A method and system for evaluating the quality of a vee-die forming based on digitized sample boxes

Abstract

The application provides a kind of based on digital sample box's curved plate forming quality evaluation method and system, comprising: obtaining the point cloud data of the measured forming curved plate;Theoretical curved plate digital model is constructed;The point cloud data of the measured curved plate is registered with the theoretical curved plate digital model;The point cloud data of the measured curved plate is compared with the theoretical curved plate digital model;Based on the two kinds of numerical model after registration and comparison, the accuracy of the measured curved plate is evaluated.Based on the two kinds of numerical model after registration and comparison, the forming error is fed back.Using the present application, the original sample sample box processing can be replaced, and energy consumption is saved;It omits the processing of sample sample box and the link of manual recording, saves a lot of manual working hours, and improves the detection efficiency.

Description

Technical Field

[0001] This invention relates to the field of optical measurement, and more specifically, to a method and system for evaluating the forming quality of curved plates based on a digital sample box. Background Technology

[0002] Complex free-form sheet materials have intricate shapes and are primarily formed using water-fire bending and mechanical cold bending methods, making precise forming difficult. The forming process involves multiple passes, requiring in-situ measurements of the surface shape between each pass. These measurements are then compared with the design deviations to allow for feedback control of subsequent process parameters. Currently, the forming process for complex free-form sheets still largely relies on traditional manual inspection methods such as templates and sample boxes. This results in long manufacturing preparation cycles, low inspection accuracy, high labor intensity for on-site workers, and difficulty in quantifying and tracing the inspection results. Therefore, there is an urgent need for a new digital inspection system for complex free-form sheet forming, along with data monitoring and on-site feedback technologies, to achieve efficient and high-precision in-situ measurement, deviation assessment, and on-site operation guidance for the complex free-form sheet surface shape during the forming process. Summary of the Invention

[0003] To address the shortcomings of existing technologies, the purpose of this invention is to provide a method and system for evaluating the forming quality of curved plates based on digital sample boxes.

[0004] According to one aspect of the present invention, a method for evaluating the forming quality of curved plates based on digital templates is provided, comprising:

[0005] Obtain the point cloud data of the shaped curved plate to be tested;

[0006] Construct a digital model of the theoretical curved plate;

[0007] The point cloud data of the test plate is registered with the digital model of the theoretical plate.

[0008] The point cloud data of the test plate is compared with the digital model of the theoretical plate.

[0009] The accuracy of the test plate is evaluated based on the two digital models obtained after registration and comparison.

[0010] Based on the two digital models after registration and comparison, feedback is provided on the molding error.

[0011] Preferably, acquiring the point cloud data of the shaped curved plate to be tested includes:

[0012] A binocular stereo vision system is used to project a circular spot array onto the surface of the curved plate under test.

[0013] Images of the projected circular spot array are acquired using left and right cameras, and the center point of the circular spots is extracted.

[0014] Based on neighborhood topology information, a single-frame structured light stereo matching technology is used to establish stereo matching of corresponding circular spots in the left and right cameras, and obtain the three-dimensional reconstruction of the circular spot array, that is, to obtain the point cloud data of the curved board under test from the current shooting perspective.

[0015] Acquire point cloud data of the test plate from multiple angles;

[0016] The coordinate system of the point cloud data is normalized.

[0017] Based on the normalized coordinate system, the point cloud data of the test plate from multiple angles are merged into a complete point cloud data, thus obtaining the point cloud data of the entire test plate.

[0018] Preferably, the step of merging the point cloud data of the test plate from multiple angles into a complete point cloud data based on the normalized coordinate system includes:

[0019] Initial matching of multi-view point cloud data: The corresponding points of the multi-view point cloud data are determined by using the corresponding point determination method based on the extended Gaussian sphere, and initial matching is performed.

[0020] Global stitching of multi-view point cloud data: Based on the initial matching, the ICP method is used for global stitching, that is, to find the target point cloud p of the theoretical curved plate digital model. i Compared with the actual obtained measurement point cloud p i Let the rotation matrix R and translation vector T between the data be such that F(R,T)=min∑[Rp′ i +Tp i ] 2 To achieve the optimal;

[0021] Line laser data stitching: Combining the principles of stereo vision measurement and laser triangulation, a scanning 3D measurement system is constructed using two CCD cameras and a laser emitter. The system uses grid markers to complete the stitching, matching, and fusion of point cloud data and line laser scanning data.

[0022] Preferably, the registration includes coarse registration and fine registration; the coarse registration includes centroid matching, average normal vector matching, and principal axis matching.

[0023] The centroid matching is as follows: the three-dimensional moments in the Cartesian coordinate system can be expressed by matrix theory, thereby calculating the centroids of the point cloud of the test plate and the theoretical point cloud respectively, and realizing the centroid matching of the digital model of the plate.

[0024] The average normal vector matching is achieved by calculating the point cloud data of the test plate and the digital model of the theoretical plate, respectively.

[0025] The average normal vector is then registered;

[0026] The inertial principal axis matching is as follows: the point cloud data of the curved plate to be tested is projected and registered with the inertial principal axis of the theoretical curved plate digital model, respectively.

[0027] The fine registration is achieved by using the iterative nearest point (ICP) method to achieve fine registration of the curved plate digital model.

[0028] Preferably, comparing the point cloud data of the test plate with the theoretical digital model of the test plate includes:

[0029] The registered measured point cloud and theoretical point cloud are projected onto the XOY plane respectively;

[0030] Find the four theoretical projection points on the XOY plane that are closest to the measured projection point;

[0031] Using the three-dimensional theoretical points corresponding to the four theoretical projection points, interpolate three-dimensional theoretical points that have the same projection as the corresponding measurement points; use the three-dimensional theoretical points as the corresponding points of the measurement points in the theoretical point cloud.

[0032] Preferably, the evaluation of the accuracy of the test plate based on the two digital models after registration and comparison includes:

[0033] Formed surface difference: The difference between the formed surface and the theoretical surface is evaluated by directly comparing the measured point cloud with the theoretical point cloud;

[0034] Lateral Formability: Obtain the point cloud data of the curved plate to be tested and the rib data of the theoretical curved plate digital model; keep the theoretical rib stationary, translate the measured rib along the axial direction (depth direction) within the same rib plane, so that the lowest point of the measured rib coincides with the lowest point of the theoretical rib; calculate the distance between corresponding points of the theoretical rib and the measured rib in the depth direction, thereby representing the lateral formability of the curved plate at that rib; use the maximum value and average value of the distances between all corresponding points on the theoretical rib and the measured rib to represent the lateral formability deviation of the rib line;

[0035] Longitudinal forming distortion: The longitudinal forming distortion is evaluated using the angle between the average normal vectors of the theoretical rib and the measured rib.

[0036] Longitudinal forming accuracy: After the digital model of the formed curved plate is registered, the centerline data on the actual measured point cloud of the formed curved plate is obtained by using the projection interpolation method; the axial depth value of the centerline is compared with the theoretical data and the measured data to represent the longitudinal forming accuracy of the formed curved plate.

[0037] Preferably, the step of feeding back the molding error based on the two digital models after registration and comparison includes:

[0038] First, the coded coordinates of each projected circular spot in the acquired image and the projection base plate are kept in a one-to-one correspondence, thereby establishing a one-to-one mapping of the circular spots in the projector and camera.

[0039] Secondly, according to the principle of binocular stereo vision 3D reconstruction, the circular spots in the images captured by the left and right cameras and the spatial points in the 3D reconstruction actually have a one-to-one mapping relationship.

[0040] Finally, by using the imaging circular spots in the images captured by the camera as a bridge, a one-to-one mapping relationship can be established between the two-dimensional projection circular spots and the three-dimensional reconstructed circular spots in the projector.

[0041] Based on the mapping relationship, once the digital model has completed registration and comparison, the deviation between each three-dimensional measurement point and the theoretical point can be calculated and used as the forming error.

[0042] The forming error of each circular spot center point on the projection base plate corresponds to a three-dimensional measurement point. This forming error is converted into color and rendered on the projection base plate pattern, or the corresponding error is directly marked on the projection circular spot in numerical form.

[0043] A projector is used to project the base plate pattern with added error information onto the surface of the shaped curved plate being tested.

[0044] According to a second aspect of the present invention, a curved plate forming quality evaluation system based on a digital template is provided, comprising:

[0045] The module for acquiring point cloud data of the shaped curved plate under test acquires the point cloud data of the shaped curved plate under test.

[0046] Theoretical Curve Module: This module constructs a digital model of the theoretical curve.

[0047] The registration module registers the point cloud data of the test plate with the digital model of the theoretical plate.

[0048] The comparison module compares the point cloud data of the test plate with the theoretical digital model of the plate.

[0049] The evaluation module evaluates the accuracy of the test plate based on the two digital models after registration and comparison.

[0050] Based on the two digital models after registration and comparison, feedback is provided on the molding error.

[0051] According to a third aspect of the present invention, a terminal is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, can be used to perform the methods described above, or to run the systems described above.

[0052] According to a fourth aspect of the present invention, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, can be used to perform the methods described above, or to run the system described above.

[0053] Compared with the prior art, the present invention has the following beneficial effects:

[0054] The method and system for evaluating the forming quality of curved plates based on digital template boxes in this invention embodiment can replace the original template box processing, saving energy consumption; it omits the template box processing and manual recording steps, saving a lot of manpower and improving testing efficiency. Attached Figure Description

[0055] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0056] Figure 1 This is a flowchart of a curved plate forming quality evaluation method based on a digital sample box according to an embodiment of the present invention;

[0057] Figure 2 This is a schematic diagram of the structure of a binocular stereo vision measurement system according to a preferred embodiment of the present invention;

[0058] Figure 3 This is a framework diagram of acquiring three-dimensional model data in a preferred embodiment of the present invention. Detailed Implementation

[0059] The present invention will now be described in detail with reference to specific embodiments. These embodiments will help those skilled in the art to further understand the present invention, but do not limit the invention in any way. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention. These all fall within the scope of protection of the present invention.

[0060] based on Figure 1 The present invention provides an embodiment of a method for evaluating the forming quality of curved plates based on digital sample boxes, which includes the following process:

[0061] S100, acquire point cloud data of the shaped curved plate to be tested;

[0062] S200, a model for building digital sample boxes;

[0063] S300, register the point cloud data of the shaped curved plate to be tested in S100 with the model of the digital sample box in S200.

[0064] S400 compares the point cloud data of the shaped curved plate to be tested in S100 with the model of the digital sample box in S200.

[0065] S500 evaluates the accuracy of the test plate based on the two digital models after registration and comparison;

[0066] The S600 provides feedback on molding errors based on two digital models after registration and comparison.

[0067] The method in this embodiment can replace the original sample box processing, saving energy consumption; it omits the sample box processing and manual recording steps, saving a lot of manpower and improving testing efficiency.

[0068] In a preferred embodiment of the present invention, implementing S100 includes two steps:

[0069] S101, acquire point cloud data of the shaped curved plate to be tested from multiple angles;

[0070] S102 stitches together the point cloud data of the shaped curved plate to be tested from multiple angles.

[0071] Before implementing 101, it is necessary to establish a binocular stereo vision measurement system. The specific process for establishing the system is as follows:

[0072] S01: Optimization of structural parameters for in-situ measurement of binocular stereo vision: Designing a reasonable baseline distance between binocular measurement sensors;

[0073] S02: Development of Binocular Stereo Vision In-Situ Measurement Sensor: Selection of binocular stereo vision sensors, including: camera, lens, projector, and system setup; see [link / reference] Figure 2

[0074] S03: In-situ calibration of the binocular stereo vision measurement sensor: The calibration parameters of binocular stereo vision mainly consist of two parts: the intrinsic parameters of the two cameras and the pose parameters between the two cameras (i.e., the structural parameters of binocular stereo vision). A small-sized two-dimensional target is used to calibrate the structural parameters of binocular stereo vision;

[0075] S04: Multi-Vision Sensor Integration and Spatial Optimization Layout: This paper proposes a multi-vision sensor spatial layout planning method that optimizes measurement uncertainty. The feasibility of the multi-vision sensor spatial layout planning strategy is ensured by the three-dimensional measurement uncertainty of the target point. First, a discretized geometric model of multi-vision sensor measurement is established based on the measured object, and the decision variables of the spatial layout planning problem are defined. Then, the measurement uncertainty is set as the objective function of spatial layout planning. Combined with the constraints in multi-vision sensor measurement, a genetic algorithm is applied to optimize the decision variables of multi-vision sensor spatial layout planning, and finally, the optimal multi-vision sensor spatial layout planning scheme is obtained.

[0076] In S01, different baseline distances of the binocular measurement sensor affect its measurement accuracy and field of view. The baseline distance directly impacts the measurement error; as the baseline distance increases, the measurement error first decreases and then increases. Therefore, finding the optimal baseline distance is crucial. First, a large baseline distance increases the sensor's structural volume, reducing its operational flexibility. Second, a large baseline distance increases the sensor's sensitivity to vibration. Finally, a large baseline distance also increases the parallax between the left and right cameras on the same object, making occlusion more likely and increasing the difficulty of stereo matching. Therefore, to obtain a reasonable design result, a balance must be struck between the sensor's baseline distance and measurement accuracy to achieve the best measurement effect.

[0077] Based on the above-mentioned binocular vision stereo system, implement S101 and S102:

[0078] In a preferred embodiment, step S101 is performed to acquire point cloud data of the shaped curved plate to be tested from multiple angles. The process is as follows:

[0079] S1011, firstly, by projecting a circular spot array onto the surface of the shaped curved plate being measured;

[0080] S1012, then the left and right cameras acquire images of the projected circular spot array and extract the center point of the circular spot;

[0081] S1013 Finally, based on the single-frame structured light stereo matching technology based on neighborhood topology information, stereo matching of corresponding circular spots in the left and right cameras is established, thereby realizing the three-dimensional reconstruction of the circular spot array.

[0082] In a preferred embodiment, step S102 is performed to stitch together the point cloud data of the shaped curved plate to be tested from multiple angles. The process is as follows:

[0083] S1021, Initial matching of multi-view point cloud data (point cloud data obtained after processing data collected from different angles): A corresponding point determination method based on an extended Gaussian sphere is adopted to determine the corresponding points of each multi-view point cloud data to achieve initial matching, such as normal vector, tangent vector, and curvature.

[0084] S1022, Global stitching of multi-view point cloud data: Based on initial matching, the ICP algorithm is used. Specifically, the basic idea of ​​the ICP method is to find the guiding target point cloud p. i The rotation matrix R and translation vector T between the actual obtained measurement point cloud data and the following equation are optimized:

[0085] F(R,T)=min∑[Rp′ i +Tp i ] 2

[0086] S1023: Combining the principles of stereo vision measurement and laser triangulation, a scanning 3D measurement system was constructed using two CCD cameras and a laser emitter. Data stitching and matching were achieved using grid markers, and a richer 3D object model with more point cloud information was obtained through data fusion. Specifically, this embodiment uses a sub-pixel laser center extraction algorithm based on the HSV color space, which further improves the extraction accuracy of the laser center compared to the original grayscale centroid method. A grid marker board was designed, using the intersections of grid lines as markers for inter-frame matching. The laser data was stitched together based on the spatial position invariance of the markers and the principles of stereo vision. Then, the spatial 3D model of the object under test was obtained using both stereo vision measurement and laser triangulation principles. Finally, by establishing a global coordinate system and solving the pose transformation relationship between the spatial coordinate systems using markers, the point cloud data obtained from the two methods were fused to obtain a richer 3D point cloud model of the object surface. This improves measurement accuracy and is less affected by the surface texture characteristics of the object. See [link to documentation]. Figure 3 As shown.

[0087] Through the above embodiments, in S101, point cloud data from multiple perspectives are acquired; if the shaped curved plate to be tested is to be assembled in S102, the point cloud data of the entire shaped curved plate to be tested can be obtained. The point cloud data of the shaped curved plate to be tested obtained through the above embodiments has high accuracy, laying a solid data foundation for subsequent calibration, comparison, and evaluation steps.

[0088] In a preferred embodiment of the present invention, step S200 is implemented to construct a model of the digital prototype box template. Specifically, a reverse-engineering method for constructing the digital prototype box template based on a three-dimensional surface design model is used, and a theoretical surface design model is derived using ship-aided design software such as Tribon or AM. This theoretical surface design model replaces the sample prototype box in the prior art, saving material costs and manual processing time.

[0089] In a preferred embodiment of the present invention, step S300 is performed to register the point cloud data of the shaped curved plate to be tested with the model of the digital template box.

[0090] This process primarily utilizes globally invariant features of the digital models, such as the centroid (first-order moment), average normal vector, and principal axes of inertia (second-order moments), to establish coarse registration between the models. This effectively solves the registration difficulties caused by low shape similarity and large pose differences between the measured and theoretical point clouds in the early stages of ship curved plate forming. The registration results provided in this embodiment can be used as the initial solution for optimization methods (such as ICP) to reduce the risk of the iterative process getting trapped in local optima. Therefore, the overall registration process mainly consists of the following four steps: centroid matching, average normal vector matching, principal axis of inertia matching, and fine registration.

[0091] Specifically, S301, centroid matching: the three-dimensional moments in the Cartesian coordinate system can be expressed through matrix theory, thereby calculating the centroids of the measured point cloud data and the theoretical point cloud of the curved plate respectively, thus realizing the centroid matching of the digital model of the curved plate.

[0092] S302, Average Normal Vector Matching: The average normal vector of the digital model is used as a global matching feature to further register the measured point cloud and the theoretical point cloud of the curved plate. For the 3D measured point cloud of the curved plate, the average normal vector of the curved plate should be the average of the sum of the normal vectors of each point in the point cloud.

[0093] S303, Inertial Principal Axis Matching: Registration of the measured point cloud and the theoretical point cloud of a curved plate can be achieved using the inertial principal axis of the projected point cloud. The specific process is as follows:

[0094] S3031, First, calculate the two-dimensional projected point cloud m″. i and d″ i The angle between the maximum principal axis of inertia and the X-axis and

[0095] S3032, and then use the included angles respectively and For the measured point cloud M″ i And theoretical point cloud D″ i Rotating around the Z-axis achieves inertial principal axis matching between the two projected point clouds, and also realizes approximate registration between the measured point cloud and the theoretical point cloud of the curved plate.

[0096] S3034, Fine Registration: The Iterative Closest Point (ICP) algorithm is used to achieve fine registration of the curved plate digital model. ICP registers two point cloud datasets iteratively. In each iteration, the algorithm selects the closest point as the corresponding matching point and calculates a set of rigid body transformation parameters (rotation matrix and translation vector) to minimize the following formula:

[0097]

[0098] N in the formula p and N q These represent the total number of points in the point cloud P of the test surface and the point cloud Q of the theoretical surface, respectively; w i,j This represents the weight coefficient of the corresponding matching point pair, and its values ​​are typically as follows: If p i It is q j The closest point, then w i,j =1; otherwise w i,j =0. Therefore, the formula can be further simplified to:

[0099]

[0100] in the formula Once the corresponding matching points are determined, the rigid body transformation between the two point clouds can be directly solved; after obtaining the rigid body transformation, the matching of the two point clouds is completed, and point cloud fusion can be performed.

[0101] Therefore, registration refers to the process of stitching or aligning two geometric models of different objects or the same object in three-dimensional space using rigid body transformation.

[0102] Model registration is a prerequisite for model comparison; only with good registration can the comparison be valuable. Furthermore, the quality of registration directly affects the comparison results.

[0103] In this embodiment, the data registration of the point cloud of the curved plate under test and the theoretical point cloud was completed through coarse registration and fine registration. The registration was carried out from multiple aspects to ensure the accuracy of the registration, which provides a solid foundation for the subsequent evaluation of the forming quality of the curved plate.

[0104] In one embodiment of the present invention, S400 is implemented to compare the point cloud data of the shaped curved plate to be tested with the model of the digital sample box.

[0105] After the measured point cloud and theoretical point cloud of the shaped curved plate are registered, it is difficult to guarantee absolute consistency between the two point clouds in terms of spatial sampling interval and sampling position. This means that when searching for corresponding points in the theoretical point cloud along a direction parallel to the Z-axis, there is often no corresponding point in the theoretical point cloud whose line connecting the corresponding measured point is exactly parallel to the Z-axis. To accurately calculate the true depth difference of the point cloud model in the Z-axis direction, a method based on XOY plane projection interpolation is used to calculate the corresponding points of the measured points in the theoretical point cloud. The specific process is as follows: S401, project the registered measured point cloud and theoretical point cloud onto the XOY plane respectively;

[0106] S402, find the four theoretical projection points on the XOY plane that are closest to the measured projection point;

[0107] S403, using the three-dimensional theoretical points corresponding to the four theoretical projection points mentioned above, interpolate to obtain a three-dimensional theoretical point with the same projection as the measured point. This point is the corresponding point of the measured point in the theoretical point cloud. Once the measured point and its corresponding theoretical point are determined, the Z coordinate values ​​of the two points can be directly compared to calculate the forming deviation between the two points.

[0108] In this embodiment, the digital model comparison is the quantitative deviation between the measured point cloud and the theoretical point cloud, providing guidance for on-site operations.

[0109] In a preferred embodiment of the present invention, S500 is implemented to evaluate the accuracy of the shaped curved plate to be tested based on the two digital models after registration and comparison. The specific process is as follows:

[0110] S501, Surface Asymmetry of the Formed Surface: Based on the digital model registration described above, the surface symmetry between the formed surface and the theoretical surface is evaluated by directly comparing the measured point cloud with the theoretical point cloud. A measurement method based on a projection circular spot array is used; compared with template measurement, the measurement density in this embodiment is higher.

[0111] S502, based on the previous model registration, regarding lateral forming accuracy: For binocular stereo vision measurement methods based on circular spot array structured light, it is difficult to ensure that the center of the projected circular spot is exactly located on the rib line marked on the surface of the formed curved plate during actual measurement. Therefore, it is impossible to directly obtain the measurement data of the actual rib line on the curved plate, and thus impossible to directly compare the measured rib values ​​and theoretical rib values. In the subsequent registration process between the measured point cloud and the theoretical point cloud of the curved plate, the theoretical rib data is kept synchronized. After the measured point cloud and the theoretical point cloud of the curved plate are registered, the corresponding rib data in the measured point cloud can be interpolated using the above-mentioned corresponding point lookup method based on projection interpolation. After obtaining the theoretical rib data and the actual measured rib data, the models on different ribs can be directly compared to evaluate the lateral forming accuracy of the curved plate on different ribs. Specifically,

[0112] S5021, first keep the theoretical rib position still, and translate the measured rib position along the axial direction (depth direction) within the same rib position plane so that the lowest point of the measured rib position coincides with the lowest point of the theoretical rib position.

[0113] S5022, then calculate the distance between the theoretical rib and the corresponding point in the depth direction of the measured rib, so as to represent the transverse forming degree of the curved plate at the rib.

[0114] S5023 Finally, for quantitative description, the maximum and average distances of all corresponding points on the theoretical rib and the measured rib are used to represent the lateral forming deviation of the rib line.

[0115] S503, Longitudinal forming distortion: evaluated using the angle between the average normal vectors of the theoretical rib and the measured rib.

[0116] S504, Longitudinal forming accuracy: The same detection method as for transverse forming accuracy is adopted, that is, the theoretical centerline of the formed curved plate is mapped into the corresponding measurement centerline in the measurement point cloud and then compared. To achieve the digital model comparison of the centerline, the following is included:

[0117] S5041, First, after the digital model of the formed curved plate is registered, the centerline data on the actual measured point cloud of the formed curved plate is obtained by using the projection interpolation method mentioned above.

[0118] S5042, then compare the theoretical data and the measured axial depth value of the centerline to detect the longitudinal forming accuracy of the curved plate.

[0119] In the above embodiments, the surface difference of the formed surface is used to evaluate the smoothness of the formed surface of the curved surface. The transverse formability reflects the forming error of the rib line of the curved plate, the longitudinal formability reflects the forming error of the centerline of the curved plate, and the longitudinal forming distortion reflects the forming error of the centerline of the curved plate. These indicators together reflect the forming accuracy of the curved surface, and the subsequent forming process parameters can be set or corrected according to the deviation of these forming parameters.

[0120] In a preferred embodiment of the present invention, S600 is implemented, whereby the forming error is fed back based on the two digital models after registration and comparison. Specifically, this involves combining single-frame circular array structured light stereo matching technology based on neighborhood topology information to establish a one-to-one mapping relationship between the forming error of the ship's curved plate and the projection features, thereby directly projecting the digital model comparison error of the formed curved plate onto the surface of the measured formed curved plate. This specifically includes:

[0121] S601, mapping between two-dimensional projection features and three-dimensional reconstruction features; specifically, including:

[0122] S601, First, the coded coordinates of each projected circular spot in the acquired image and the projection base plate are kept in a one-to-one correspondence, thereby establishing a one-to-one mapping of the circular spots in the projector and the camera.

[0123] S6012, Secondly, according to the principle of binocular stereo vision 3D reconstruction, the circular spots in the images captured by the left and right cameras and the spatial points of 3D reconstruction actually have a one-to-one mapping relationship.

[0124] S6013 Finally, based on the imaging circular spots in the images captured by the camera as a bridge, a one-to-one mapping relationship can be established between the two-dimensional projection circular spots and the three-dimensional reconstructed circular spots in the projector.

[0125] Once the digital model has been registered and compared, the deviation between each three-dimensional measurement point and the theoretical point can be calculated.

[0126] S602, each circular spot on the projection base plate corresponds to a forming error of a three-dimensional measurement point. Therefore, this forming error can be converted into color and rendered on the projection base plate pattern, or the corresponding error can be directly marked on the projection circular spot in numerical form. Finally, a projector is used to project the base plate pattern with added error information onto the surface of the shaped curved plate being measured.

[0127] This embodiment directly feeds back the forming error of the ship's curved plate to the surface of the formed curved plate, which is beneficial for on-site technicians to formulate or correct subsequent forming process parameters.

[0128] Based on the same inventive concept, this invention provides a curved plate forming quality evaluation system based on a digital template, including a point cloud acquisition module for the curved plate under test, a theoretical curved plate module, a registration module, a comparison module, and an evaluation module. The curved plate point cloud acquisition module acquires point cloud data of the curved plate under test; the theoretical curved plate module constructs a theoretical curved plate digital model; the registration module registers the point cloud data of the curved plate under test with the theoretical curved plate digital model; the comparison module compares the point cloud data of the curved plate under test with the theoretical curved plate digital model; the evaluation module evaluates the accuracy of the curved plate under test based on the two digital models obtained from the registration and comparison; and provides feedback on forming errors based on the two digital models obtained from the registration and comparison.

[0129] The specific implementation techniques of each module / unit in the above examples of the present invention can be referred to the steps of the curved plate forming quality evaluation method based on digital sample box in the above embodiments, and will not be repeated here.

[0130] Based on the same inventive concept, the present invention provides a terminal, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it can be used to perform the above-described method or to run the above-described system.

[0131] Based on the same inventive concept, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, can be used to perform the above-described method or to run the above-described system.

[0132] Specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art can make various modifications or variations within the scope of the claims, which do not affect the essence of the present invention. The above preferred features can be used in any combination without conflict.

Claims

1. A method for evaluating the forming quality of curved plates based on digital templates, characterized in that, include: Obtain the point cloud data of the curved surface to be tested; Construct a digital model of the theoretical curved plate; The point cloud data of the test plate is registered with the digital model of the theoretical plate. The point cloud data of the test plate is compared with the digital model of the theoretical plate. Based on the two digital models after registration and comparison, the accuracy of the curved plate under test is evaluated; and / or, based on the two digital models after registration and comparison, the forming error is fed back. The accuracy of the test plate is evaluated based on the two digital models after registration and comparison, including: Formed surface difference: The difference between the formed surface and the theoretical surface is evaluated by directly comparing the measured point cloud with the theoretical point cloud; Lateral Formability: Obtain the point cloud data of the curved plate to be tested and the rib data of the theoretical curved plate digital model; keep the theoretical rib stationary, and translate the measured rib along the axial direction within the same rib plane so that the lowest point of the measured rib coincides with the lowest point of the theoretical rib; calculate the distance between corresponding points in the depth direction of the theoretical rib and the measured rib respectively, thereby representing the lateral formability of the curved plate at that rib; use the maximum value and average value of the distances between all corresponding points on the theoretical rib and the measured rib respectively to represent the lateral formability deviation of the rib line; Longitudinal forming distortion: The longitudinal forming distortion is evaluated using the angle between the average normal vectors of the theoretical rib and the measured rib. Longitudinal forming accuracy: After the digital model of the formed curved plate is registered, the centerline data on the actual measured point cloud of the formed curved plate is obtained by using the projection interpolation method; the axial depth value of the centerline is compared with the theoretical data and the measured data to represent the longitudinal forming accuracy of the formed curved plate. The feedback of molding error based on the two digital models after registration and comparison includes: First, the coded coordinates of each projected circular spot in the acquired image and the projection base plate are kept in a one-to-one correspondence, thereby establishing a one-to-one mapping of the circular spots in the projector and camera. Secondly, according to the principle of binocular stereo vision 3D reconstruction, the circular spots in the images captured by the left and right cameras and the spatial points in the 3D reconstruction actually have a one-to-one mapping relationship. Finally, by using the imaging circular spots in the images captured by the camera as a bridge, a one-to-one mapping relationship can be established between the two-dimensional projection circular spots and the three-dimensional reconstructed circular spots in the projector. Based on the mapping relationship, once the digital model has completed registration and comparison, the deviation between each three-dimensional measurement point and the theoretical point can be calculated and used as the forming error. The forming error of each circular spot center point on the projection base plate corresponds to a three-dimensional measurement point. This forming error is converted into color and rendered on the projection base plate pattern, or the corresponding error is directly marked on the projection circular spot in numerical form. A projector is used to project the base plate pattern with added error information onto the surface of the shaped curved plate being tested.

2. The method for evaluating the forming quality of curved plates based on a digital template according to claim 1, characterized in that, The acquisition of point cloud data of the shaped curved plate to be tested includes: A binocular stereo vision system is used to project a circular spot array onto the surface of the curved plate under test. Images of the projected circular spot array are acquired using left and right cameras, and the center point of the circular spots is extracted. Based on neighborhood topology information, a single-frame structured light stereo matching technology is used to establish stereo matching of corresponding circular spots in the left and right cameras, and obtain the three-dimensional reconstruction of the circular spot array, that is, to obtain the point cloud data of the curved board under test from the current shooting perspective. Acquire point cloud data of the test plate from multiple angles; The coordinate system of the point cloud data is normalized. Based on the normalized coordinate system, the point cloud data of the test plate from multiple angles are merged into a complete point cloud data, thus obtaining the point cloud data of the entire test plate.

3. The method for evaluating the forming quality of curved plates based on a digital sample box according to claim 2, characterized in that, The normalized coordinate system merges the point cloud data of the test plate from multiple angles into a complete point cloud data, including: Initial matching of multi-view point cloud data: The corresponding points of the multi-view point cloud data are determined by using the corresponding point determination method based on the extended Gaussian sphere, and initial matching is performed. Global stitching of multi-view point cloud data: Based on the initial matching, the ICP method is used for global stitching, that is, to find the target point cloud of the theoretical curved plate digital model. Compared with the actual obtained measurement point cloud Rotation matrix between data Translation vector ,make To achieve the optimal; Line laser data stitching: Combining the principles of stereo vision measurement and laser triangulation, a scanning 3D measurement system is constructed using two CCD cameras and a laser emitter. The system uses grid markers to complete the stitching, matching, and fusion of point cloud data and line laser scanning data.

4. The method for evaluating the forming quality of curved plates based on a digital sample box according to claim 1, characterized in that, The registration includes coarse registration and fine registration; the coarse registration includes centroid matching, average normal vector matching, and principal axis of inertia matching. The centroid matching is achieved by using matrix theory to express the three-dimensional moments in the Cartesian coordinate system, thereby calculating the centroids of the point cloud of the test plate and the theoretical point cloud, and realizing the centroid matching of the digital model of the plate. The average normal vector matching is as follows: the point cloud data of the test plate and the average normal vector of the theoretical plate digital model are calculated and registered respectively. The inertial principal axis matching is as follows: the point cloud data of the curved plate to be tested is projected and registered with the inertial principal axis of the theoretical curved plate digital model, respectively. The fine registration is achieved by using the iterative nearest point (ICP) method to achieve fine registration of the curved plate digital model.

5. The method for evaluating the forming quality of curved plates based on a digital template according to claim 1, characterized in that, The step of comparing the point cloud data of the test curved plate with the theoretical curved plate digital model includes: The registered measured point cloud and theoretical point cloud are projected onto the XOY plane respectively; Find the four theoretical projection points on the XOY plane that are closest to the measured projection point; Using the three-dimensional theoretical points corresponding to the four theoretical projection points, interpolate three-dimensional theoretical points that have the same projection as the corresponding measurement points; use the three-dimensional theoretical points as the corresponding points of the measurement points in the theoretical point cloud.

6. A curved plate forming quality evaluation system based on a digital template box, characterized in that, include: The module for acquiring point cloud data of the curved board under test acquires the point cloud data of the curved board under test. The theoretical curve board module constructs a digital model of the theoretical curve board; The registration module registers the point cloud data of the test plate with the digital model of the theoretical plate. The comparison module compares the point cloud data of the test plate with the theoretical digital model of the plate. The evaluation module evaluates the accuracy of the test plate based on the two digital models after registration and comparison. Based on the two digital models after registration and comparison, feedback is provided on the forming error; The accuracy of the test plate is evaluated based on the two digital models after registration and comparison, including: Formed surface difference: The difference between the formed surface and the theoretical surface is evaluated by directly comparing the measured point cloud with the theoretical point cloud; Lateral Formability: Obtain the point cloud data of the curved plate to be tested and the rib data of the theoretical curved plate digital model; keep the theoretical rib stationary, and translate the measured rib along the axial direction within the same rib plane so that the lowest point of the measured rib coincides with the lowest point of the theoretical rib; calculate the distance between corresponding points in the depth direction of the theoretical rib and the measured rib respectively, thereby representing the lateral formability of the curved plate at that rib; use the maximum value and average value of the distances between all corresponding points on the theoretical rib and the measured rib respectively to represent the lateral formability deviation of the rib line; Longitudinal forming distortion: The longitudinal forming distortion is evaluated using the angle between the average normal vectors of the theoretical rib and the measured rib. Longitudinal forming accuracy: After the digital model of the formed curved plate is registered, the centerline data on the actual measured point cloud of the formed curved plate is obtained by using the projection interpolation method; the axial depth value of the centerline is compared with the theoretical data and the measured data to represent the longitudinal forming accuracy of the formed curved plate. The feedback of molding error based on the two digital models after registration and comparison includes: First, the coded coordinates of each projected circular spot in the acquired image and the projection base plate are kept in a one-to-one correspondence, thereby establishing a one-to-one mapping of the circular spots in the projector and camera. Secondly, according to the principle of binocular stereo vision 3D reconstruction, the circular spots in the images captured by the left and right cameras and the spatial points in the 3D reconstruction actually have a one-to-one mapping relationship. Finally, by using the imaging circular spots in the images captured by the camera as a bridge, a one-to-one mapping relationship can be established between the two-dimensional projection circular spots and the three-dimensional reconstructed circular spots in the projector. Based on the mapping relationship, once the digital model has completed registration and comparison, the deviation between each three-dimensional measurement point and the theoretical point can be calculated and used as the forming error. The forming error of each circular spot center point on the projection base plate corresponds to a three-dimensional measurement point. This forming error is converted into color and rendered on the projection base plate pattern, or the corresponding error is directly marked on the projection circular spot in numerical form. A projector is used to project the base plate pattern with added error information onto the surface of the shaped curved plate being tested.

7. A terminal, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it can be used to perform the method of any one of claims 1-5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, this program can be used to perform the method described in any one of claims 1-5.

Images