A full-process supervision system based on visual offline identification

Abstract

The application discloses a kind of full-process supervision systems based on visual offline identification, and the application relates to component production technical field, it is solved that existing technology mostly stays in single size comparison level, lack of the problem of full-process integrated supervision mechanism from model generation, standard matching, automatic splicing to assembly verification, the unified check feature of the present application is constructed by built-in midpoint, plane midpoint and feature vector, corresponding three-dimensional standard model can be matched from model database quickly and accurately, the identification logic is stable, and the anti-interference is strong, effectively avoid the model mismatching problem caused by similar appearance;Through the vector comparison of reference feature and component feature, length difference and included angle calculation, the assembly accuracy can be quantitatively judged, and the assembly standard, correctable and substandard state can be distinguished, the visualization and digitization of assembly quality are realized, the on-site quick error correction and traceability are facilitated, and the assembly reliability and overall production quality of flow line are effectively improved.

Description

Technical Field

[0001] This invention relates to the field of component manufacturing technology, specifically to a full-process monitoring system based on offline visual recognition. Background Technology

[0002] In modern industrial automated production lines, the identification, assembly, and quality verification of various components are key aspects of the production process.

[0003] Traditional production methods rely heavily on manual visual inspection, two-dimensional image recognition, or simple dimensional checks, which suffer from limited recognition accuracy, susceptibility to lighting and environmental interference, and an inability to fully characterize the three-dimensional structural features of devices. For complex devices with similar shapes, misclassification of models and misalignment during assembly can easily occur, making precise matching difficult.

[0004] With the gradual application of 3D vision technology, although spatial information of devices can be obtained through 3D scanning, most existing technologies only remain at the level of 3D model reconstruction or single-dimensional comparison, lacking a comprehensive integrated monitoring mechanism covering the entire process from model generation, standard matching, automatic stitching to assembly verification. In the model matching process, simple comparison of the overall contour or local feature points is usually adopted. This results in inconsistent feature definitions, poor robustness, and easy matching failure due to scanning errors or pose differences. In the assembly stage, stitching is mostly performed using preset programs, lacking quantitative verification methods for the actual assembly results. This makes it difficult to accurately judge the degree of assembly deviation and distinguish between minor correctable deviations and serious non-conformities, leading to unstable assembly quality, high rework rates, and low production efficiency, failing to meet the requirements of high-precision, high-stability automated production line monitoring.

[0005] Therefore, how to achieve closed-loop supervision of the entire process of automated generation of 3D models of production line components, accurate feature matching, intelligent assembly and splicing, and quantitative verification has become a technical problem that urgently needs to be solved in the field of industrial vision offline recognition and assembly supervision. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a full-process monitoring system based on offline visual recognition, which solves the problem that most existing technologies are limited to single-size comparison and lack an integrated monitoring mechanism covering the entire process from model generation, standard matching, automatic splicing to assembly verification.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a full-process monitoring system based on offline visual recognition, comprising:

[0008] The 3D scanning processing unit performs 3D scanning on the devices present on the production line and generates a three-dimensional single-unit model associated with the corresponding device.

[0009] The model database contains a large number of 3D standard models, all of which are preset models.

[0010] The model verification and recognition end performs 3D entity verification on the generated 3D single-unit model, confirming the built-in midpoint of the 3D single-unit model and the midpoints of different planes. Then, based on the built-in midpoint and the midpoints of the planes, it confirms the verification features associated with the 3D single-unit model. Finally, based on the verification features, it selects and marks the suitable model from the model database. The specific method is as follows:

[0011] Combine the generated 3D single-unit model with the 3D coordinate system, confirm the 3D coordinates associated with different points on the outer surface of the 3D single-unit model, perform mean processing on several confirmed sets of 3D coordinates, confirm the mean coordinates associated with several 3D coordinates, confirm the spatial position associated with the mean coordinates in the 3D coordinate system, and mark the associated spatial position in the 3D single-unit model, which is recorded as the built-in midpoint of the 3D single-unit model.

[0012] Next, the midpoints of the planes associated with different outer surfaces of the three-dimensional single model are determined, so that the single set of outer surfaces is combined with the two-dimensional coordinate system. The two-dimensional coordinates associated with different points on the outer surface are confirmed. Then, the average value of several sets of two-dimensional coordinates is processed to confirm the average coordinates. The point where the average coordinates are located is marked on the outer surface and recorded as the midpoint of the plane associated with the outer surface.

[0013] Starting from the built-in midpoint and ending from the plane midpoint, construct the feature vectors associated with the starting point and the ending point. Confirm the feature vectors associated with several outer surfaces in sequence, and record the confirmed feature vectors as the verification features associated with the three-dimensional single-unit model.

[0014] The specific method for selecting the adaptation model is as follows:

[0015] Extract the 3D standard model from the model database, confirm the marked model midpoints and surface midpoints within the 3D standard model, and confirm the standard features associated with the 3D standard model using the same confirmation method as the verification features.

[0016] The verification features are compared with the standard features to ensure that the starting points associated with the verification features and the standard features coincide. Then, it is determined whether the verification features and the standard features are completely consistent. If they are, the 3D standard model associated with the standard features is recorded as the adaptation model of the 3D single model. If not, no marking is made, and the process of confirming the adaptation model for other 3D standard models continues.

[0017] On the splicing processing end, based on the adapter model marked by the 3D single-unit model, the splicing model and splicing points are confirmed from the model information associated with the adapter model. Different devices are then spliced ​​to generate a spliced ​​assembly. The specific method is as follows:

[0018] Based on the splicing model associated with the adaptation model, the three-dimensional single-unit model associated with the splicing model is identified. The components of the identified three-dimensional single-unit model are recorded as secondary components to be assembled, and the components of the three-dimensional single-unit model associated with the adaptation model are recorded as primary components to be assembled.

[0019] Based on the matching model and the splicing points marked in the splicing model, the main component to be assembled and the secondary component to be assembled are spliced ​​together so that the marked splicing points coincide, and a splicing component is generated.

[0020] The comprehensive verification end performs verification processing on the generated splicing components. It uses the splicing model body associated with the adaptation model and the splicing model as the standard model body, locks the baseline features from the standard model body, and then confirms the component features associated with the splicing components. The baseline features and component features are compared and verified to identify whether the splicing components are assembled correctly. The specific method is as follows:

[0021] Generate a standard model body from the fitted model and the spliced ​​model according to the marked splicing points. In the standard model body, take the midpoint of the fitted model as the starting point and the midpoint of the spliced ​​model as the ending point, and generate the point vector associated with the starting point to the ending point. This vector is denoted as the reference feature associated with the standard model body.

[0022] The two sets of three-dimensional single-unit models associated with the splicing component are confirmed. The built-in midpoint of the three-dimensional single-unit model associated with the fitting model is taken as the starting point, and the built-in midpoint of the three-dimensional single-unit model associated with the splicing model is taken as the ending point. The point vector associated from the starting point to the ending point is generated and recorded as the component feature associated with the splicing component.

[0023] The baseline feature and component feature are compared to ensure that the starting points associated with the two sets of features coincide. The included angle A between the two sets of features is then confirmed. The vector lengths associated with the two sets of features are then confirmed. The vector length of the baseline feature is denoted as L1, and the vector length of the component feature is denoted as L2.

[0024] If L1=L2 and the included angle is 0, no processing is required, which means that the splicing component assembly meets the standard, and an assembly compliance signal is generated for display.

[0025] If L1≠L2, then use: |L1-L2|÷L1=B1 and A÷360=B2 to confirm the two sets of proportional parameters B1 and B2, and use: PD=B1×C1+B2×C2 to confirm the evaluation value PD, where C1 and C2 are preset fixed coefficient factors. If PD≤Y1, then generate an assembly correction signal for display, where Y1 is a preset value. If PD>Y1, then generate an assembly non-compliance signal and display it in time.

[0026] Preferably, if there is no included angle, the included angle is 0.

[0027] Preferably, if L1≠L2, another specific method for identifying whether the splicing components are assembled to standard is also included:

[0028] Based on the generated reference features, a set of spherical ranges are generated with the end point of the reference features as the center of the sphere and the preset radius R as the radius of the sphere, where R is a preset value;

[0029] Align the starting points of the component features with the reference features, and identify whether the ending points of the component features are within the spherical range. If so, generate an assembly correction signal and display it; otherwise, generate an assembly non-compliance signal and display it promptly.

[0030] Preferably, the bottom surface of the three-dimensional single-unit model is the upper base surface of the pipeline.

[0031] This invention provides a full-process monitoring system based on offline visual recognition. Compared with existing technologies, it has the following advantages:

[0032] By constructing unified verification features using built-in midpoints, plane midpoints, and feature vectors, it can quickly and accurately match corresponding 3D standard models from the model database. The recognition logic is stable and has strong anti-interference capabilities, effectively avoiding model mismatch problems caused by similar appearances.

[0033] In the splicing and assembly process, different components are automatically spliced ​​based on the preset adaptation model and splicing points. Standard splicing components can be generated without manual intervention, which significantly improves assembly efficiency and consistency.

[0034] The integrated verification terminal can quantitatively determine the assembly accuracy by comparing the vectors of the baseline features and component features, calculating the length difference and the included angle, and distinguishing between the assembly that meets the standards, can be corrected, and does not meet the standards. This enables visualized and digital supervision of assembly quality, facilitates rapid on-site error correction and traceability, effectively improves the reliability of assembly lines and the overall production quality, and is suitable for automated identification and assembly supervision scenarios of various types of industrial components. Attached Figure Description

[0035] Figure 1 This is a schematic diagram of the principle framework of the present invention. Detailed Implementation

[0036] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0037] First Embodiment

[0038] Please see Figure 1 This application provides a full-process monitoring system based on visual offline recognition, including a 3D scanning processing end, a model database, a model verification and recognition end, a stitching processing end, and a comprehensive verification end. The 3D scanning processing end, the model verification and recognition end, the stitching processing end, and the comprehensive verification end are electrically connected from the output node to the input node in sequence, and the model database is electrically connected to the input node of the model verification and recognition end.

[0039] The 3D scanning processing unit performs 3D scanning on the devices on the production line, generating a three-dimensional single-unit model associated with the corresponding device. The bottom surface of the three-dimensional single-unit model is the upper base surface of the production line. During the scanning process, there are scanning devices around the production line, which can fully scan the outer surface features of the device. Based on the point cloud data confirmed during the scanning process, a single-unit model associated with the corresponding device is generated. The device moves with the production line to the designated scanning position. After the sensor detects that the device is in place, it automatically triggers the 3D scanning equipment (laser / depth camera) to start acquisition. The scanning equipment performs multi-view / single-line scanning on the device to generate raw 3D point cloud data containing the spatial coordinates of the device surface; it automatically filters out noise points and background interference points (such as conveyor belts and fixtures) in the point cloud, retaining only the effective point cloud of the device itself; it segments the independent point cloud of a single device from the cleaned point cloud, removing all background data that is not related to the device; if it is a multi-view scan, it automatically aligns and merges point clouds from different angles to form a complete point cloud of the entire surface of the device; it converts the complete point cloud into a closed, continuous triangular mesh surface, repairs small holes formed by scanning, and obtains a basic 3D model. The method of generating the basic 3D model is quite common in existing technologies, so it will not be described in detail here.

[0040] Among them, the model database contains a large number of 3D standard models, all of which are preset models that are set up and stored in the model database in advance by relevant personnel;

[0041] The model verification and recognition end performs 3D entity verification on the generated 3D single-unit model, confirming the built-in midpoint and the midpoints of different planes of the 3D single-unit model. Based on the built-in midpoint and the midpoints of the planes, it confirms the verification features associated with the 3D single-unit model. Then, based on the verification features, it selects and marks the matching model from the model database. Specifically, the bottom surface of the 3D single-unit model is generally the base surface of the pipeline. This part is generally not scannable, so the surface of the pipeline is used as the bottom surface of the corresponding model. Based on the specific base surface performance characteristics, it confirms the feature points associated with different surfaces of the corresponding model. Based on the comprehensive performance between the feature points, it confirms the multi-directional verification vector associated with the corresponding model. Subsequently, it performs the comparison process between models based on this verification vector to achieve effective adaptation between models, thereby selecting the matching model associated with the corresponding device.

[0042] The specific method for verifying the features of a three-dimensional single-unit model is as follows:

[0043] Combine the generated 3D single-unit model with the 3D coordinate system, confirm the 3D coordinates associated with different points on the outer surface of the 3D single-unit model, perform mean processing on several confirmed sets of 3D coordinates, confirm the mean coordinates associated with several 3D coordinates, confirm the spatial position associated with the mean coordinates in the 3D coordinate system, and mark the associated spatial position in the 3D single-unit model, which is recorded as the built-in midpoint of the 3D single-unit model.

[0044] Next, the midpoints of the planes associated with different outer surfaces of the three-dimensional single model are determined, so that the single set of outer surfaces is combined with the two-dimensional coordinate system. The two-dimensional coordinates associated with different points on the outer surface are confirmed. Then, the average value of several sets of two-dimensional coordinates is processed to confirm the average coordinates. The point where the average coordinates are located is marked on the outer surface and recorded as the midpoint of the plane associated with the outer surface.

[0045] Starting from the built-in midpoint and ending from the plane midpoint, construct the feature vectors associated with the starting point and ending point (the direction of the vectors is the direction of movement from the starting point to the ending point). Confirm the feature vectors associated with several outer surfaces in sequence, and record the confirmed feature vectors as the verification features associated with the three-dimensional single-unit model.

[0046] The specific method for selecting a suitable model from the model database based on verification features is as follows:

[0047] Extract the 3D standard model from the model database, and confirm the marked model midpoints (equivalent to built-in midpoints) and surface midpoints (equivalent to plane midpoints) within the 3D standard model. Then, confirm the standard features associated with the 3D standard model using the same confirmation method as the verification features.

[0048] The verification features are compared with the standard features to ensure that the starting points associated with the verification features and the standard features coincide. Then, it is determined whether the verification features and the standard features are completely consistent (different vectors within the two features completely coincide). If so, the 3D standard model associated with the standard features is recorded as the adaptation model of the 3D single-unit model. If not, no marking is made, and the process of confirming the adaptation model for other 3D standard models continues.

[0049] Specifically, in order to effectively compare and confirm the identified 3D single-unit model with the standard model, it is necessary to confirm the point features of the 3D single-unit model with respect to different base surfaces, thereby effectively confirming the position vectors associated with different point features. The standard model is synchronously set with associated position vectors. By comparing and verifying the verification features of the two confirmed models, the overlap process between the two models can be effectively identified. If the two models completely overlap, it means that the corresponding standard model is the adapted model of the corresponding entity model, which facilitates the subsequent assembly processing of different adapted models and locks the assembly points existing in the corresponding model.

[0050] Second Embodiment

[0051] In the specific implementation process, compared with the above embodiments, this embodiment mainly focuses on the assembly process between different devices, and its specific execution end is the splicing processing end;

[0052] In the splicing processing section, based on the adapter model marked by the 3D single-unit model, the splicing model and splicing points are confirmed from the model information associated with the adapter model. Different devices are then spliced ​​together to generate a splicing assembly. The specific splicing process includes:

[0053] Based on the splicing model associated with the adaptation model, the three-dimensional single-unit model associated with the splicing model is identified. The components of the identified three-dimensional single-unit model are recorded as secondary components to be assembled, and the components of the three-dimensional single-unit model associated with the adaptation model are recorded as primary components to be assembled.

[0054] Based on the splicing points marked in the adaptation model and splicing model, the main component to be assembled and the secondary component to be assembled are spliced ​​together so that the marked splicing points coincide, thus generating a spliced ​​component.

[0055] The comprehensive verification end verifies the generated splicing components. It uses the splicing model body associated with the adaptation model and the splicing model as the standard model body, locks the benchmark features from the standard model body, confirms the component features associated with the splicing components, and compares and verifies the benchmark features with the component features to identify whether the splicing components are assembled correctly.

[0056] Generate a standard model body from the fitted model and the spliced ​​model according to the marked splicing points. In the standard model body, take the midpoint of the fitted model as the starting point and the midpoint of the spliced ​​model as the ending point, and generate the point vector associated with the starting point to the ending point. This vector is denoted as the reference feature associated with the standard model body.

[0057] The two sets of three-dimensional single-unit models associated with the splicing component are confirmed. The built-in midpoint of the three-dimensional single-unit model associated with the fitting model is taken as the starting point, and the built-in midpoint of the three-dimensional single-unit model associated with the splicing model is taken as the ending point. The point vector associated from the starting point to the ending point is generated and recorded as the component feature associated with the splicing component.

[0058] The baseline feature and component feature are compared to ensure that the starting points associated with the two sets of features coincide. The included angle A between the two sets of features is then confirmed. If no included angle exists, the included angle is set to 0. The vector lengths associated with the two sets of features are then confirmed. The vector length of the baseline feature is denoted as L1, and the vector length of the component feature is denoted as L2.

[0059] If L1=L2 and the included angle is 0, no processing is required, which means that the splicing component assembly meets the standard, and an assembly compliance signal is generated for display.

[0060] If L1≠L2, then the following methods are used to confirm the two sets of proportional parameters B1 and B2: |L1-L2|÷L1=B1 and A÷360=B2. The following method is used to confirm the evaluation value PD: PD=B1×C1+B2×C2. C1 and C2 are preset fixed coefficient factors, and their specific values ​​are determined by the operator based on experience. If PD≤Y1, then an assembly correction signal is generated and displayed. The operator then uses this signal to make real-time corrections to the splicing components. Y1 is a preset value, and its specific value is determined by the operator based on experience. If PD>Y1, then an assembly non-compliance signal is generated and displayed in a timely manner.

[0061] If L1≠L2, another specific method for identifying whether the splicing components are assembled correctly is as follows:

[0062] Based on the generated reference features, a set of spherical ranges are generated with the end point of the reference features as the center of the sphere and the preset radius R as the radius of the sphere. R is a preset value, and its specific value is determined by the operator based on experience.

[0063] Align the starting points of the component features with the reference features, and identify whether the ending points of the component features are within the spherical range. If so, generate an assembly correction signal and display it; otherwise, generate an assembly non-compliance signal and display it promptly.

[0064] Some of the data in the above formulas are numerical calculations with dimensions removed, and the contents not described in detail in this specification are all prior art known to those skilled in the art.

[0065] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A full-process monitoring system based on visual offline recognition, characterized in that, include: The 3D scanning processing unit performs 3D scanning on the devices present on the production line and generates a three-dimensional single-unit model associated with the corresponding device. The model database contains a large number of 3D standard models, all of which are preset models. The model verification and recognition end performs 3D entity verification on the generated 3D single model, confirms the built-in midpoint of the 3D single model and the midpoint of different planes, and then confirms the verification features associated with the 3D single model based on the built-in midpoint and the midpoint of the planes, and then selects and marks the matching model from the model database based on the verification features. On the splicing processing end, based on the adapter model marked by the 3D single-unit model, the splicing model and splicing point are confirmed from the model information associated with the adapter model, and different devices are spliced ​​to generate splicing components. The integrated verification end performs verification processing on the generated splicing components. It takes the splicing model body associated with the adaptation model and the splicing model as the standard model body, locks the benchmark features from the standard model body, confirms the component features associated with the splicing components, compares and verifies the benchmark features and component features, and identifies whether the splicing components are assembled to standard.

2. The full-process monitoring system based on visual offline recognition according to claim 1, characterized in that, The specific method by which the model verification and recognition terminal confirms the verification features of the three-dimensional single-unit model is as follows: Combine the generated 3D single-unit model with the 3D coordinate system, confirm the 3D coordinates associated with different points on the outer surface of the 3D single-unit model, perform mean processing on several confirmed sets of 3D coordinates, confirm the mean coordinates associated with several 3D coordinates, confirm the spatial position associated with the mean coordinates in the 3D coordinate system, and mark the associated spatial position in the 3D single-unit model, which is recorded as the built-in midpoint of the 3D single-unit model. Next, the midpoints of the planes associated with different outer surfaces of the three-dimensional single model are determined, so that the single set of outer surfaces is combined with the two-dimensional coordinate system. The two-dimensional coordinates associated with different points on the outer surface are confirmed. Then, the average value of several sets of two-dimensional coordinates is processed to confirm the average coordinates. The point where the average coordinates are located is marked on the outer surface and recorded as the midpoint of the plane associated with the outer surface. Starting from the built-in midpoint and ending at the plane midpoint, construct the feature vectors associated with the starting and ending points. Confirm the feature vectors associated with several outer surfaces in sequence, and record the confirmed feature vectors as the verification features associated with the three-dimensional single-unit model.

3. The full-process monitoring system based on visual offline recognition according to claim 1, characterized in that, The specific method for selecting the appropriate model in the model verification and recognition terminal is as follows: Extract the 3D standard model from the model database, confirm the marked model midpoints and surface midpoints within the 3D standard model, and confirm the standard features associated with the 3D standard model using the same confirmation method as the verification features. The verification features are compared with the standard features to ensure that the starting points associated with the verification features and the standard features coincide. Then, it is determined whether the verification features and the standard features are completely consistent. If they are, the 3D standard model associated with the standard features is recorded as the adapted model of the 3D single-unit model. If not, no marking is made, and the process of confirming the adapted model for other 3D standard models continues.

4. The full-process monitoring system based on visual offline recognition according to claim 1, characterized in that, The splicing processing terminal performs splicing processing on different devices in the following specific way: Based on the splicing model associated with the adaptation model, the three-dimensional single-unit model associated with the splicing model is identified. The components of the identified three-dimensional single-unit model are recorded as secondary components to be assembled, and the components of the three-dimensional single-unit model associated with the adaptation model are recorded as primary components to be assembled. Based on the splicing points marked in the adaptation model and splicing model, the main component to be assembled and the secondary component to be assembled are spliced ​​together so that the marked splicing points coincide, thus generating a spliced ​​component.

5. The full-process monitoring system based on visual offline recognition according to claim 1, characterized in that, The comprehensive verification terminal identifies whether the splicing components are assembled to meet the standards in the following specific way: Generate a standard model body from the fitted model and the spliced ​​model according to the marked splicing points. In the standard model body, take the midpoint of the fitted model as the starting point and the midpoint of the spliced ​​model as the ending point, and generate the point vector associated with the starting point to the ending point. This vector is denoted as the reference feature associated with the standard model body. The two sets of three-dimensional single-unit models associated with the splicing component are confirmed. The built-in midpoint of the three-dimensional single-unit model associated with the fitting model is taken as the starting point, and the built-in midpoint of the three-dimensional single-unit model associated with the splicing model is taken as the ending point. The point vector associated from the starting point to the ending point is generated and recorded as the component feature associated with the splicing component. The baseline feature and component feature are compared to ensure that the starting points associated with the two sets of features coincide. The included angle A between the two sets of features is then confirmed. The vector lengths associated with the two sets of features are then confirmed. The vector length of the baseline feature is denoted as L1, and the vector length of the component feature is denoted as L2. If L1=L2 and the included angle is 0, no processing is required, indicating that the splicing components have met the assembly standards, and an assembly compliance signal will be generated for display.

6. The full-process monitoring system based on visual offline recognition according to claim 5, characterized in that, If there is no included angle, the included angle is 0.

7. A full-process monitoring system based on offline visual recognition according to claim 5, characterized in that, If L1≠L2, then use: |L1-L2|÷L1=B1 and A÷360=B2 to confirm the two sets of proportional parameters B1 and B2, and use: PD=B1×C1+B2×C2 to confirm the evaluation value PD, where C1 and C2 are preset fixed coefficient factors. If PD≤Y1, then generate an assembly correction signal for display, where Y1 is a preset value. If PD>Y1, then generate an assembly non-compliance signal and display it in time.

8. A full-process monitoring system based on visual offline recognition according to claim 5, characterized in that, If L1≠L2, another specific method for identifying whether the splicing components are assembled correctly is also included: Based on the generated reference features, a set of spherical ranges are generated with the end point of the reference features as the center of the sphere and the preset radius R as the radius of the sphere, where R is a preset value; Align the starting points of the component features with the reference features, and identify whether the ending points of the component features are within the spherical range. If so, generate an assembly correction signal and display it; otherwise, generate an assembly non-compliance signal and display it promptly.

9. A full-process monitoring system based on visual offline recognition according to claim 1, characterized in that, The bottom surface of the three-dimensional single-unit model is the upper base surface of the pipeline.

Images