified, then the corresponding actual hyperbolic panel is unqualified; z1 represents the number of matching point pairs with a curvature deviation greater than or equal to the preset deviation threshold, Z represents the number of matching point pairs, YZ represents the preset ratio, JZ represents the mean value of the sum of the curvature deviations of each matching point pair, and YJZ represents the preset deviation mean value.

[0039] The specific implementation of the functions of the above modules corresponds to the above hyperbolic panel finished product acceptance detection method and will not be elaborated here.

[0040] Embodiment III This invention also provides an electronic device, including: a processor, a sending device, an input device, an output device and a memory. The processor can be implemented by using a general CPU (Central Processing Unit), a microprocessor, an application specific integrated circuit, or one or more integrated circuits, etc., and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present application. The memory can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device or a random access memory (RAM), etc., and is used to store computer program codes. The computer program codes include computer instructions. When the processor executes the computer instructions, the electronic device executes the method in any one of the above possible implementation manners.

[0041] Embodiment IV This invention also provides a computer-readable storage medium, in which a computer program is stored. The computer program includes program instructions. When the program instructions are executed by the processor of an electronic device, the processor is enabled to execute the method in any one of the above possible implementation manners.

[0042] The beneficial effects of this invention are as follows: In this invention, the ideal required hyperbolic panel and the actual hyperbolic panel are respectively divided into multiple ideal single-curved plates and multiple actual single-curved plates, and the corresponding ideal single-plate three-dimensional point cloud models and actual single-plate three-dimensional point cloud models are obtained. Then, multiple matching point pairs are obtained and pose alignment is performed. After that, according to the curvature deviation of each matching point pair, the qualification situation of the actual hyperbolic panel is analyzed. The overall detection is transformed into local single-plate analysis, effectively solving the detection blind area caused by the geometric complexity of the hyperboloid, and eliminating the pose deviation between the actual and ideal models through pose alignment to ensure the consistency of the comparison benchmark and improve the detection accuracy; The area with a curvature entropy value greater than or equal to a preset threshold is used as a complex area, that is, in the high-entropy area, further encrypted scanning is performed, which can ensure that the area with a large curvature change is fully detected, and the scanning step size is adaptively adjusted according to each curvature gradient. The scanning step size is reduced in the area with a large curvature change to increase the scanning density, and the scanning step size is increased in the area with a gentle curvature change to reduce the scanning density, so as to adapt to the curvature change near the current point, improve the scanning accuracy and efficiency, and thus obtain an accurate three-dimensional point cloud model for the actual hyperbolic panel to ensure the accuracy of the finished product detection.

[0043] In the description of the specification, the descriptions referring to terms such as "one embodiment", "example", "specific example", etc. mean that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0044] In addition, in each embodiment of the present application, each functional unit can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above integrated unit can be implemented in the form of hardware or in the form of a software functional unit. If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods in each embodiment of the present application. And the aforementioned storage medium includes: various media that can store programs such as USB flash drives, mobile hard disks, read-only memories (ROM), random access memories (RAM), magnetic disks, or optical discs.

[0045] The above are only specific embodiments of the present invention, enabling those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but rather will conform to the broadest scope consistent with the principles and novel features claimed herein.

Claims

1. A method for acceptance inspection of finished hyperbolic panels, characterized in that: The following steps are involved: Acquire an ideal three-dimensional point cloud model and an actual three-dimensional point cloud model respectively; the ideal three-dimensional point cloud model and the actual three-dimensional point cloud model are three-dimensional point cloud models of an ideal required hyperbolic panel and an actual hyperbolic panel respectively; The ideal required hyperbolic panel and the actual hyperbolic panel are divided into a plurality of ideal single-curved panels and a plurality of actual single-curved panels respectively, and the corresponding ideal single-curved panel three-dimensional point cloud model and the actual single-curved panel three-dimensional point cloud model are obtained; the ideal single-curved panel and the actual single-curved panel correspond one to one; Map the actual single-board 3D point cloud model and the corresponding ideal single-board 3D point cloud model to the same coordinate system, obtain multiple matching point pairs, and perform pose alignment; Analyze the curvature deviation of each matching point pair between the actual single board 3D point cloud model after pose alignment and the corresponding ideal single board 3D point cloud model; If (z1 / Z)≥YZ, and / or JZ is greater than or equal to YJZ, the corresponding actual single-curved plate is unqualified. If any actual single-curved plate is unqualified, the corresponding actual hyperbolic panel is unqualified; z1 represents the number of matching point pairs whose curvature deviation is greater than or equal to the preset deviation threshold, Z represents the number of matching point pairs, YZ represents the preset ratio, JZ represents the mean of the sum of the curvature deviations of each matching point pair, and YJZ represents the preset deviation mean.

2. The finished product acceptance inspection method of the hyperbolic panel according to claim 1 is characterized in that: The ideal three-dimensional point cloud model is constructed and acquired according to the design parameters of the ideal hyperbolic panel required; The actual three-dimensional point cloud model is obtained specifically as follows: Generate multiple mesh regions on the actual hyperbolic panel surface; Perform 3D scanning on the actual hyperbolic panel to obtain the initial point cloud of each grid area and analyze the complex area; Adaptively adjust the scanning step size to perform point cloud encryption scanning in complex areas.

3. The finished product acceptance inspection method of the hyperbolic panel according to claim 2 is characterized in that: The complex area is analyzed as follows: Calculate the curvature entropy value of each grid area, and regard the grid area whose curvature entropy value is greater than or equal to a preset threshold as a complex area; The curvature entropy value: ; Among them, h represents the curvature entropy value, It is represented as the curvature value of the i-th initial point position in the corresponding grid area, n is represented as the number of initial points of the initial point cloud in the corresponding grid area, and i is the index symbol.

4. The finished product acceptance inspection method of the hyperbolic panel according to claim 2, characterized in that: The adaptive adjustment of the scanning step size to perform point cloud encryption scanning on complex areas is specifically as follows: ; in, It is represented as the scanning step length used when scanning from the i-th initial point position to other initial point positions in the corresponding grid area. It is represented as the curvature gradient of the i-th initial point in the corresponding grid area, ||·|| represents the modular calculation, and i is the index symbol.

5. The finished product acceptance inspection method of the hyperbolic panel according to claim 1, characterized in that: The actual single board 3D point cloud model and the corresponding ideal single board 3D point cloud model are mapped to the same coordinate system, multiple matching point pairs are obtained, and pose alignment is performed, specifically including the following steps: Extract feature points from the actual single-board 3D point cloud model and the corresponding ideal single-board 3D point cloud model, and describe the feature points; The feature points of the actual single-board 3D point cloud model are matched with the feature points of the corresponding ideal single-board 3D point cloud model through a feature matching algorithm to obtain a plurality of matching point pairs; According to the matching point pairs, the SVD algorithm is used to analyze the rotation matrix R and the translation vector t, and the actual single board 3D point cloud model is transformed according to the rotation matrix and the translation vector to complete the pose alignment with the corresponding ideal single board 3D point cloud model.

6. The finished product acceptance inspection method of the hyperbolic panel according to claim 5, characterized in that: After obtaining a plurality of matching point pairs, the method further includes using a random sampling consensus algorithm to eliminate mismatched point pairs.

7. The finished product acceptance inspection method of the hyperbolic panel according to claim 5, characterized in that: The SVD algorithm is used to analyze the rotation matrix R and the translation vector t, which is specifically: Obtain the centroid coordinate positions of the actual single board 3D point cloud model and the corresponding ideal single board 3D point cloud model, which are denoted as P1 and P2 respectively; Analyze the covariance matrix H based on P1 and P2; Performing SVD decomposition on the covariance matrix to obtain a left singular vector matrix U and a right singular vector matrix V; According to the left singular vector matrix U and the right singular vector matrix V, the rotation matrix R and the translation vector t are analyzed: ; T represents matrix transpose.

8. The method for inspection and acceptance of finished hyperbolic panels according to claim 5, characterized in that: The actual single board three-dimensional point cloud model is transformed according to the rotation matrix and translation vector, specifically: ; Indicates the coordinate position of the i-th point after the actual single-board three-dimensional point cloud model is transformed, It represents the coordinate position of the i-th point in the actual single board 3D point cloud model, where i is the index symbol.

9. The method for inspection and acceptance of finished hyperbolic panels according to claim 1, characterized in that: The curvature deviation of each matching point pair between the actual single board 3D point cloud model after pose alignment and the corresponding ideal single board 3D point cloud model is specifically: Obtain the maximum principal curvature and the minimum principal curvature of each point belonging to the matching point pair in the actual three-dimensional point cloud model after posture alignment and the corresponding ideal single-board three-dimensional point cloud model; Compute the Gaussian curvature of each point belonging to a matching pair of points: ; Calculate the curvature deviation of each matching point pair separately: ; in, Represents the Gaussian curvature of the i-th point belonging to the matching point pair in the actual 3D point cloud model after pose alignment, It represents the maximum principal curvature of the i-th point belonging to the matching point pair in the actual 3D point cloud model after pose alignment, which means, It represents the minimum principal curvature of the i-th point belonging to the matching point pair in the actual 3D point cloud model after pose alignment. represents the Gaussian curvature of the i-th point belonging to the matching point pair in the corresponding ideal single-board 3D point cloud model, represents the maximum principal curvature of the i-th point belonging to the matching point pair in the corresponding ideal single-plate 3D point cloud model, represents the minimum principal curvature of the i-th point belonging to the matching point pair in the corresponding ideal single-plate 3D point cloud model, Represents the curvature deviation of the i-th matching point pair, where i is the index symbol.

10. A finished product acceptance inspection system for a hyperbolic panel, using the finished product acceptance inspection method for a hyperbolic panel as claimed in any one of claims 1 to 9, characterized in that: include: A point cloud model acquisition module, used to respectively acquire an ideal three-dimensional point cloud model and an actual three-dimensional point cloud model; The ideal three-dimensional point cloud model and the actual three-dimensional point cloud model are three-dimensional point cloud models of the ideal required hyperbolic panel and the actual hyperbolic panel respectively; A single-curved plate division module is used to divide the ideal required hyperbolic panel and the actual hyperbolic panel into multiple ideal single-curved plates and multiple actual single-curved plates respectively, and obtain the corresponding ideal single-plate three-dimensional point cloud model and the actual single-plate three-dimensional point cloud model; the ideal single-curved plate and the actual single-curved plate correspond one to one; The matching point acquisition and pose alignment module is used to map the actual single board 3D point cloud model and the corresponding ideal single board 3D point cloud model to the same coordinate system, obtain multiple matching point pairs, and perform pose alignment; The calculation and judgment module is used to analyze the curvature deviation of each matching point pair between the actual single board 3D point cloud model after posture alignment and the corresponding ideal single board 3D point cloud model; if (z1 / Z)≥YZ, and / or JZ is greater than or equal to YJZ, the corresponding actual single curved board is unqualified, and if any actual single curved board is unqualified, the corresponding actual hyperbolic panel is unqualified; z1 represents the number of matching point pairs whose curvature deviation is greater than or equal to the preset deviation threshold, Z represents the number of matching point pairs, YZ represents the preset ratio, JZ represents the mean of the sum of the curvature deviations of each matching point pair, and YJZ represents the preset deviation mean.

Images