A label identification method, device and equipment of an ionization separation panel

Abstract

This disclosure provides a method, apparatus, and device for tag identification of an ionization separation panel. The method includes: acquiring a test image associated with the ionization separation panel under test, and determining a first test point cloud associated with the test image; processing the first test point cloud and a first standard point cloud using the same rules to obtain a first feature matrix and a second feature matrix; obtaining a target registration matrix corresponding to the second test point cloud based on the first and second feature matrices; performing rotation correction on the test image according to the target registration matrix to obtain a corrected test image; and performing information recognition on the test image based on a visual recognition model to output the tag content sequence and tag number in the test image. The technical solution of this disclosure, based on the first standard point cloud and the first test point cloud of the ionization separation panel under test, determines the tag content sequence and tag number, thereby improving the accuracy and stability of tag content recognition.

Description

Technical Field

[0001] This disclosure relates to the field of industrial automation testing technology, and in particular to a label identification method, apparatus and equipment for an ionization separation panel. Background Technology

[0002] For ionization separation panels, label inspection is a core process for ensuring product quality and safety. Labels carry the identification information and traceability data of the parts, and are a necessary means to achieve full lifecycle management of parts and ensure consistent assembly.

[0003] Currently, the main method for detecting labels on ionization separation panels is a combination of manual visual inspection and QR code scanning verification. This method suffers from high reliance on manual labor, low detection efficiency, and low detection accuracy. Summary of the Invention

[0004] This disclosure provides a label recognition method, apparatus, and device for ionization separation panels to improve the accuracy and stability of label content recognition.

[0005] In a first aspect, embodiments of this disclosure provide a tag identification method for an ionization separation panel, the method comprising: Acquire a test image associated with the ionization separation panel under test, and determine a first test point cloud associated with the test image; wherein, the ionization separation panel under test is affixed with a label; The same rules are applied to the first test point cloud and the first standard point cloud to obtain a first feature matrix corresponding to the first test point cloud and a second feature matrix corresponding to the first standard point cloud; wherein, the first standard point cloud is determined based on the CAD model of the ionization separation panel to be tested; Based on the first feature matrix and the second feature matrix, the second point cloud to be tested and the second standard point cloud are processed to obtain a target registration matrix corresponding to the second point cloud to be tested. The second point cloud to be tested is a point cloud after downsampling the first point cloud to be tested, and the second standard point cloud is a point cloud after downsampling the first standard point cloud. The downsampling rules are the same. Based on the target registration matrix, the image to be tested is rotated and corrected to obtain the corrected image to be used; The image to be tested is identified based on a trained visual recognition model, and the sequence of label content and the number of labels in the image to be tested are output. The sequence of label content is determined based on the position of the labels attached to the ionization separation panel to be tested.

[0006] Secondly, embodiments of the present invention also provide a tag identification device for an ionization separation panel, the device comprising: The first test point cloud determination module is used to acquire the test image associated with the test ionization separation panel and determine the first test point cloud associated with the test image; wherein, the test ionization separation panel is affixed with a label; The feature matrix determination module is used to process the first test point cloud and the first standard point cloud using the same rules to obtain a first feature matrix corresponding to the first test point cloud and a second feature matrix corresponding to the first standard point cloud; wherein, the first standard point cloud is determined based on the CAD model of the ionization separation panel to be tested; The target registration matrix determination module is used to process the second point cloud to be tested and the second standard point cloud based on the first feature matrix and the second feature matrix to obtain a target registration matrix corresponding to the second point cloud to be tested. The second point cloud to be tested is a point cloud after downsampling the first point cloud to be tested, and the second standard point cloud is a point cloud after downsampling the first standard point cloud. The downsampling rules are the same. The candidate image determination module is used to perform rotation correction on the candidate image according to the target registration matrix to obtain the corrected candidate image; The label content sequence output module is used to perform information recognition on the image to be used based on the trained visual recognition model, and output the label content sequence and the number of labels in the image to be used; wherein, the label content sequence is determined based on the position of the labels attached to the ionization separation panel to be tested.

[0007] Thirdly, embodiments of the present invention also provide an electronic device, the electronic device comprising: One or more processors; Storage device for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the tag identification method for the ionization separation panel as described in any embodiment of the present invention.

[0008] Fourthly, embodiments of the present invention also provide a storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to perform a tag identification method for an ionization separation panel as described in any of the embodiments of the present invention.

[0009] Fifthly, embodiments of the present invention also provide a computer program product, including a computer program, characterized in that, when executed by a processor, the computer program implements the tag identification method for an ionization separation panel as described in any embodiment of the present invention.

[0010] The technical solution of this disclosure first acquires a test image associated with the ionization separation panel under test and determines a first test point cloud associated with the test image. Then, the first test point cloud and a first standard point cloud are processed using the same rules to obtain a first feature matrix corresponding to the first test point cloud and a second feature matrix corresponding to the first standard point cloud. Next, based on the first and second feature matrices, the second test point cloud and the second standard point cloud are processed to obtain a target registration matrix corresponding to the second test point cloud. Further, based on the target registration matrix, the test image is rotated and corrected to obtain a corrected image for use. Finally, based on a trained visual recognition model, information recognition is performed on the image for use, outputting the label content sequence and the number of labels in the image for use. This solves the problems of high reliance on manual inspection, low detection efficiency, and low detection accuracy in the prior art for detecting labels on ionization separation panels, which relies on a combination of manual visual inspection and QR code scanning verification. This embodiment of the disclosure realizes the determination of the tag content sequence and the number of tags based on the first standard point cloud and the first test point cloud of the ionization separation panel, thereby improving the accuracy and stability of tag content recognition. Attached Figure Description

[0011] To more clearly illustrate the technical solutions of exemplary embodiments of the present invention, the accompanying drawings used in describing the embodiments are briefly introduced below. Obviously, the accompanying drawings described are only a portion of the drawings of the embodiments to be described in this invention, and not all of the drawings. For those skilled in the art, other drawings can be obtained from these drawings without any creative effort.

[0012] Figure 1 This is a schematic flowchart of a tag identification method for an ionization separation panel provided in an embodiment of this disclosure; Figure 2 This is a schematic diagram of a first standard point cloud provided in an embodiment of this disclosure; Figure 3 This is a schematic diagram of an image to be tested provided in an embodiment of this disclosure; Figure 4 This is a schematic diagram of a first point cloud to be measured provided in an embodiment of this disclosure; Figure 5 This is a schematic diagram of a point cloud after coarse registration provided in an embodiment of this disclosure; Figure 6 This is a schematic diagram of a finely registered point cloud provided in an embodiment of this disclosure; Figure 7 This is a schematic diagram of a potential image provided in an embodiment of this disclosure; Figure 8 This is a schematic diagram of a label provided in an embodiment of this disclosure; Figure 9 This is a schematic diagram of a tag content sequence provided in an embodiment of this disclosure; Figure 10 This is a schematic flowchart of a tag identification method for an ionization separation panel provided in an embodiment of this disclosure; Figure 11 This is a schematic diagram of a second point cloud to be measured provided in an embodiment of this disclosure; Figure 12 This is a schematic diagram of a second standard point cloud provided in an embodiment of this disclosure; Figure 13 This is a schematic diagram of the structure of a tag recognition device for an ionization separation panel provided in an embodiment of this disclosure; Figure 14 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this disclosure. Detailed Implementation

[0013] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, the accompanying drawings show only the parts relevant to the present invention, and not all of the structures.

[0014] Before introducing the technical solutions provided by the embodiments of this disclosure, the application scenarios can be illustrated first. The technical solutions provided by the embodiments of this disclosure can be applied to scenarios involving the identification of tags on ionization separation panels. Based on the technical solutions of the embodiments of this disclosure, the tag content sequence and the number of tags can be determined based on a first standard point cloud and a first test point cloud of the ionization separation panel under test, thereby improving the accuracy and stability of tag content identification.

[0015] Example 1 Figure 1 This is a schematic flowchart of a label identification method for an ionization separation panel provided in this disclosure. This disclosure is applicable to the situation of identifying labels on ionization separation panels. The method can be executed by a label identification device for an ionization separation panel. The device can be implemented in the form of software and / or hardware. The hardware can be a mobile electronic device. The electronic device can execute the label identification method for an ionization separation panel provided in this technical solution.

[0016] like Figure 1 As shown, the method includes: S110. Acquire the image to be tested associated with the ionization separation panel to be tested, and determine the first point cloud to be tested associated with the image to be tested.

[0017] The panel to be tested for ionization separation has a label attached to it.

[0018] It should be noted that the ionization separation panel under test refers to an ionization separation panel that requires label testing. For example, the ionization separation panel under test can be a sheet metal part manufactured for aerospace. A label is affixed to the surface of the ionization separation panel under test. The label is used to identify relevant information about the ionization separation panel under test. For example, the relevant information about the ionization separation panel under test can be its model number and specifications. See also... Figure 2 Tags can include regular tags and general tags. 1-1, 1-2....6-1 represent regular tags, and general tags are like... Figure 2 As shown. 1-1 and 1-2 represent the contents of several labels in column 1. , ... indicates the first The column label content. It should be noted that after aligning the ionization separation panel under test, the column direction is... The axis is parallel, and the row direction is parallel to the axis. The axes are parallel. It should also be noted that... Figure 2 The distinction between the 6-1 tag and the main tag can be determined by their content.

[0019] It should be noted that the image under test refers to an image specifically used for detecting the label on the ionization separation panel under test. The image under test can be a color image captured by a color depth camera, containing RGB visual information of the ionization separation panel and label under test, and serves as the raw image material for subsequent label recognition. See also... Figure 3 , Figure 3 The image to be tested is associated with the ionization separation panel under test. The first test point cloud refers to the three-dimensional point cloud data that corresponds one-to-one with the test image. The first test point cloud contains the three-dimensional spatial coordinate information of the ionization separation panel under test, and each pixel in the test image corresponds strictly to a three-dimensional spatial point in the first test point cloud. See also Figure 4 , Figure 4 This is the first point cloud associated with the image to be tested.

[0020] Optionally, the ionization separation panel to be tested is placed on the platform, and the image to be tested corresponding to the ionization separation panel is acquired based on a color depth camera deployed at a position associated with the platform; the image to be tested is converted into a first point cloud to be tested.

[0021] The workbench refers to a dedicated testing platform, which serves as the carrier for placing the ionization separation panel to be tested. The color depth camera is a camera device that is fixed in position relative to the workbench; it can be fixedly mounted above the workbench.

[0022] It should be noted that when acquiring the image corresponding to the ionization separation panel under test, the first step is to place the panel. Specifically, the panel is placed stably on the table, ensuring there is no significant tilt or obstruction, facilitating scanning by the color depth camera. Then, the large-scale visual model monitors the working area in real time through a monitoring camera deployed above the table. Upon detecting the placement of the panel, the color depth camera is automatically activated. Finally, the image corresponding to the panel is acquired. Specifically, the color depth camera works synchronously, acquiring an image containing the RGB information of the panel and its label; this image is the test image.

[0023] It should also be noted that while the color depth camera acquires the image to be tested, it automatically and simultaneously generates the first point cloud to be tested. The color depth camera has a built-in coordinate mapping mechanism, which enables simultaneous acquisition of the image to be tested and the first point cloud to be tested. Each pixel in the image to be tested corresponds to the visual information of a point on the surface of the ionization separation panel to be tested. Therefore, the three-dimensional spatial coordinates corresponding to that pixel are captured simultaneously, and the three-dimensional coordinates corresponding to all pixels are integrated to form the first point cloud to be tested, which contains the complete three-dimensional geometric features of the ionization separation panel to be tested.

[0024] Specifically, the ionization separation panel to be tested is placed on a platform. Then, a color depth camera fixed at a relevant position on the platform is used to capture the image corresponding to the ionization separation panel. Finally, the captured image can be directly converted into a corresponding first point cloud based on the guessed depth camera, laying the data foundation for subsequent point cloud registration and label detection.

[0025] S120. The same rules are used to process the first point cloud to be tested and the first standard point cloud to obtain the first feature matrix corresponding to the first point cloud to be tested and the second feature matrix corresponding to the first standard point cloud.

[0026] The first standard point cloud is determined based on the CAD model of the ionization separation panel under test.

[0027] It should be noted that the "same rule" refers to using completely identical algorithms, parameters, and processes when processing both the first test point cloud and the first standard point cloud. The first standard point cloud is a standard 3D point cloud generated based on the CAD model of the ionization separation panel under test. The first standard point cloud refers to the ideal point cloud of the ionization separation panel in its standard assembly position, fully representing the standard 3D geometric features and standard label position information of the ionization separation panel under test. The first standard point cloud can serve as a benchmark for subsequent comparison and registration with the first test point cloud.

[0028] It should be noted that the CAD model of the ionization separation panel under test refers to the three-dimensional computer-aided design model generated during the design phase of the panel. This model serves as the original basis for generating the first standard point cloud and includes all design information such as the standard dimensions, shape, and standard label positions of the ionization separation panel under test. See also... Figure 2 , Figure 2 This is the first standard point cloud.

[0029] It should also be noted that the first feature matrix is ​​a matrix obtained by processing the first test point cloud according to the same rules, used to characterize the features of the test point cloud. The core of the first feature matrix is ​​a set of FPFH feature descriptors used to describe the three-dimensional features of the first test point cloud. The second feature matrix is ​​a matrix obtained by processing the first standard point cloud according to the same rules, used to characterize the features of the standard point cloud. The second feature matrix is ​​a set of FPFH feature descriptors of the first standard point cloud, serving as a benchmark for similarity matching with the first feature matrix.

[0030] Specifically, after determining the first test point cloud based on the image to be tested and the first standard point cloud based on the CAD model of the ionization separation panel to be tested, the first test point cloud and the first standard point cloud are processed using the same rules. This yields a first feature matrix characterizing the features of the first test point cloud and a second feature matrix characterizing the features of the first standard point cloud.

[0031] S130. Based on the first feature matrix and the second feature matrix, process the second point cloud to be measured and the second standard point cloud to obtain the target registration matrix corresponding to the second point cloud to be measured.

[0032] The second point cloud to be tested is a point cloud after downsampling the first point cloud to be tested, and the second standard point cloud is a point cloud after downsampling the first standard point cloud. The downsampling rules are the same.

[0033] It should be noted that the second point cloud to be tested can be a simplified version of the first point cloud to be tested, obtained by voxel mesh downsampling. The second standard point cloud is a simplified version of the first standard point cloud, obtained by voxel mesh downsampling. The second standard point cloud and the second point cloud to be tested use the same downsampling rules for subsequent synchronous processing and matching.

[0034] It should also be noted that the target registration matrix refers to the mathematical matrix obtained after solving the spatial attitude and position relationship between the second point cloud to be measured and the second standard point cloud through a specific registration algorithm. This matrix is ​​used to characterize the spatial transformation relationship required to map from the coordinate system of the second standard point cloud to the coordinate system of the second point cloud to be measured.

[0035] Optionally, the first feature matrix and the second feature matrix are used as input data for the global registration algorithm to obtain the rigid transformation matrix, and the second point cloud to be measured is registered based on the rigid transformation matrix; a point-to-surface distance error function is constructed, and the target registration matrix with the smallest error is determined based on the registered second point cloud to be measured and the second standard point cloud.

[0036] The rigid transformation matrix includes at least a rotation matrix and a translation matrix.

[0037] The global registration algorithm can be a random sample consensus algorithm. As a highly robust iterative algorithm, the random sample consensus algorithm is used to find the optimal matching relationship from two sets of feature matrices (first and second feature matrices) through random sampling and verification in the presence of noise or outliers, thus solving for the initial rigid transformation relationship. The random sample consensus algorithm can effectively eliminate interference points and find approximate alignment parameters. The rigid transformation matrix is ​​a mathematical matrix describing the rigid body motion of a spatial object, i.e., changing only its position and orientation without altering its shape and size. The rigid transformation matrix includes at least a rotation matrix and a translation matrix. The rotation matrix is ​​used to control the rotation and orientation of the point cloud. The translation matrix is ​​used to control the movement and position of the point cloud. The rigid transformation matrix characterizes the spatial transformation relationship from the coordinate system of the second standard point cloud to the coordinate system of the second point cloud to be measured. The first registration refers to the initial spatial transformation of the second point cloud to be measured, based on the rigid transformation matrix obtained by the global registration algorithm, involving rotation and translation, to roughly align it spatially with the second standard point cloud. See also... Figure 5 , Figure 5 The red point cloud in the image is the registered second point cloud obtained after performing the first registration based on the rigid transformation matrix on the second point cloud to be measured.

[0038] It should be noted that the first and second feature matrices are used as input data to the global registration algorithm. The global registration algorithm randomly selects a subset of feature points for matching, and through multiple iterations and verifications, calculates an initial rigid transformation matrix that allows most point pairs to match successfully. Using this optimal initial rigid transformation matrix, a spatial transformation involving rotation and translation is performed on the second point cloud to be measured, completing the first registration. At this point, the registered second point cloud to be measured is roughly aligned with the second standard point cloud, but slight errors still exist.

[0039] It should be noted that the point-to-plane distance error function is the objective function used to quantify and evaluate registration accuracy. The point-to-plane distance error function calculates the distance from each point in the registered second test point cloud to the plane containing the corresponding point in the second standard point cloud, and uses the sum of all distances as the optimization objective. By minimizing this error value, the most perfect alignment can be found. The target registration matrix with the minimum error is the optimal rigid transformation matrix obtained iteratively by minimizing the point-to-plane distance error. The target registration matrix is ​​the final and most accurate registration parameter, containing the precise rotation angle and translation of the ionization separation panel under test relative to the standard position. See also... Figure 6 , Figure 6 The red point cloud in the image represents the final registered second point cloud obtained after registering it with the target registration matrix that minimizes error. The target registration matrix that minimizes error contains at least the following parameters: a rotation angle θ around the Z-axis representing the registered second point cloud, and a translation vector describing the spatial translation.

[0040] It should also be noted that, in determining the target registration matrix with the minimum error, firstly, a point-to-plane distance error function is constructed based on the registered second test point cloud and the second standard point cloud. The objective of this function is to minimize the sum of the distances from points on the registered second test point cloud to the planes containing corresponding points on the second standard point cloud. Then, iterative optimization is performed, i.e., using the point-to-plane ICP algorithm, with the point-to-plane distance error function as the optimization objective, the obtained rigid transformation matrix is ​​iteratively adjusted. Further, the algorithm continuously calculates and updates the transformation matrix until the value of the error function converges to the global minimum. Finally, the target registration matrix is ​​obtained, which is the rigid transformation matrix with the minimum error obtained at the final convergence, and this is the target registration matrix.

[0041] Specifically, after obtaining the first feature matrix corresponding to the first point cloud to be tested and the second feature matrix corresponding to the first standard point cloud, a coarse matching is first performed using a global registration algorithm to obtain a preliminary spatial transformation relationship. Then, an iterative optimization algorithm is used for fine matching to finally obtain the target registration matrix with the highest accuracy, thereby achieving accurate spatial mapping between the ionization separation panel to be tested and the standard ionization separation panel.

[0042] S140. Based on the target registration matrix, perform rotation correction on the image to be tested to obtain the corrected image to be used.

[0043] Rotation correction, based on the target registration matrix, is an operation performed on the image under test. It can correct orientation deviations of corresponding physical objects in the image, such as tilting or rotating physical objects, which can cause image orientation errors.

[0044] It should be noted that the corrected image to be used refers to the final image that has been rotated and corrected, meets the specifications, and can be directly used for subsequent operations.

[0045] Optionally, based on the rotation angle on the Z-axis in the target registration matrix, the pixels in the image to be tested are rotated and offset corrected to obtain the image to be used.

[0046] It should be noted that the rotation angle on the Z-axis in the target registration matrix is ​​a parameter describing the rotation angle of the physical object around the vertical axis, i.e., the Z-axis. Rotation offset correction refers to the fine-grained positional offset correction of each pixel in the image under test based on the rotation angle on the Z-axis. This not only adjusts the overall orientation but also accurately corrects the spatial position of individual pixels, completely eliminating the minute positional deviations in the image caused by the rotation or placement of physical objects.

[0047] Specifically, spatial transformation parameters determined by the target registration matrix are used to correct position and orientation deviations in the image under test.

[0048] For example, based on the Z-axis rotation angle θ extracted from the target registration matrix, and using the center coordinates of the image under test as the rotation reference point, a bilinear interpolation algorithm is used to perform a rotation transformation on the image under test, generating a corrected image to eliminate image distortion caused by the placement angle deviation of the ionization separation panel under test. Assuming the calculated rotation angle θ is -3.001°, the image under test is subjected to rotation offset correction, resulting in the following image: Figure 7 As shown.

[0049] S150. Based on the trained visual recognition model, perform information recognition on the image to be used, and output the label content sequence and the number of labels in the image to be used.

[0050] The label content sequence is determined based on the label positions affixed to the ionization separation panel under test. The label count is the actual number of labels identified by the visual recognition model from the image to be used.

[0051] It should be noted that the sequence of label content is not random, but rather strictly ordered according to the actual vertical and horizontal spatial positions of the labels on the ionization separation panel under test. For example, see... Figure 8 After the trained visual recognition model recognizes information from the image to be used, the output label is a blue box.

[0052] Optionally, the text information in the image to be used is identified based on a visual recognition model, and the text information in the preset area is extracted based on a semantic segmentation network; the text information is sorted in a grid according to a preset direction to generate a sequence of label content attached to the ionization separation panel to be tested according to spatial location; the number of labels is counted.

[0053] It should be noted that the semantic segmentation network is used to accurately segment the regions containing labels from the image, distinguishing the label regions from the background regions, and retaining only the text information within the label regions. The preset direction can be vertical, meaning the text information is sorted from top to bottom and left to right. See also... Figure 9 The label area is divided into virtual grids by rows and columns, and the label text information is sorted according to the grid position to ensure that the sequence order is consistent with the physical position.

[0054] In this embodiment, the visual recognition model also outputs component labels of the ionization separation panel to be tested. The method further includes: retrieving the label specification sequence of the target component based on the component label; determining the label recognition result as the first result when the label content corresponding to the same position in the label content sequence and the label specification sequence is the same; and determining the label recognition result as the first result when the number of labels is the same as the number of label information corresponding to the label specification sequence.

[0055] The label specification sequence includes information on multiple labels arranged spatially. The first result is used to characterize the accuracy of the labels attached to the ionization separation panel under test.

[0056] It should be noted that the component label on the ionization separation panel under test is output by the visual recognition model and is used to identify the ionization separation panel itself. Its core function is to quickly locate the corresponding label specification for that component in the standard database. See [link to documentation]. Figure 9 D1213-FIR and other labels are component tags of the ionization separation panel under test. The label specification sequence of the target component is a set of standard tags corresponding to a specific ionization separation panel under test, stored in a standard database. It contains multiple tag information, and all tag information is arranged according to the preset spatial position of the component's tags. The first result is one type of tag identification result, used to characterize the accuracy of the tags attached to the ionization separation panel under test. The tag identification result as the first result must simultaneously meet two conditions: the tag content sequence is consistent with the content at the corresponding position of the label specification sequence, and the number of tags is consistent with the number of tags in the label specification sequence.

[0057] Specifically, the image to be used is input into the visual recognition model. The pre-trained visual recognition model is used to automatically recognize the image to be used, and finally obtains the sequence of label content arranged according to the actual spatial position of the label, as well as the total number of recognized labels.

[0058] The technical solution of this disclosure first acquires a test image associated with the ionization separation panel under test and determines a first test point cloud associated with the test image. Then, the first test point cloud and a first standard point cloud are processed using the same rules to obtain a first feature matrix corresponding to the first test point cloud and a second feature matrix corresponding to the first standard point cloud. Next, based on the first and second feature matrices, the second test point cloud and the second standard point cloud are processed to obtain a target registration matrix corresponding to the second test point cloud. Further, based on the target registration matrix, the test image is rotated and corrected to obtain a corrected image for use. Finally, based on a trained visual recognition model, information recognition is performed on the image for use, outputting the label content sequence and the number of labels in the image for use. This solves the problems of high reliance on manual inspection, low detection efficiency, and low detection accuracy in the prior art for detecting labels on ionization separation panels, which relies on a combination of manual visual inspection and QR code scanning verification. This embodiment of the disclosure realizes the determination of the tag content sequence and the number of tags based on the first standard point cloud and the first test point cloud of the ionization separation panel, thereby improving the accuracy and stability of tag content recognition.

[0059] Example 2 Figure 10 This is a flowchart illustrating the tag identification method for an ionization separation panel provided in this embodiment of the invention. Based on the aforementioned embodiments, a more detailed explanation is provided regarding the processing of a first test point cloud and a first standard point cloud using the same rules to obtain a first feature matrix corresponding to the first test point cloud and a second feature matrix corresponding to the first standard point cloud. Specific implementation methods can be found in the technical solution of this embodiment. Technical terms that are the same as or corresponding to those in the above embodiments will not be repeated here.

[0060] like Figure 10 As shown, the method specifically includes the following steps: S210. Acquire the image to be tested associated with the ionization separation panel to be tested, and determine the first point cloud to be tested associated with the image to be tested.

[0061] S220. Perform voxel mesh downsampling on the first point cloud to be measured and the first standard point cloud to obtain the second point cloud to be measured corresponding to the first point cloud to be measured and the second standard point cloud corresponding to the first standard point cloud.

[0062] Among them, voxel mesh downsampling processing, as a point cloud simplification algorithm, can divide the three-dimensional space into uniform small cubes, i.e., voxels. Within each voxel, only one representative point is retained, which can reduce the number of point clouds while preserving the core geometric features of the ionization separation panel under test, thereby improving the efficiency of subsequent processing.

[0063] It should be noted that the second point cloud to be measured is a simplified version of the first point cloud to be measured, obtained by voxel mesh downsampling. See [link / reference] Figure 11 , Figure 11 The second test point cloud has a reduced number of points, improving computational efficiency while still retaining the core 3D features of the ionization separation panel under test. The second standard point cloud is a simplified version obtained by downsampling the first standard point cloud using a voxel grid. See also Figure 12 , Figure 12 This serves as the second standard point cloud, employing the same downsampling rules as the second point cloud to be tested, for subsequent synchronization processing and matching.

[0064] Specifically, when performing voxel grid downsampling on the first test point cloud and the first standard point cloud, firstly, the size of the voxel grid can be determined according to the size of the ionization separation panel under test and the detection accuracy, ensuring that core features are preserved after simplification. Then, downsampling processing is performed, that is, the space of the first test point cloud is divided using a voxel grid downsampling algorithm, and representative points of each voxel are retained to obtain the second test point cloud. The first standard point cloud is processed synchronously, that is, downsampling is performed on the first standard point cloud using the same voxel size and algorithm to obtain the second standard point cloud.

[0065] S230. Perform normal estimation on the second point cloud to be measured and the second standard point cloud to obtain the first feature matrix corresponding to the second point cloud to be measured and the second feature matrix corresponding to the second standard point cloud.

[0066] The moving least squares method can be used to estimate the normals of the second test point cloud and the second standard point cloud. During normal estimation, the second test point cloud and the second standard point cloud are processed separately, analyzing the three-dimensional spatial relationship between each point and its neighboring points to calculate the normal direction of each point, i.e., the three-dimensional direction vector perpendicular to the surface of the ionization separation panel where that point is located. Then, the normal directions of all points are integrated to form a set of normal information for the corresponding point cloud.

[0067] It should be noted that the first feature matrix refers to the feature matrix calculated after estimating the normals of the second point cloud to be tested. The first feature matrix can completely characterize the local three-dimensional features of the actual point cloud of the ionization separation panel to be tested, and is used for subsequent similarity matching with the second feature matrix. The second feature matrix refers to the feature matrix calculated after estimating the normals of the second standard point cloud. The second feature matrix can completely characterize the local three-dimensional features of the standard point cloud of the ionization separation panel to be tested, and serves as the benchmark for similarity matching with the first feature matrix.

[0068] Optionally, normal estimation is performed on the second point cloud to be measured and the second standard point cloud to obtain the first normal information corresponding to the second point cloud to be measured and the second normal information corresponding to the second standard point cloud; a fast point feature histogram is calculated based on the first normal information to obtain the first feature matrix, and a fast point feature histogram is calculated based on the second normal information to obtain the second feature matrix.

[0069] It should be noted that the moving least squares method is used to analyze the spatial position of each point's neighboring points in the second point cloud to be measured, calculate the normal direction of each point, and integrate all normal directions to obtain the first normal information. The second standard point cloud is processed simultaneously, i.e., the same moving least squares method and parameters are used to analyze the spatial position of each point's neighboring points in the second standard point cloud, calculate the normal direction of each point, and integrate all normal directions to obtain the second normal information.

[0070] It should also be noted that the first normal information is input into the fast point feature histogram algorithm. By statistically analyzing the normal direction of each point in the first normal information and its relationship with the normals of adjacent points, a feature histogram for each point is generated. The feature histograms of all points are then integrated to obtain the first feature matrix. Using the same fast point feature histogram algorithm and parameters, based on the second normal information, the feature histogram of each point in the second standard point cloud is calculated, and the two are then integrated to obtain the second feature matrix.

[0071] Specifically, after obtaining the second point cloud corresponding to the first point cloud to be tested, and the second standard point cloud corresponding to the first standard point cloud, normal estimation is performed on the second point cloud to obtain the first normal information corresponding to the second point cloud to be tested. Normal estimation is performed on the second standard point cloud to obtain the second normal information corresponding to the second standard point cloud. The first normal information is input into the fast point feature histogram algorithm to obtain the first feature matrix, and the second normal information is input into the fast point feature histogram algorithm to obtain the second feature matrix.

[0072] S240. Based on the first feature matrix and the second feature matrix, process the second point cloud to be measured and the second standard point cloud to obtain the target registration matrix corresponding to the second point cloud to be measured.

[0073] S250. Based on the target registration matrix, perform rotation correction on the image to be tested to obtain the corrected image to be used.

[0074] S260. Based on the trained visual recognition model, perform information recognition on the image to be used, and output the label content sequence and the number of labels in the image to be used.

[0075] The technical solution of this disclosure involves acquiring a test image associated with an ionization separation panel under test and determining a first test point cloud associated with the test image. Then, voxel grid downsampling processing is performed on the first test point cloud and a first standard point cloud to obtain a second test point cloud corresponding to the first test point cloud and a second standard point cloud corresponding to the first standard point cloud. Normal estimation is performed on the second test point cloud and the second standard point cloud to obtain a first feature matrix corresponding to the second test point cloud and a second feature matrix corresponding to the second standard point cloud. Then, based on the first and second feature matrices, the second test point cloud and the second standard point cloud are processed to obtain a target registration matrix corresponding to the second test point cloud. Further, based on the target registration matrix, rotation correction is performed on the test image to obtain a corrected test image. Finally, based on the trained visual recognition model, information recognition is performed on the image to be used, and the label content sequence and label number in the image to be used are output. Through a series of operations such as voxel grid downsampling and normal estimation, the recognition error caused by the placement offset of the ionization separation panel to be tested and image distortion is effectively eliminated, ensuring the accuracy of the label content sequence and label number recognition.

[0076] Example 3 Figure 13 This is a schematic diagram of the structure of the tag recognition device for the ionization separation panel provided in the embodiments of this disclosure, as shown below. Figure 13 As shown, the device includes: a first point cloud determination module 310, a feature matrix determination module 320, a target registration matrix determination module 330, a candidate image determination module 340, and a label content sequence output module 350.

[0077] A first test point cloud determination module is used to acquire a test image associated with the ionization separation panel under test, and determine a first test point cloud associated with the test image; wherein, a label is attached to the ionization separation panel under test; a feature matrix determination module is used to process the first test point cloud and the first standard point cloud using the same rules to obtain a first feature matrix corresponding to the first test point cloud and a second feature matrix corresponding to the first standard point cloud; wherein, the first standard point cloud is determined based on the CAD model of the ionization separation panel under test; a target registration matrix determination module is used to determine the second test point cloud and the second standard point cloud based on the first feature matrix and the second feature matrix. The point cloud is processed to obtain a target registration matrix corresponding to the second test point cloud, wherein the second test point cloud is a point cloud downsampled from the first test point cloud, and the second standard point cloud is a point cloud downsampled from the first standard point cloud, with the same downsampling rules; a candidate image determination module is used to perform rotation correction on the test image according to the target registration matrix to obtain a corrected candidate image; a label content sequence output module is used to perform information recognition on the candidate image based on a trained visual recognition model and output the label content sequence and the number of labels in the candidate image; wherein the label content sequence is determined based on the label positions attached to the ionization separation panel under test.

[0078] The technical solution of this disclosure first acquires a test image associated with the ionization separation panel under test and determines a first test point cloud associated with the test image. Then, the first test point cloud and a first standard point cloud are processed using the same rules to obtain a first feature matrix corresponding to the first test point cloud and a second feature matrix corresponding to the first standard point cloud. Next, based on the first and second feature matrices, the second test point cloud and the second standard point cloud are processed to obtain a target registration matrix corresponding to the second test point cloud. Further, based on the target registration matrix, the test image is rotated and corrected to obtain a corrected image for use. Finally, based on a trained visual recognition model, information recognition is performed on the image for use, outputting the label content sequence and the number of labels in the image for use. This solves the problems of high reliance on manual inspection, low detection efficiency, and low detection accuracy in the prior art for detecting labels on ionization separation panels, which relies on a combination of manual visual inspection and QR code scanning verification. This embodiment of the disclosure realizes the determination of the tag content sequence and the number of tags based on the first standard point cloud and the first test point cloud of the ionization separation panel, thereby improving the accuracy and stability of tag content recognition.

[0079] Based on the above technical solutions, the first test point cloud determination module 310 is further configured to place the test ionization separation panel on the table and acquire the test image corresponding to the test ionization separation panel based on a color depth camera deployed at a position associated with the table; and convert the test image into the first test point cloud.

[0080] Based on the above technical solutions, the feature matrix determination module 320 further includes: a voxel grid downsampling processing submodule and a normal estimation submodule.

[0081] The voxel grid downsampling processing submodule is used to perform voxel grid downsampling processing on the first test point cloud and the first standard point cloud to obtain a second test point cloud corresponding to the first test point cloud and a second standard point cloud corresponding to the first standard point cloud. The normal estimation submodule is used to perform normal estimation on the second point cloud to be measured and the second standard point cloud to obtain a first feature matrix corresponding to the second point cloud to be measured and a second feature matrix corresponding to the second standard point cloud.

[0082] Based on the above technical solutions, the normal estimation submodule includes: a normal estimation unit and a fast point feature histogram calculation unit.

[0083] The normal estimation unit is used to perform normal estimation on the second point cloud to be measured and the second standard point cloud to obtain first normal information corresponding to the second point cloud to be measured and second normal information corresponding to the second standard point cloud. The fast point feature histogram calculation unit is used to calculate the fast point feature histogram based on the first normal information to obtain the first feature matrix, and to calculate the fast point feature histogram based on the second normal information to obtain the second feature matrix.

[0084] Based on the above technical solutions, the target registration matrix determination module 330 is further configured to use the first feature matrix and the second feature matrix as input data for the global registration algorithm to obtain a rigid transformation matrix, and perform a first registration on the second point cloud to be measured based on the rigid transformation matrix, wherein the rigid transformation matrix includes at least a rotation matrix and a translation matrix; construct a point-to-surface distance error function, and determine the target registration matrix with the smallest error based on the registered second point cloud to be measured and the second standard point cloud.

[0085] Based on the above technical solutions, the candidate image determination module 340 is further used to perform rotation offset correction on the pixels in the candidate image based on the rotation angle on the Z-axis in the target registration matrix, so as to obtain the candidate image.

[0086] Based on the above technical solutions, the label content sequence output module 350 is further configured to identify text information in the image to be used based on the visual recognition model, and extract text information in a preset area according to the semantic segmentation network; sort the text information in a grid according to a preset direction to generate a label content sequence attached to the ionization separation panel to be tested according to spatial location; and count the number of labels.

[0087] Based on the above technical solutions, the device further includes: a tag recognition result determination module, used to retrieve the tag specification sequence of the target component according to the component tag; wherein, the tag specification sequence includes multiple tag information arranged in spatial position; when the tag content corresponding to the same position in the tag content sequence and the tag specification sequence is the same, the tag recognition result is determined as a first result; when the number of tags is the same as the number of tag information corresponding to the tag specification sequence, the tag recognition result is determined as a first result; wherein, the first result is used to characterize the accuracy of the tag attached to the ionization separation panel under test.

[0088] The tag identification device for the ionization separation panel provided in this disclosure can execute the tag identification method for the ionization separation panel provided in any embodiment of this disclosure, and has the corresponding functional modules and beneficial effects for executing the method.

[0089] It is worth noting that the various units and modules included in the above-mentioned device are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be realized; in addition, the specific names of each functional unit are only for easy differentiation and are not used to limit the protection scope of the embodiments of this disclosure.

[0090] Example 4 Figure 14 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this disclosure. Refer to the following... Figure 14 It illustrates an electronic device suitable for implementing embodiments of the present disclosure (e.g., Figure 14 The diagram below shows the structure of the terminal device or server 500. The terminal device in this embodiment may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), and vehicle terminals (e.g., vehicle navigation terminals). Figure 14 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.

[0091] like Figure 14As shown, electronic device 500 may include a processing unit (e.g., a central processing unit, a graphics processing unit, etc.) 501, which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 502 or a program loaded from storage device 508 into random access memory (RAM) 503. The RAM 503 also stores various programs and data required for the operation of electronic device 500. The processing unit 501, ROM 502, and RAM 503 are interconnected via bus 504. An edit / output (I / O) interface 505 is also connected to bus 504.

[0092] Typically, the following devices can be connected to I / O interface 505: input devices 506 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 507 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 508 including, for example, magnetic tapes, hard disks, etc.; and communication devices 509. Communication device 509 allows electronic device 500 to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 14 An electronic device 500 with various devices is shown; however, it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed alternatively.

[0093] In particular, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 509, or installed from a storage device 508, or installed from a ROM 502. When the computer program is executed by the processing device 501, it performs the functions defined in the methods of embodiments of this disclosure.

[0094] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0095] The electronic device provided in this embodiment and the tag identification method for the ionization separation panel provided in the above embodiment belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0096] Example 5 This disclosure provides a computer storage medium storing a computer program that, when executed by a processor, implements the tag identification method for the ionization separation panel provided in the above embodiments.

[0097] It should be noted that the computer-readable medium described in this disclosure can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this disclosure, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this disclosure, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0098] In some implementations, the server may communicate using any currently known or future-developed network protocol such as HTTP (Hypertext Transfer Protocol) and may interconnect with digital data communication (e.g., communication networks) of any form or medium. Examples of communication networks include local area networks (“LANs”), wide area networks (“WANs”), the Internet (e.g., the Internet), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future-developed networks.

[0099] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device.

[0100] The aforementioned computer-readable medium carries one or more programs that, when executed by the electronic device, cause the electronic device to: Acquire a test image associated with the ionization separation panel under test, and determine a first test point cloud associated with the test image; wherein, the ionization separation panel under test is affixed with a label; The same rules are applied to the first test point cloud and the first standard point cloud to obtain a first feature matrix corresponding to the first test point cloud and a second feature matrix corresponding to the first standard point cloud; wherein, the first standard point cloud is determined based on the CAD model of the ionization separation panel to be tested; Based on the first feature matrix and the second feature matrix, the second point cloud to be tested and the second standard point cloud are processed to obtain a target registration matrix corresponding to the second point cloud to be tested. The second point cloud to be tested is a point cloud after downsampling the first point cloud to be tested, and the second standard point cloud is a point cloud after downsampling the first standard point cloud. The downsampling rules are the same. Based on the target registration matrix, the image to be tested is rotated and corrected to obtain the corrected image to be used; The image to be tested is identified based on a trained visual recognition model, and the sequence of label content and the number of labels in the image to be tested are output. The sequence of label content is determined based on the position of the labels attached to the ionization separation panel to be tested.

[0101] Computer program code for performing the operations of this disclosure can be written in one or more programming languages ​​or a combination thereof, including but not limited to object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0102] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0103] The units described in the embodiments of this disclosure can be implemented in software or hardware. The names of the units are not, in some cases, intended to limit the specific unit.

[0104] The functions described above in this document can be performed at least in part by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), system-on-a-chip (SoCs), complex programmable logic devices (CPLDs), and so on.

[0105] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0106] The above description is merely a preferred embodiment of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features disclosed in this disclosure that have similar functions.

[0107] Furthermore, while the operations are described in a specific order, this should not be construed as requiring these operations to be performed in the specific order shown or in a sequential order. In certain environments, multitasking and parallel processing may be advantageous. Similarly, while several specific implementation details are included in the above discussion, these should not be construed as limiting the scope of this disclosure. Certain features described in the context of individual embodiments may also be implemented in combination in a single embodiment. Conversely, various features described in the context of a single embodiment may also be implemented individually or in any suitable sub-combination in multiple embodiments.

[0108] Although the subject matter has been described using language specific to structural features and / or methodological logic, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. Rather, the specific features and actions described above are merely illustrative examples of implementing the claims.

Claims

1. A label identification method for an ionization separation panel, characterized in that, include: Acquire a test image associated with the ionization separation panel under test, and determine a first test point cloud associated with the test image; wherein, the ionization separation panel under test is affixed with a label; The same rules are applied to the first test point cloud and the first standard point cloud to obtain a first feature matrix corresponding to the first test point cloud and a second feature matrix corresponding to the first standard point cloud; wherein, the first standard point cloud is determined based on the CAD model of the ionization separation panel to be tested; Based on the first feature matrix and the second feature matrix, the second point cloud to be tested and the second standard point cloud are processed to obtain a target registration matrix corresponding to the second point cloud to be tested. The second point cloud to be tested is a point cloud after downsampling the first point cloud to be tested, and the second standard point cloud is a point cloud after downsampling the first standard point cloud. The downsampling rules are the same. Based on the target registration matrix, the image to be tested is rotated and corrected to obtain the corrected image to be used; The image to be tested is identified based on a trained visual recognition model, and the sequence of label content and the number of labels in the image to be tested are output. The sequence of label content is determined based on the position of the labels attached to the ionization separation panel to be tested.

2. The method according to claim 1, characterized in that, The step of acquiring the image to be tested associated with the ionization separation panel under test, and determining the first point cloud to be tested associated with the image to be tested, includes: The ionization separation panel to be tested is placed on a platform, and the test image corresponding to the ionization separation panel is acquired by a color depth camera deployed at a position associated with the platform. The image to be tested is converted into the first point cloud to be tested.

3. The method according to claim 1, characterized in that, The process of processing the first point cloud to be tested and the first standard point cloud using the same rules to obtain a first feature matrix corresponding to the first point cloud to be tested and a second feature matrix corresponding to the first standard point cloud includes: Voxel grid downsampling processing is performed on the first test point cloud and the first standard point cloud to obtain the second test point cloud corresponding to the first test point cloud and the second standard point cloud corresponding to the first standard point cloud. Normal estimation is performed on the second point cloud to be measured and the second standard point cloud to obtain a first feature matrix corresponding to the second point cloud to be measured and a second feature matrix corresponding to the second standard point cloud.

4. The method according to claim 3, characterized in that, The step of estimating the normals of the second point cloud to be measured and the second standard point cloud to obtain a first feature matrix corresponding to the second point cloud to be measured and a second feature matrix corresponding to the second standard point cloud includes: Normal estimation is performed on the second point cloud to be measured and the second standard point cloud to obtain the first normal information corresponding to the second point cloud to be measured and the second normal information corresponding to the second standard point cloud. Based on the first normal information, a fast point feature histogram is calculated to obtain the first feature matrix, and based on the second normal information, a fast point feature histogram is calculated to obtain the second feature matrix.

5. The method according to claim 1, characterized in that, The step of processing the second point cloud to be measured and the second standard point cloud based on the first feature matrix and the second feature matrix to obtain a target registration matrix corresponding to the second point cloud to be measured includes: The first feature matrix and the second feature matrix are used as input data for the global registration algorithm to obtain a rigid transformation matrix. The second point cloud to be measured is then registered based on the rigid transformation matrix. The rigid transformation matrix includes at least a rotation matrix and a translation matrix. A point-to-area distance error function is constructed, and based on the registered second point cloud to be measured and the second standard point cloud, the target registration matrix with the minimum error is determined.

6. The method according to claim 1, characterized in that, The step of performing rotation correction on the image to be tested according to the target registration matrix to obtain the corrected image to be used includes: Based on the rotation angle on the Z-axis in the target registration matrix, the pixel points in the image to be tested are rotated and offset corrected to obtain the image to be used.

7. The method according to claim 1, characterized in that, The process of recognizing information from the candidate image based on a trained visual recognition model, and outputting the label content sequence and label count in the candidate image, includes: The visual recognition model identifies the text information in the image to be used, and the semantic segmentation network extracts the text information in the preset region. The text information is sorted in a grid according to a preset direction to generate a sequence of label contents attached to the ionization separation panel under test according to spatial location; Count the number of tags.

8. The method according to claim 1, characterized in that, The visual recognition model also outputs component labels for the ionization separation panel under test, and the method further includes: Based on the component label, retrieve the label specification sequence of the target component; wherein, the label specification sequence includes multiple label information arranged according to spatial location; When the tag content at the same position in the tag content sequence and the tag specification sequence is the same, the tag recognition result is determined as the first result; When the number of tags is the same as the number of tag information corresponding to the tag specification sequence, the tag recognition result is determined as the first result; The first result is used to characterize the accuracy of the label attached to the ionization separation panel under test.

9. A label recognition device for an ionization separation panel, characterized in that, include: The first test point cloud determination module is used to acquire the test image associated with the test ionization separation panel and determine the first test point cloud associated with the test image; wherein, the test ionization separation panel is affixed with a label; The feature matrix determination module is used to process the first test point cloud and the first standard point cloud using the same rules to obtain a first feature matrix corresponding to the first test point cloud and a second feature matrix corresponding to the first standard point cloud; wherein, the first standard point cloud is determined based on the CAD model of the ionization separation panel to be tested; The target registration matrix determination module is used to process the second point cloud to be tested and the second standard point cloud based on the first feature matrix and the second feature matrix to obtain a target registration matrix corresponding to the second point cloud to be tested. The second point cloud to be tested is a point cloud after downsampling the first point cloud to be tested, and the second standard point cloud is a point cloud after downsampling the first standard point cloud. The downsampling rules are the same. The candidate image determination module is used to perform rotation correction on the candidate image according to the target registration matrix to obtain the corrected candidate image; The label content sequence output module is used to perform information recognition on the image to be used based on the trained visual recognition model, and output the label content sequence and the number of labels in the image to be used; wherein, the label content sequence is determined based on the position of the labels attached to the ionization separation panel to be tested.

10. An electronic device, characterized in that, The electronic device includes: One or more processors; Storage device for storing one or more programs. When one or more programs are executed by one or more processors, the one or more processors implement the tag identification method for an ionization separation panel as described in any one of claims 1-8.

Images