Composite material profile quality joint detection method based on image data

Abstract

This invention provides a method for joint quality inspection of composite profiles based on image data, belonging to the field of image processing technology. Specifically, it includes: determining a control strategy for the production parameters of the composite profile based on the distribution data of defect identification focus cameras and the defect identification results of different defect identification focus cameras; adjusting the production parameters of the composite profile during changes based on the control strategy; determining the types of defects to be detected by the defect identification focus cameras based on the changes in defect identification data of different defect identification focus cameras during the adjustment process; and determining a joint quality inspection method for composite profiles based on image data according to the defect types detected by different defect identification focus cameras and the control strategy for the production parameters of the composite profile, thereby improving the reliability of the quality inspection process.

Description

Technical Field

[0001] This invention belongs to the field of image processing technology, and in particular relates to a method for joint detection of composite material profile quality based on image data. Background Technology

[0002] Fiberglass pultruded profiles for doors and windows have complex cross-sections (such as multi-cavity, with sealing grooves, and with decorative curved surfaces), and defects can appear on any surface of the profile—the top surface, bottom surface, both sides, and even the inner wall of the groove. Existing inspection methods usually only photograph the top surface (the largest surface) of the profile, ignoring the sides and bottom surface, resulting in a large number of blind spots.

[0003] To address the aforementioned technical problems, existing solutions employ multiple cameras to simultaneously capture images of multiple surfaces, enabling synchronized identification and processing of defects at multiple locations. However, these solutions suffer from the following drawbacks: When performing defect identification processing, the same profile may have quality defects in multiple cameras simultaneously. This requires a high degree of reliability in combining and reconstructing different cameras under multiple quality defect types to generate a complete profile. Therefore, it is an urgent technical problem to solve how to determine the requirements for combined reconstruction analysis and processing under multiple quality defect types based on the quality defect identification results of different cameras, and how to determine the joint quality detection method of different cameras based on the requirements for combined reconstruction analysis and processing under multiple quality defect types. This will improve the observation reliability of the combined reconstruction processing of different cameras under multiple quality defect types.

[0004] Specifically, this application provides a method for joint quality inspection of composite material profiles based on image data. Summary of the Invention

[0005] To achieve the objectives of this invention, the following technical solution is adopted: Specifically, this application provides a method for joint quality inspection of composite material profiles based on image data, which includes: S1 uses the quality inspection data of composite profiles to determine the defect identification results of cameras with different phases, determines the degree of dispersion of identification processing under different quality defect types based on the defect identification results, and determines the defect identification focus camera in the camera based on the degree of dispersion of identification processing of cameras with different phases. S2 determines the control strategy for the production parameters of composite profiles based on the distribution data of the defect identification focus camera and the defect identification results of different defect identification focus cameras. Based on the control strategy, it performs adjustment processing during the change of the production parameters of the composite profiles. Based on the change of the defect identification data of different defect identification focus cameras during the adjustment processing, it determines the type of defect that the defect identification focus camera focuses on. S3 determines a joint quality inspection method for composite material profiles based on image data, according to the defect type data of different defect identification cameras and the control strategy of the production parameters of the composite material profiles.

[0006] The beneficial effects of this invention are as follows: Based on the defect identification results of different cameras, the degree of dispersion of associated defect types in different cameras is determined. That is, the more associated defect types in different cameras, the higher the degree of dispersion. The higher the degree of dispersion, the greater the need for combined reconstruction processing under multiple quality defect types. Therefore, based on the degree of dispersion of associated defect types in different cameras, the need for combined reconstruction of images from cameras with different phases on the same load material profile is determined. Using the need for combined reconstruction, the defect identification focus cameras that need to be focused on are specifically identified, which also lays the foundation for further determination of the quality detection method for defect identification focus cameras.

[0007] Based on the defect type data from different defect identification cameras, the requirements for combined reconstruction analysis and processing of images from different defect identification cameras under different defect types are determined. Specifically, the larger the defect type of different defect identification cameras, the higher the requirements for combined reconstruction analysis and processing of images under that defect type. Simultaneously, the requirements for the reliability of quality defect type identification are also higher. Therefore, by considering the requirements for the reliability of quality defect type identification and the control strategy for the production parameters of composite profiles, a joint quality inspection method for composite profiles based on image data is determined. This further improves the reliability of quality defect type identification and processing from cameras with different phases while reducing data processing and control difficulties. Furthermore, based on the identified quality defect types, a foundation is laid for further improving the reliability of combined reconstruction processing under multiple defect types of the same composite profile.

[0008] Furthermore, the quality inspection data includes the quality defect detection results of cameras at different locations in the production line of the composite material profile.

[0009] Furthermore, the defect identification results of the camera include the identification results of the camera under different types of quality defects.

[0010] Furthermore, the method for determining the camera of interest for defect identification in the camera is as follows: S11 determines the number of times the camera identifies different types of quality defects based on the defect identification results of the camera; S12 determines the type of quality defect in the camera whose recognition count is above a preset recognition count threshold based on the recognition count, and uses it as an associated defect type; S13 determines the camera of interest for defect identification based on the degree of dispersion of associated defect types in different cameras.

[0011] Furthermore, the method for determining the quality of composite profiles based on image data is as follows: S41 determines the number of defect types in the defect identification camera based on the defect type data in different defect identification cameras. S42 uses the number of the types of defects of interest to determine the cameras with recognition needs among the defect recognition cameras of interest; S43 determines the image data-based joint inspection method for composite material profile quality based on the control strategy of the identification requirement camera and the production parameters of the composite material profile.

[0012] Other features and advantages will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention are realized and obtained through the structures particularly pointed out in the description and the drawings.

[0013] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0014] The above and other features and advantages of the present invention will become more apparent from a detailed description of exemplary embodiments thereof with reference to the accompanying drawings.

[0015] Figure 1 This is a flowchart of a joint quality inspection method for composite material profiles based on image data; Figure 2 This is a flowchart of a method for identifying defects in a camera and determining the camera to be inspected; Figure 3 This is a flowchart illustrating the method for determining the control strategy for the production parameters of composite material profiles. Detailed Implementation

[0016] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the embodiments set forth herein; rather, they are provided so that the invention will be thorough and complete, and the concept of the exemplary embodiments will be fully conveyed to those skilled in the art. The same reference numerals in the drawings denote the same or similar structures, and therefore their detailed description will be omitted.

[0017] The terms “a,” “one,” “the,” and “the” are used to indicate the existence of one or more elements / components / etc.; the terms “including” and “having” are used to indicate an open-ended meaning of inclusion and that there may be other elements / components / etc. in addition to the listed elements / components / etc.

[0018] Example 1 To solve the above problems, according to one aspect of the present invention, such as Figure 1 As shown, a method for joint quality inspection of composite material profiles based on image data is provided, specifically including: S1 uses the quality inspection data of composite profiles to determine the defect identification results of cameras with different phases, determines the degree of dispersion of identification processing under different quality defect types based on the defect identification results, and determines the defect identification focus camera in the camera based on the degree of dispersion of identification processing of cameras with different phases. Furthermore, the quality inspection data includes the quality defect detection results of cameras at different locations in the production line of the composite material profile.

[0019] Furthermore, the defect identification results of the camera include the identification results of the camera under different types of quality defects.

[0020] Specifically, such as Figure 2 As shown, the method for determining the camera of interest for defect identification in the camera is as follows: In this embodiment, the degree of dispersion of associated defect types in different cameras is determined based on the defect identification results of different cameras. That is, the more associated defect types in different cameras, the higher the degree of dispersion. Based on the degree of dispersion of associated defect types in different cameras, the requirement for combining and reconstructing images from cameras with different phases onto the same load material profile is determined. The requirement for combined reconstruction is used to determine the defect identification focus cameras that need to be focused on. That is, the higher the requirement for combined reconstruction, the more relaxed the screening conditions for defect identification focus cameras are, which also lays the foundation for further determining the quality detection method for defect identification focus cameras.

[0021] S11 determines the number of times the camera identifies different types of quality defects based on the defect identification results of the camera; The defect identification results of the cameras refer to the identification and judgment results of various quality defect types output by the image recognition algorithm after the cameras of different phases in the production line acquire images of composite material profiles. The quality defect types refer to various quality defects that may occur in the composite material profiles during the pultrusion molding process, including surface bubbles, internal cracks, dimensional deviations, surface scratches, uneven color, and exposed fibers. The identification count refers to the cumulative number of profiles with a certain quality defect type identified by a certain camera within a statistical period, used to quantify the camera's identification activity for that defect type.

[0022] Assuming that multiple phase cameras are arranged in the production line, each camera acquires images and identifies defects from different angles of each composite profile it passes through. For each type of quality defect, the number of profiles with that type of defect identified by the camera within the statistical period is counted. The number of times each camera identifies different quality defect types is obtained, which is then compared with a preset identification number threshold to determine which defect types belong to the associated defect types of that camera.

[0023] This step involves statistically analyzing the number of times each camera identifies different quality defect types. The significance of this step lies in the fact that the number of identifications is a core indicator for measuring the activity level of a camera in identifying a certain type of defect. Only by accurately grasping the identification frequency of each camera for various types of defects can we objectively determine which defect types belong to the associated defect types of that camera in subsequent steps. This provides reliable statistical data support for calculating the dispersion of associated defect types and determining the cameras of interest for defect identification.

[0024] S12 determines the type of quality defect in the camera whose recognition count is above a preset recognition count threshold based on the recognition count, and uses it as an associated defect type; The preset recognition frequency threshold refers to the critical number of recognitions required to determine whether a camera is highly active in recognizing a certain type of quality defect and whether that defect type should be included in the camera's associated defect type range. The associated defect type refers to those quality defect types for which a camera's recognition frequency reaches or exceeds the preset recognition frequency threshold, i.e., the set of defect types that the camera recognizes frequently and has high recognition activity within the statistical period.

[0025] Assuming that the number of times each camera identifies different quality defect types is compared with a preset identification number threshold, those quality defect types with identification numbers greater than or equal to the preset identification number threshold are extracted and classified into the set of associated defect types for that camera. This set is used to calculate the number of associated defect types for that camera and the degree of dispersion of associated defect types between different cameras.

[0026] This step determines the associated defect types for each camera by filtering based on a preset threshold for the number of recognitions. The significance of this step is that not all identified defect types need to be included in the associated defect type range. Only defect types that have reached a sufficient number of recognitions indicate that the camera has a continuous and stable ability to recognize such defects. Threshold filtering can eliminate the interference of occasional recognition noise, ensure the representativeness and reliability of the associated defect type set, and provide accurate basic data for subsequent dispersion assessment.

[0027] S13 determines the camera of interest for defect identification based on the degree of dispersion of associated defect types in different cameras.

[0028] The dispersion of the associated defect types refers to the degree of difference between the sets of associated defect types from different cameras. Specifically, it is quantified by the differences in the quantity and types of associated defect types between different cameras. A higher dispersion indicates greater differences and lower overlap in the defect types that different cameras focus on, thus increasing the need for multi-camera image combination and reconstruction. The defect identification focus camera refers to the camera that needs to be prioritized and used in the multi-camera quality joint detection system. Its selection criteria are dynamically adjusted based on the dispersion of the associated defect types; a higher dispersion results in more lenient selection criteria.

[0029] Assuming that the number of associated defect types is calculated for each camera, and then the sets of associated defect types of all cameras are compared pairwise, the number of overlaps and differences of associated defect types between different cameras are counted. The degree of dispersion is measured by the proportion of the number of differences to the total number of defect types. Camera groups with high degree of dispersion need to undergo multi-camera image combination and reconstruction processing to obtain the complete quality status of the profile. Therefore, more cameras are included in the scope of cameras of interest for defect identification.

[0030] This step determines the camera to focus on for defect identification by the degree of dispersion of associated defect types. The significance of this step is that different phase cameras have inherent differences in their ability to identify different quality defect types due to differences in installation location, lighting conditions, and shooting angle. The degree of dispersion of associated defect types quantifies the degree of this difference. The higher the degree of dispersion, the less sufficient the identification results of a single camera are to reflect the complete quality status of the profile. More cameras need to be included in the scope of focus to initiate multi-camera combination reconstruction processing to ensure the comprehensiveness and reliability of quality defect identification.

[0031] Furthermore, based on the degree of dispersion of associated defect types in different cameras, the defect identification focus camera in the cameras is determined, specifically including: It should be noted that, in case 1: S131, it is determined whether there are cameras with a number of associated defect types greater than the preset defect type number threshold. If so, in order to realize the combined reconstruction processing of images from cameras with different phases under multiple associated defect types, and thus determine the true state of the composite material profile, all cameras are determined to be defect identification focus cameras. If not, proceed to step S132. The preset defect type number threshold refers to the critical number of associated defect types used to determine whether the number of associated defect types of a certain camera is already too large and whether the overall defect identification complexity of the production line where that camera is located is already high. The phrase "all cameras are defect identification focus cameras" means that when there are cameras with an excessive number of associated defect types, due to the high demand for multi-defect type combination reconstruction, all phase cameras in the production line need to be included in the focus scope to ensure that multi-camera joint detection can cover the identification needs of all associated defect types.

[0032] If the number of associated defect types for a certain camera exceeds the preset threshold for the number of defect types, it means that the shooting position of that camera can identify a large number of defect types, the overall defect identification complexity of the production line is high, and the need for multi-camera image combination reconstruction is more urgent. In this case, all cameras in the production line are identified as defect identification focus cameras, so as to make full use of the identification capabilities of all cameras for multi-camera joint detection.

[0033] In this situation, when there are cameras with an excessive number of associated defect types, all cameras are identified as cameras of interest for defect identification. The significance of this is that when the number of associated defect types of a certain camera is already large, it indicates that the distribution of quality defect types of composite profiles in the production line is relatively complex. The identification results of a single camera are far from sufficient to cover all defect types. It is necessary to include all cameras in the scope of focus and start multi-camera image combination reconstruction processing involving all cameras in order to accurately restore the true quality status of composite profiles and avoid missed defects due to insufficient camera coverage.

[0034] Additionally, it should be noted that in case 2: S132, cameras with the type of defect of interest are used as screening cameras. It is determined whether the number of screening cameras is less than the preset threshold for the number of screening cameras. If so, the need for combined reconstruction processing of cameras with different phases under different fault types is low. Therefore, it is determined that the screening camera belongs to the defect identification focus camera. If not, proceed to step S133. The selected cameras refer to cameras with at least one associated defect type or a type of defect of interest. The preset threshold for the number of selected cameras is a critical number of cameras used to determine whether the number of cameras with the type of defect of interest is already low and the demand for multi-camera combination reconstruction processing is low. The selection of cameras as defect identification focus cameras means that when the number of selected cameras is small, due to the low demand for multi-camera combination reconstruction, all selected cameras are directly included in the scope of defect identification focus cameras without further weighted selection.

[0035] If the number of screening cameras for the type of defect of interest is less than the preset threshold for the number of screening cameras, it means that the number of cameras that can identify quality defects in the production line is relatively small. At this time, the production quality is relatively high, and the demand for combined reconstruction processing between multiple cameras is relatively low. In this case, all screening cameras are directly identified as cameras of interest for defect identification to simplify the subsequent processing flow and ensure basic multi-camera joint detection capabilities.

[0036] In this case, when the number of screened cameras is small, they are directly identified as cameras of interest for defect identification. The significance of this is that when the number of cameras with a large number of quality defects is limited, the available data sources for multi-camera image combination reconstruction are limited. At this time, the production quality is high, and the demand and complexity of combination reconstruction are low. In this case, there is no need to perform complex weight screening. All screened cameras are directly included in the scope of key attention, which simplifies the judgment process and ensures the full utilization of limited camera resources in joint quality inspection.

[0037] It also includes the following: Case 3: S133 determines the basic weight value of the screening camera based on the number of associated defect types of the screening camera, and determines whether the sum of the basic weight values ​​of different screening cameras is greater than a preset weight threshold. If so, it determines that all cameras belong to the defect identification focus cameras. If not, it takes the camera with a quality defect type that has a number of identifications above a preset number threshold (less than the preset number of identifications threshold) as the defect identification focus camera.

[0038] The basic weight value of the selected cameras refers to the weight value calculated based on the number of associated defect types of the selected cameras. The more associated defect types, the higher the basic weight value, which is used to quantify the importance of each selected camera in the overall defect identification system. The preset weight threshold refers to the critical weight and value used to determine whether the sum of the importance of all selected cameras is high enough to require all cameras to be included in the scope of key attention. The preset number of recognitions threshold (less than the preset number of recognitions threshold) refers to the critical number of recognitions threshold used to include some cameras in the scope of defect identification at a lower recognition number of recognitions when the sum of the importance of all selected cameras does not reach the preset weight threshold, thereby ensuring that a certain number of cameras are included in the scope of key attention by lowering the threshold.

[0039] The formula for calculating the basic weight value of the camera selection is: Basic weight value = Number of associated defect types ÷ Total number of all quality defect types.

[0040] If the sum of the basic weight values ​​of each selected camera does not exceed the preset weight threshold, it means that the sum of the importance of each selected camera is still low. In this case, a preset number threshold (less than the preset recognition number threshold) is used as the threshold, and cameras corresponding to the quality defect type whose recognition number reaches the threshold are included in the scope of defect recognition focus cameras. If the sum of the basic weight values ​​of each selected camera exceeds the preset weight threshold, it means that the sum of the importance of each selected camera is already high, and the need for multi-camera combination reconstruction is high. All cameras are determined to be defect recognition focus cameras.

[0041] This step uses a comprehensive judgment based on the sum of basic weight values ​​to finally screen cameras for defect identification. Its significance lies in quantifying the overall importance of each screened camera by using basic weight values ​​when there are a large number of screened cameras but no specific camera with an excessive number of associated defect types. If the sum of importance is already high, the full-camera focus mode needs to be activated to cover diverse defect identification needs. If the sum of importance is still low, cameras are added with a lower preset threshold, ensuring that the number of cameras for defect identification meets the basic requirements of combination and reconstruction while avoiding the over-inclusion of unimportant cameras that would increase the data processing burden.

[0042] There are 6 cameras (C1 to C6) in the production line. There are a total of 8 types of quality defects during the statistical period (surface bubbles, internal cracks, dimensional deviations, surface scratches, uneven color, exposed fibers, damaged ends, and uneven wall thickness).

[0043] In S11, the number of times each camera from C1 to C6 identifies different quality defect types within the statistical period is counted. The number of times each camera identifies defects is compared with the preset identification threshold of 100 times to determine the associated defect types of each camera: C1 associated defect types: surface bubbles, size deviation, a total of 2 types; C2 associated defect types: internal cracks, surface scratches, a total of 2 types.

[0044] Enter S13. S131 determines that the maximum number of associated defect types for each camera is 3 types of C4 and C6, which is not greater than the preset defect type threshold of 5 types. Enter S132.

[0045] In S132, the screening cameras with associated defect types are C1, C2, C3, C4, C5, and C6, a total of 6 cameras, which is not less than the preset screening camera quantity threshold of 3 cameras, and then proceed to S133.

[0046] In S133, the basic weight value of each screened camera (number of associated defect types ÷ 8) is calculated: C1: 2 ÷ 8 = 0.25; C2: 2 ÷ 8 = 0.25; C3: 2 ÷ 8 = 0.25; C4: 3 ÷ 8 = 0.375; C5: 2 ÷ 8 = 0.25; C6: 3 ÷ 8 = 0.375. The sum of the basic weight values ​​= 0.25 + 0.25 + 0.25 + 0.375 + 0.25 + 0.375 = 1.75, which is greater than the preset weight threshold of 0.60. Therefore, all cameras (C1 to C6) are determined to be defect identification cameras of interest, and the process proceeds to S2.

[0047] S2 determines the control strategy for the production parameters of composite profiles based on the distribution data of the defect identification focus camera and the defect identification results of different defect identification focus cameras. Based on the control strategy, it performs adjustment processing during the change of the production parameters of the composite profiles. Based on the change of the defect identification data of different defect identification focus cameras during the adjustment processing, it determines the type of defect that the defect identification focus camera focuses on. Specifically, such as Figure 3 As shown, the method for determining the control strategy of the production parameters of the composite material profile is as follows: In this embodiment, based on the distribution data of defect identification cameras and the number of associated defect types in the defect identification cameras, the number of defect identification cameras and the number of defect types requiring special attention are determined. Using these numbers, the control requirements for the production parameters of the production equipment are determined. That is, the more defect identification cameras and defect types requiring special attention, the more necessary it is to perform multi-phase camera image combination reconstruction processing under multiple quality defect types. The control requirements for the production parameters of the production equipment are determined based on these requirements, i.e., production control processing is performed under multiple production parameters, thereby identifying more quality defect types and laying the foundation for further improving the reliability and sufficiency of multi-phase camera image combination reconstruction processing.

[0048] S21 Based on the distribution data of the defect identification focus camera, determine the proportion of the defect identification focus camera among all cameras, and use the proportion of the defect identification focus camera among all cameras as the focus camera proportion. The distribution data of the defect identification focus cameras refers to the identification information of which cameras are identified as defect identification focus cameras among each phase camera and their distribution location information. The focus camera ratio refers to the ratio of the number of defect identification focus cameras to the total number of cameras in the production line. It is used to measure the coverage of the cameras that currently need to be focused on in the overall camera system. The higher the focus camera ratio, the wider the range of cameras that need to participate in multi-camera joint inspection. The control strategy of production parameters should be adjusted accordingly to adapt to a wider range of quality inspection needs.

[0049] Assuming there are 6 cameras in the production line, and S13 determines that all 6 cameras are cameras of interest for defect identification, then the proportion of cameras of interest = 6 ÷ 6 = 1.00. This high proportion indicates that the scope of multi-camera joint detection is relatively wide, and the production parameter control strategy needs to consider the need for more comprehensive defect type coverage.

[0050] This step calculates the proportion of cameras of interest, which is significant because the proportion of cameras of interest is a core indicator for measuring the size of the joint detection range of multiple cameras. A higher proportion of cameras of interest means that the recognition results of more cameras need to be included in the scope of joint quality detection. The production parameter control strategy needs to be adjusted to a more lenient mode to obtain more comprehensive defect type data, thus providing a sufficient data foundation for subsequent multi-camera image combination and reconstruction processing.

[0051] It is understandable that if the proportion of the focus camera is greater than the preset focus camera proportion threshold in the above steps, then the control strategy for the production parameters of the composite material profile is determined to be a relaxed control strategy. That is, only when the production parameters of the production equipment are within the qualified range and the rate of change of the production parameters from the optimal production parameters is greater than the preset rate of change threshold, is it necessary to adjust the production equipment to the optimal production parameters. In this way, the defect types under various operating parameters can be obtained, and the cameras of different phases can be reliably combined and reconstructed.

[0052] The preset camera coverage threshold refers to the critical percentage value used to determine whether the coverage of the defect identification cameras is already high enough, requiring a more lenient production parameter control strategy. The lenient control strategy refers to the production parameter control mode used when the camera coverage is high. In this mode, parameter adjustments are only made when the production parameters are within acceptable limits and the rate of change between the production parameters and the optimal production parameters exceeds the preset rate of change threshold, avoiding excessively frequent adjustments that could lead to incomplete defect type data acquisition. The optimal production parameter refers to the production parameter that minimizes the occurrence of production quality defects, i.e., the production parameter setting value that minimizes the occurrence rate of quality defects in composite material profiles, obtained through statistical analysis of historical production data. The preset rate of change threshold refers to the critical rate of change value used to determine whether the difference between the production parameters and the optimal production parameters is too large and requires adjustment.

[0053] If the proportion of cameras under observation exceeds the preset threshold, it indicates that the multi-camera joint detection range is relatively wide. A relaxed control strategy is adopted, under the premise that the production parameters are qualified, adjustments are only made when the production parameters deviate from the optimal production parameters by more than the preset rate of change threshold. This retains a certain range of parameter changes within the qualified range to obtain defect type data under various operating parameters, providing more sufficient defect type coverage for the combined reconstruction processing of multi-camera images.

[0054] This judgment adopts a relaxed control strategy when the proportion of cameras of interest is high. The significance of this strategy is that when the coverage of cameras of interest in defect identification is high, multi-camera joint detection can cover a wider shooting range and defect types. At this time, if the production parameters are controlled too strictly and kept near the optimal value, it will be difficult to obtain diverse defect type data under different parameter settings. This is not conducive to the multi-camera combined reconstruction processing fully covering various defect types. Adopting a relaxed control strategy can obtain more comprehensive defect type data while ensuring product quality compliance, thereby improving the reliability of combined reconstruction processing.

[0055] It is also understood that if the proportion of the camera in focus is not greater than the preset threshold for the proportion of the camera in focus, then proceed to step S22.

[0056] The transition step S22 indicates that when the proportion of cameras under observation does not exceed the preset threshold for the proportion of cameras under observation, it means that the coverage of multi-camera joint detection is relatively limited. It is necessary to further analyze the proportion of associated defect types of each defect identification camera to more precisely determine the control strategy of production parameters, so as to ensure the reliability of defect identification while avoiding an overly lenient control strategy that leads to a decline in product quality.

[0057] If the proportion of cameras under focus does not exceed the preset threshold for the proportion of cameras under focus, it indicates that the current scope of cameras under focus is relatively narrow. Further analysis is needed based on the proportion of associated defect types of each camera under focus to determine whether a more stringent production parameter control strategy is required. Proceed to S22 for further judgment.

[0058] S22 determines the number of associated defect types in the defect identification camera based on the defect identification results of different defect identification focus cameras, and determines the proportion of associated defect types in the defect identification focus camera based on the proportion of the number of associated defect types in all defect types identified by the defect identification focus camera. The "all defect types" identified by the defect identification focus camera refer to all quality defect types actually identified by a particular defect identification focus camera within the statistical period, regardless of whether the number of identifications reaches the preset identification threshold. The "associative defect type ratio" refers to the ratio of the number of associated defect types of a particular defect identification focus camera to the total number of defect types identified by that camera. It measures the coverage of qualified associated defect types within the camera's total identified defect types. A lower associated defect type ratio indicates that a significant number of defect types have not yet met the identification threshold, and the camera's defect identification stability is relatively low.

[0059] Suppose that a certain defect identification camera identifies a total of 6 defect types, of which 2 are associated defect types. Then the ratio of associated defect types for this camera is approximately 0.33 (2 ÷ 6). This low ratio indicates that there are still many defect types that have not been identified enough times, and this camera is likely to be a high-risk camera.

[0060] This step calculates the proportion of associated defect types for each camera of interest in defect identification. The significance of this step is that the proportion of associated defect types quantifies the coverage relationship between the qualified defect types in each camera of interest and all the identified defect types. The lower the proportion, the more unstable the defect identification capability of the camera is, and the more defect types have not yet reached the standard for stable identification. After such cameras are marked as risk cameras, compensation for their defect identification stability needs to be considered in the production parameter control strategy. The reliability of their defect identification can be improved by adjusting the control strategy.

[0061] Furthermore, the above steps include the following: Based on the proportion of associated defect types of different defect identification focus cameras, it is determined whether there are defect identification focus cameras whose proportion of associated defect types is less than a preset defect type proportion threshold. If so, proceed to the next step; otherwise, determine that the control strategy for the production parameters of the composite material profile is a strict control strategy. That is, as long as the production parameters of the production equipment are within the qualified range, and there is no production equipment that needs to be adjusted due to the rate of change of the production parameters from the optimal production parameters being greater than a preset rate of change threshold within the most recent preset time period, then the production equipment needs to be adjusted only when the rate of change of the production parameters from the optimal production parameters is greater than the preset rate of change threshold. In other words, the production parameters are adjusted to the optimal production parameters, so that the defect types under various operating parameters can be obtained, and the cameras of different phases can be reliably combined and reconstructed.

[0062] The preset defect type ratio threshold refers to the critical ratio value used to determine whether the ratio of associated defect types of a certain defect identification camera is already low and whether the camera belongs to the risk identification camera category. The strict control strategy refers to the production parameter control mode adopted when there are no defect identification cameras with an associated defect type ratio lower than the preset defect type ratio threshold. In this mode, if there are no devices that have been processed due to excessive parameter change rate within the most recent preset time period, adjustments are only made when the change rate between the production parameters and the optimal production parameters exceeds the preset change rate threshold. The strictness of the control is between the lenient control strategy and other control strategies, and it appropriately obtains diverse defect type data while ensuring product quality.

[0063] Cameras with a defect type ratio less than a preset defect type ratio threshold are identified as risk cameras. It is determined whether the number of risk cameras exceeds a preset risk camera number threshold. If so, the control strategy for the production parameters of the composite material profile is determined to be a lenient control strategy. That is, only when the production parameters of the production equipment are within the qualified range and the rate of change between the production parameters and the optimal production parameters is greater than a preset rate of change threshold, is it necessary to adjust the production equipment to the optimal production parameters. This allows for the acquisition of defect types under various operating parameters and the reliable combination and reconstruction of cameras with different phases. If not, proceed to step S23.

[0064] The risk-identifying camera refers to a defect identification camera whose proportion of associated defect types is less than a preset defect type proportion threshold. This means that the camera has insufficient coverage of eligible defect types and relatively low defect identification stability. The preset risk camera quantity threshold is a critical risk camera quantity value used to determine whether the number of risk-identifying cameras is already too large and a lenient production parameter control strategy is needed.

[0065] If there are defect identification cameras whose proportion of associated defect types is less than a preset defect type proportion threshold, these cameras are identified as risk cameras. If the number of risk cameras exceeds a preset risk camera number threshold, it indicates that many cameras have insufficient defect identification stability. In this case, a relaxed control strategy is needed to obtain more comprehensive defect type data by allowing a wider range of production parameter variations, thereby improving the defect identification stability of each risk camera. If the number of risk cameras does not exceed the limit, proceed to S23 for comprehensive judgment.

[0066] This step determines the production parameter control strategy by identifying the number of risk cameras. Its significance lies in the fact that when many cameras have insufficient defect recognition stability, if the production parameter control is too strict, it will be difficult to obtain enough diverse defect type data to improve the recognition stability of these cameras. Adopting a relaxed control strategy can help each risk camera accumulate more types of defect recognition data by appropriately changing the parameters, provided that the product quality is qualified. This will gradually improve the stability and reliability of its defect recognition, and lay a more solid data foundation for subsequent multi-camera combination reconstruction processing.

[0067] S23 uses the proportion of the focus camera and the proportion of associated defect types in different defect identification focus cameras to determine the control strategy for the production parameters of the composite material profile.

[0068] Furthermore, based on the proportion of cameras of interest and the proportion of risk identification cameras among defect identification cameras, the combined reconstruction requirement value for cameras of different phases under different quality defect types is determined. It is then determined whether the combined reconstruction requirement value is greater than a preset requirement threshold. If so, the control strategy for the production parameters of the composite material profile is determined to be a relaxed control strategy. That is, only when the production parameters of the production equipment are within the acceptable range and the rate of change between the production parameters and the optimal production parameters is greater than a preset rate of change threshold, is it necessary to adjust the production equipment to the optimal production parameters. This allows for the acquisition of defect types under various operating parameters and the determination of different phases. If the cameras of different phases can be reliably combined and reconstructed, then the control strategy for the production parameters of the composite material profile is determined to be another control strategy. That is, as long as the production parameters of the production equipment are within the qualified range, and there are no multiple production equipment that need to be adjusted due to the rate of change of the production parameters from the optimal production parameters being greater than the preset rate of change threshold within the most recent preset time period, then the production equipment needs to be adjusted only when the rate of change of the production parameters from the optimal production parameters is greater than the preset rate of change threshold. That is, the production parameters are adjusted to the optimal production parameters, so as to obtain the defect types under various operating parameters and determine that the cameras of different phases can be reliably combined and reconstructed.

[0069] The proportion of risk-identifying cameras among defect-identifying focus cameras refers to the ratio of the number of risk-identifying cameras to the total number of defect-identifying focus cameras, used to measure the proportion of cameras with insufficient stability among defect-identifying focus cameras. The combined reconstruction demand value is a comprehensive index calculated based on the proportion of focus cameras and the proportion of risk-identifying cameras, used to quantify the overall demand for combined reconstruction processing of multiple quality defect types using different phase cameras under current operating conditions. A higher combined reconstruction demand value indicates a more lenient production parameter control strategy is needed to obtain more comprehensive defect type data. The preset demand threshold is a critical demand value used to determine whether the combined reconstruction demand value is already high and requires a lenient control strategy. Other control strategies refer to the production parameter control modes adopted when the combined reconstruction demand value does not exceed the preset demand threshold. In this mode, the control is stricter than the lenient control strategy. If there are no multiple production devices adjusted due to excessive parameter variation rates within the recent preset time period, adjustments are only made when the variation rate between the production parameters and the optimal production parameters exceeds the preset variation rate threshold. This aims to ensure product quality stability through a stricter control strategy while appropriately acquiring diverse defect type data within permissible limits.

[0070] The formula for calculating the combined reconstruction requirement value is: Combined reconstruction requirement value = Percentage of cameras of interest × 0.5 + Percentage of risk-identifying cameras among defect-identifying cameras of interest × 0.5.

[0071] Assuming the proportion of cameras under observation is A, and the proportion of risk-identifying cameras among defect-identifying cameras under observation is B, calculate the combined reconstruction requirement value. If the combined reconstruction requirement value exceeds the preset requirement threshold, a lenient control strategy is adopted. If it does not exceed the threshold, other control strategies are adopted. Under the premise that the production parameters are qualified and no multiple devices have been adjusted due to excessive parameter change rate within the recent preset time period, adjustments are only made when the parameter change rate exceeds the preset change rate threshold, thus achieving stricter quality control.

[0072] This step finalizes the production parameter control strategy through comprehensive calculation of the combined reconstruction demand value. Its significance lies in the fact that when both the proportion of cameras under observation and the proportion of cameras identifying risks are at a moderate level, the combined indicators can more comprehensively assess the current multi-camera joint detection system's demand for diverse defect type data. When the combined reconstruction demand value is high, a relaxed control strategy is adopted to fully acquire various defect type data and improve the reliability of multi-camera combined reconstruction. When the combined reconstruction demand value is low, other control strategies are adopted to ensure product quality in a more stringent manner, while appropriately acquiring defect type data within the allowable range, achieving a fine balance between quality control and the sufficiency of defect identification.

[0073] Following the results of Example S1: all 6 cameras (C1 to C6) are defect identification cameras of interest, with a total of 8 types of quality defects. The preset threshold for the proportion of cameras of interest is set at 0.80. The optimal production parameters are: pultrusion speed 0.8 m / min, die temperature 180°C, holding pressure 12 MPa, preset variation rate threshold 15%, preset defect type proportion threshold 0.50, preset risk camera quantity threshold 2, and preset demand threshold 0.60.

[0074] In S21, the number of defect identification cameras is 6, the total number of cameras is 6, and the proportion of cameras under observation is 6 ÷ 6 = 1.00, which is greater than the preset threshold of 0.80. Therefore, the control strategy for the production parameters of composite material profiles is determined to be a relaxed control strategy.

[0075] Based on a relaxed control strategy, under the premise that the production parameters of the production equipment are within the acceptable range (pultrusion speed 0.6~1.0m / min, die temperature 170~190℃, holding pressure 10~14MPa), the rate of change between the production parameters and the optimal production parameters is monitored in real time. When the rate of change of the pultrusion speed from 0.8m / min exceeds 15%, or the rate of change of the die temperature from 180℃ exceeds 15%, or the rate of change of the holding pressure from 12MPa exceeds 15%, the corresponding production parameters are adjusted to the optimal value, thereby retaining an appropriate range of parameter variation within the acceptable range and obtaining defect type data under various operating parameters.

[0076] Specifically, the method for determining the type of defect that the defect identification camera is interested in is as follows: In this embodiment, based on the changes in defect identification data from the defect identification focus camera, the addition of new quality defect types in different defect identification focus cameras is determined. That is, the more defect identification focus cameras there are with new quality defect types, the more quality defect types are monitored. This determines the reliability of the multi-phase camera combination reconstruction processing for composite material profiles that simultaneously have both monitored and newly added quality defect types. Therefore, by determining the monitored defect types of the defect identification focus cameras, a foundation is laid for further determining the quality detection methods for different cameras, and for improving the reliability of quality defect type identification and the reliability of the observation processing of the combination reconstruction processing.

[0077] S31 determines the newly identified quality defect type in the defect identification camera by analyzing the changes in defect identification data from different defect identification focus cameras. The changes in defect identification data refer to the changes in the types of quality defects identified by each defect identification camera when identifying defects in a new batch of composite material profiles during the execution of the production parameter control strategy, relative to the types of defects identified in the historical statistical period. The newly identified quality defect types refer to those that newly appear in the defect identification data of a particular defect identification camera after the current production parameter adjustment, and were not identified by that camera in the historical statistical period, reflecting the impact of production parameter changes on the scope of defect type identification.

[0078] Assuming that during the execution of a relaxed production parameter control strategy, multi-camera defect identification is performed on each batch of composite profiles. The current set of identified defect types of each camera is compared with the set of identified defect types in its historical statistical period. Defect types that exist in the current set but not in the historical set are extracted as the newly identified quality defect types of that camera, and are used for the subsequent summary of newly added defect types and the determination of defect types of concern.

[0079] This step involves monitoring changes in defect identification data during the execution of production parameter control strategies. The significance of this step lies in the fact that adjustments to production parameter control strategies can alter the production conditions of composite profiles, potentially leading to the new identification of certain quality defect types that only manifest under specific parameter conditions. By extracting these newly identified quality defect types, the effectiveness of production parameter control strategies in expanding defect type coverage can be objectively evaluated, providing a real-time data foundation for subsequent monitoring of dynamic updates to defect types.

[0080] S32 takes the newly identified quality defect type in the defect identification focus camera as the newly added defect type of the defect identification focus camera; The newly added defect types refer to the set of newly added defect types of a camera after deduplication and confirmation, based on the newly identified quality defect types. This set reflects the expansion of the camera's defect identification capability during the execution of production parameter control strategies.

[0081] Suppose that the newly identified quality defect types of each defect identification focus camera are deduplicated, and the confirmed valid newly identified defect types are classified as new defect types of that camera. This is used to determine whether it is necessary to combine and reconstruct the identification results of all cameras to verify the authenticity of the new defect types.

[0082] This step confirms the newly identified defect types as new defect types, providing a reliable data foundation for subsequently determining the defect types of interest based on the new defect types. It also determines the need for combining images from multiple cameras under the defect types of interest and the new defect types. The more new defect types there are, the higher the need for combining images from multiple cameras under the defect types of interest and the new defect types.

[0083] S33 determines the type of defect that the defect recognition camera is interested in based on the newly added defect types of different defect recognition cameras.

[0084] Specifically, based on the newly added defect types of different defect identification focus cameras, the type of defect to be focused on by the defect identification focus camera is determined, including: It should be noted that, in case 1: S331 determines whether all defect recognition cameras have new defect types. If so, all quality defect types under the defect recognition cameras with a recognition count of more than a preset threshold (less than the preset recognition count threshold) are regarded as defect types of concern. This determines whether the combined reconstruction processing of images from different phase cameras can be achieved under the above-mentioned defect types of concern and under the new defect types. If not, proceed to step S332. The preset number threshold (less than the preset recognition number threshold) refers to the lower recognition number threshold used when determining the type of defect of interest. When all cameras of interest have new defect types, it indicates that the demand for multi-camera combination reconstruction is high. Using a lower threshold will include more quality defect types in the range of defect types of interest, so as to ensure that the combination reconstruction processing of multi-camera images can cover all relevant defect types.

[0085] Assuming that all defect identification cameras have new defect types, it means that each camera has identified new defect types during the execution of the production parameter control strategy. The demand for multi-camera image combination reconstruction is high. In this case, all quality defect types in each defect identification camera with a recognition count above a preset threshold (less than the preset recognition count threshold) are taken as defect types of concern to expand the coverage of defect types of concern and ensure that the combination reconstruction processing can cover more comprehensive defect types.

[0086] This situation expands the coverage of the defect types of interest when all cameras focused on defect identification have newly added defect types. The significance of this is that when all key cameras have identified the newly added defect types, it indicates that the current production parameter control strategy has successfully expanded the defect identification range of each camera. The number of defect types that need to be covered in the multi-camera image combination reconstruction processing increases accordingly. By using a lower preset threshold for the number of times to include more defect types in the scope of interest, it is possible to ensure that the combination reconstruction processing fully covers all types of defect types and improve the reliability of the combination reconstruction processing under multiple defect types.

[0087] Additionally, it should be noted that in case 2: S332, the defect identification focus camera with newly added defect types is used as the reconstruction verification requirement camera. The number of newly added defect types in the reconstruction verification requirement camera is used to determine the newly added verification weight value of the defect type of the reconstruction requirement camera. It is then determined whether there is a reconstruction requirement camera with a newly added verification weight value greater than the preset verification weight threshold. If yes, then proceed to step S333. If no, then the focus defect type of the defect identification focus camera is the focus defect type in which the number of times the combination reconstruction of cameras with different phases in the defect identification focus camera is less than the preset reconstruction number threshold. The "reconstruction verification requirement camera" refers to a defect identification focus camera with newly added defect types. This means that the camera identified a new defect type during the execution of the production parameter control strategy, requiring verification of the identification result through multi-camera combined reconstruction processing. The "new verification weight value" is a weighted value calculated based on the number of newly added defect types in the reconstruction verification requirement camera. The more newly added defect types, the higher the new verification weight value, used to quantify the urgency of needing combined reconstruction verification for newly added defect types in each reconstruction verification requirement camera. The "preset verification weight threshold" is a critical verification weight value used to determine whether a reconstruction verification requirement camera has a large number of newly added defect types and requires initiating a full-scale expansion of focus defect types. The "preset reconstruction number threshold" is a critical reconstruction number value used to determine whether a focus defect type has been sufficiently verified through multi-camera combined reconstruction. Focus defect types with a reconstruction number less than this threshold indicate that they have not been sufficiently verified and need to be included in the focus defect type range for further combined reconstruction verification.

[0088] The formula for calculating the new verification weight value is: New verification weight value = Number of new defect types ÷ Total number of all quality defect types.

[0089] If the new verification weight values ​​of cameras with newly added defect types do not exceed the preset verification weight threshold, it means that the number of newly added defect types in each camera is small. In this case, the defect types that are of concern in each defect identification focus camera are included in the scope of focus defect types if the number of combined reconstructions of different phase cameras is less than the preset reconstruction number threshold. This is to focus on combined reconstruction verification of defect types that have not been fully verified. If there are cameras with newly added verification weight values ​​that exceed the preset verification weight threshold, then proceed to S333 for full expansion judgment.

[0090] This step focuses on including the types of defects that have not yet been fully verified when the newly added verification weight value does not exceed the limit. The significance of this step is that when the number of newly added defect types for each camera requiring reconstruction verification is small, it means that the occurrence of newly added defect types is not common. At this time, there is no need to expand all the types of defects to be of concern. Instead, the focus is on including the types of defects that have not yet been fully verified by multiple camera combination reconstructions. This concentrates the limited combination reconstruction computing resources on the defect types that need to be verified the most, thereby improving the targeting and efficiency of combination reconstruction verification.

[0091] It also includes the following: Case 3: S333 The reconstruction requirement camera with the newly added verification weight value greater than the preset verification weight threshold is used as the screening requirement camera. Based on the number of the screening requirement cameras and the number of newly added defect types in all defect recognition focus cameras, the reconstruction verification requirement coefficient is determined. It is determined whether the reconstruction verification requirement coefficient is greater than the preset requirement coefficient threshold. If so, all quality defect types in the defect recognition focus cameras with a recognition count greater than the preset count threshold (less than the preset recognition count threshold) are used as focus defect types. This determines whether the combined reconstruction processing of images from different phase cameras can be achieved under the above focus defect types and the newly added defect types. If not, all associated defect types in the defect recognition focus cameras are used as focus defect types.

[0092] The "screening demand camera" refers to a camera with a newly added verification weight value greater than a preset verification weight threshold, indicating that the number of newly added defect types in that camera is already quite large, necessitating the full expansion of all defect types of concern. The "reconstruction verification demand coefficient" is a comprehensive coefficient calculated based on the number of screened demand cameras and the number of newly added defect types in all defect identification focus cameras. It is used to quantify the urgency of whether a full expansion of the focus defect types is necessary under the current operating conditions. The preset demand coefficient threshold is the critical demand coefficient value used to determine whether the reconstruction verification demand coefficient is already high and whether all defect types that have met the recognition count requirement should be included in the focus defect types.

[0093] The formula for calculating the refactoring verification requirement coefficient is: Refactoring verification requirement coefficient = (Number of cameras for screening requirements ÷ Number of cameras for total defect identification) × 0.6 + (Total number of new defect types in all cameras for total defect identification ÷ Total number of all quality defect types) × 0.4.

[0094] If the reconstruction verification demand coefficient exceeds the preset demand coefficient threshold, it means that the types of defects of interest need to be fully expanded. In this case, all quality defect types in each defect identification camera that have been identified more than the preset number of identifications (less than the preset number of identifications threshold) will be regarded as defect types of interest to fully cover all types of defects. If the reconstruction verification demand coefficient does not exceed the threshold, all related defect types in each defect identification camera will be regarded as defect types of interest, and coverage will be based on related defect types.

[0095] This step determines the coverage of the defect types of interest through the comprehensive calculation of the reconstruction verification demand coefficient. Its significance lies in the fact that when there are newly added cameras with high reconstruction demand values, it indicates that the number of newly added defect types for some cameras is already quite large, and a more comprehensive expansion of the defect types of interest is needed to cover the needs of multi-camera combination reconstruction. By judging the reconstruction verification demand coefficient, all defect types that have met the recognition count are included in the scope of interest when the demand is high, and all related defect types are included when the demand is not high enough. This ensures that the coverage of the defect types of interest accurately matches the distribution of the currently added defect types, providing a reliable foundation for determining the subsequent joint quality detection method.

[0096] Using the results of the previous embodiments: the defect identification focuses on 4 cameras, C1, C2, C4 and C6 (the case in embodiment S22), and there are a total of 8 types of quality defects.

[0097] In S31, during the execution of the relaxed control strategy, the changes in defect identification data of each defect identification camera are monitored: C1 newly identified quality defect type: color unevenness (not previously identified by C1); C2 newly identified quality defect type: wall thickness unevenness; C4 newly identified quality defect type: surface scratches, exposed fibers; C6 newly identified quality defect type: internal cracks, color unevenness.

[0098] In S32, the newly identified quality defect types are added as new defect types: C1 new defect type: uneven color; C2 new defect type: uneven wall thickness; C4 new defect type: surface scratches, exposed fibers; C6 new defect type: internal cracks, uneven color.

[0099] Enter S33. S331 determines: If all defect identification focus cameras (C1, C2, C4, C6) have new defect types, then all quality defect types in each defect identification focus camera that have been identified more than 10 times according to the preset number threshold will be regarded as focus defect types.

[0100] Quality defect types that are identified more than 10 times by each camera: C1: Surface bubbles (12 times), size deviation (15 times), internal cracks (8 times), exposed fibers (7 times) - among which surface bubbles and size deviation reach the preset identification threshold of 10 times, which are related defect types and proceed to S4.

[0101] S3 determines a joint quality inspection method for composite material profiles based on image data, according to the defect type data of different defect identification cameras and the control strategy of the production parameters of the composite material profiles.

[0102] Specifically, the method for determining the quality of composite profiles based on image data is as follows: In this embodiment, based on the defect type data from different defect identification cameras, the requirements for combined reconstruction analysis and processing of images from different defect identification cameras under different defect types are determined. Specifically, the larger the defect type of different defect identification cameras, the higher the requirements for combined reconstruction analysis and processing of images from different defect identification cameras under that defect type. Simultaneously, the requirements for the reliability of defect type identification are also higher. Therefore, by considering the requirements for the reliability of defect type identification and the control strategy of composite material profile production parameters, a joint quality detection method for composite material profiles based on image data is determined. That is, the higher the requirement for the reliability of defect type identification, the stricter the control of the production parameters of the load-bearing material profile, and the less likely it is to obtain defect types under multiple production parameters, thus increasing the requirement for the reliability of defect type identification. Therefore, by combining the requirements for the reliability of defect type identification, a joint quality detection method for composite material profiles is determined. This further improves the reliability of defect type identification processing from cameras with different phases while reducing data processing and control difficulties. Based on the identified defect types, a foundation is laid for further improving the reliability of combined reconstruction processing under multiple defect types of the same composite material profile.

[0103] S41 determines the number of defect types in the defect identification camera based on the defect type data in different defect identification cameras. The data on the types of defects of interest refers to the quality defect types and their related identification statistics recorded in the set of defects of interest for each defect identification camera. The number of defects of interest refers to the total number of quality defect types included in the set of defects of interest for a certain defect identification camera. It is used to measure the coverage of defect types that the camera needs to focus on in the joint quality detection system. The more defects of interest the camera has, the more complex the joint detection scenario that the camera needs to participate in, and the higher the priority of the camera as a camera for identification needs.

[0104] Suppose we statistically analyze the set of defect types of each camera that is of interest for defect identification, and obtain the number of defect types of each camera: C1 has 6 defect types, C2 has 5 defect types, C4 has 7 defect types, and C6 has 6 defect types. This number is used to compare with the preset threshold for the number of defect types of interest to determine which cameras are required for identification.

[0105] This step involves counting the number of defect types that each camera focuses on for defect identification. The significance of this step is that the number of defect types is a core indicator for measuring the importance of a camera in a multi-camera quality joint detection system. The more defect types a camera focuses on, the wider the range of defect types it needs to cover, and the higher its demand for participating in multi-camera combination reconstruction processing. Count statistics provide a quantitative basis for the selection of cameras for subsequent identification needs.

[0106] S42 uses the number of the types of defects of interest to determine the cameras with recognition needs among the defect recognition cameras of interest; The camera for identifying the required defects refers to a defect identification camera whose number of defect types exceeds a preset threshold. In other words, the camera has a wide coverage of defect types. If there are many cameras for identifying defects, the observation requirements for the reliability of multiple cameras combined and reconstructed based on the aforementioned defect types will be higher.

[0107] Assuming that the number of defect types of each camera being identified is compared with a preset threshold for the number of defect types of each camera, cameras whose number of defect types of each camera exceeds the threshold are designated as cameras for identification.

[0108] Furthermore, if the number of cameras required for identification is greater than the preset threshold for the number of required cameras, it is determined that the need for reconstruction processing of cameras with different phases is high at this time. Therefore, it is determined that all cameras perform quality defect detection processing on the same composite material profile under a preset number of lighting methods.

[0109] The preset threshold for the number of required cameras refers to the critical number of cameras required to determine whether the number of cameras needed for identification is already too large and whether the full-camera multi-lighting detection mode needs to be activated. The phrase "all cameras perform quality defect detection processing on the same composite profile under a preset number of lighting modes" means that when the number of required cameras is large, due to the high demand for multi-camera reconstruction processing, all cameras on the production line need to perform image acquisition and defect detection on the same composite profile under multiple different lighting modes to fully obtain the defect manifestation of the profile under different lighting conditions, providing diversified data sources for the combined reconstruction of multi-camera images.

[0110] Furthermore, the illumination method includes at least one of front diffuse illumination, back transmission illumination, infrared illumination, side illumination, polarized illumination, and ultraviolet illumination. The reliability of identifying quality defect types varies under different illumination methods. Specifically: front diffuse illumination has high reliability in identifying surface scratches and uneven color; back transmission illumination has high reliability in identifying uneven wall thickness, end damage, and internal cracks; infrared illumination has high reliability in identifying surface bubbles, uneven fiber distribution, and internal cracks; side illumination has high reliability in identifying exposed fibers, dimensional deviations, and surface scratches; polarized illumination has high reliability in identifying surface bubbles and defects after suppressing surface reflection; and ultraviolet illumination has high reliability in identifying coating defects and fluorescent marking defects.

[0111] If the number of cameras required for identification exceeds the preset threshold for the number of cameras required, it indicates that many cameras have a wide range of defect types to cover and the demand for multi-camera reconstruction processing is high. In this case, it is determined that all cameras in the production line will perform defect detection on the same composite material profile under a preset number of lighting modes. By acquiring images from multiple cameras and under multiple lighting modes, the comprehensiveness of defect display is maximized.

[0112] This judgment initiates the full-camera multi-illumination detection mode when there are a large number of cameras required for identification. Its significance lies in the fact that when a large number of cameras have extensive coverage of the types of defects of interest, it indicates that the types of defects that the current joint quality inspection needs to cover are quite complex and diverse. Under a single illumination mode, some types of defects may not be sufficiently revealed. By having all cameras inspect the same profile under multiple illumination modes, the different revealing capabilities of different illumination modes for various defects can be fully utilized, providing the most sufficient data foundation for the combined reconstruction of multi-camera images and improving the reliability and comprehensiveness of the combined reconstruction process.

[0113] If the number of cameras required for identification is not greater than a preset threshold for the number of required cameras, the following applies: S431 determines the recognition requirement weight value of the defect recognition camera based on the product of the number of the defect types of interest in the defect recognition camera and a preset scaling factor. It then determines whether the sum of the recognition requirement weight values ​​of different defect recognition cameras is greater than the preset weight value. If yes, proceed to step S432. If no, determine that the image data-based composite material profile quality joint detection method is to perform quality defect detection processing on the same composite material profile under a preset number of lighting modes in all defect recognition cameras with the defect types of interest. The preset scaling factor refers to the conversion coefficient used to calculate the identification demand weight value, converting the number of defect types of concern into weight values. The identification demand weight value refers to the weight value calculated by multiplying the number of defect types of concern of each defect identification camera by the preset scaling factor, used to quantify the demand level of each defect identification camera in the joint quality inspection system. The preset weight value refers to the critical weight sum used to determine whether the sum of the identification demand weight values ​​of all defect identification cameras is already high and a more comprehensive detection mode needs to be adopted.

[0114] The formula for calculating the weight value of the identified demand is: Weight value of identified demand = Number of defect types to be concerned × Preset ratio factor.

[0115] If the sum of the recognition demand weights of each defect recognition camera does not exceed the preset weight value, it indicates that the overall demand is not high. In this case, it is determined that the same composite material profile will be detected and processed under a preset number of lighting methods in all defect recognition cameras with the type of defect of interest. That is, multiple lighting methods are only used for detection within the range of cameras with the type of defect of interest, so as to balance the comprehensiveness of detection and the system processing efficiency.

[0116] This step employs a limited-camera multi-illumination detection mode when the sum of the identified demand weight values ​​does not exceed the limit. Its significance lies in the fact that when the number of cameras identifying demand is small and the overall weighted demand for the coverage of the defect types of interest by each camera is not high, it is not necessary to start full-camera multi-illumination detection. Multi-illumination detection is only performed within the range of cameras where the defect types of interest actually exist. This can concentrate detection resources on cameras that truly have demand, ensuring detection reliability while avoiding excessive increase in the system's data processing burden and detection latency.

[0117] S432 determines whether the control strategy for the production parameters of the composite material profile is a lenient control strategy. If yes, it determines that the image data-based composite material profile quality joint detection method is to perform quality defect detection processing on the same composite material profile under a preset number of lighting modes in all defect recognition focus cameras with the type of defect of interest. If no, it determines that the image data-based composite material profile quality joint detection method is to perform quality defect detection processing on the same composite material profile under a preset number of lighting modes in all defect recognition focus cameras.

[0118] The relaxed control strategy refers to the relatively relaxed production parameter control mode determined in step S2. When the production parameter control strategy is relaxed, it means that the current production parameters retain a moderate range of variation within the acceptable range, allowing for the acquisition of more diverse defect type data. The defect identification results of each camera are relatively comprehensive. In this case, multi-illumination detection within the camera range containing the defect type of interest is sufficient to meet the requirements of combined reconstruction. When the production parameter control strategy is not relaxed (i.e., it is strict control or other control strategies), it means that the production parameter control is relatively strict, and the acquisition of defect type data is relatively limited. In this case, a more comprehensive detection mode is required, performing multi-illumination detection in all defect identification cameras of interest to compensate for the limited acquisition of defect type data under strict parameter control and ensure the reliability of the combined reconstruction process.

[0119] If the production parameter control strategy is a lenient control strategy, then multi-illumination detection is performed in the defect identification camera where the defect type of interest exists; if the control strategy is a strict control strategy or other control strategies, then multi-illumination detection is performed in all defect identification cameras to compensate for the limited acquisition of defect type data by expanding the detection camera range.

[0120] This step determines the camera range for multi-illumination detection by judging the type of production parameter control strategy. Its significance lies in the fact that the strictness of the production parameter control strategy directly affects the comprehensiveness of defect type data that each camera can acquire. Under a lenient control strategy, the defect type data is relatively comprehensive, and multi-illumination detection within a limited camera range can meet the requirements. Under a strict control strategy, the acquisition of defect type data is limited, and it is necessary to expand the detection camera range to all defect identification focus cameras. This is to compensate for the limited defect type data through sufficient coverage of multi-illumination detection, ensuring that the quality joint detection method can achieve reliable multi-camera image combination reconstruction processing under different levels of production parameter control strategies.

[0121] Following the results of the previous embodiments: the defect identification focuses on 4 cameras (C1, C2, C4, and C6), and there are 8 types of defects (all defect types). The production parameter control strategy is a lenient control strategy. The preset threshold for the number of defect types is 4, the preset threshold for the number of required cameras is 2, the preset scaling factor is 0.25, and the preset weight value is 1.50.

[0122] In S41, the number of defect types of each defect identification camera is as follows: C1 defect types: surface bubbles, internal cracks, dimensional deviations, surface scratches, color unevenness, and exposed fibers, totaling 6 types; C2 defect types: internal cracks, surface scratches, color unevenness, end damage, and wall thickness unevenness, totaling 5 types.

[0123] In S42, the number of defect types of interest is compared with the preset threshold of 4 defect types of interest: C1 (6 types) is greater than 4 types, belonging to the camera for identification needs; C2 (5 types) is greater than 4 types, belonging to the camera for identification needs; C4 (7 types) is greater than 4 types, belonging to the camera for identification needs; C6 (7 types) is greater than 4 types, belonging to the camera for identification needs. The number of cameras for identification needs is 4, which is greater than the preset threshold of 2 cameras for identification needs. It is determined that all cameras (C1 to C6) will perform quality defect detection processing on the same composite material profile under a preset number of lighting methods (assuming the preset number is 3: front diffuse lighting, back transmitted lighting, infrared lighting, side lighting, polarized lighting, and ultraviolet lighting, three of which are selected).

[0124] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments of apparatus, devices, and non-volatile computer storage media are basically similar to the method embodiments, so the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0125] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0126] The above description is merely one or more embodiments of this specification and is not intended to limit this specification. Various modifications and variations can be made to the one or more embodiments of this specification by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of one or more embodiments of this specification should be included within the scope of the claims of this specification.

Claims

1. A method for joint quality inspection of composite profiles based on image data, characterized in that, Specifically, it includes: Using the quality inspection data of composite profiles, the defect identification results of cameras with different phases are determined. Based on the defect identification results, the degree of dispersion of identification processing under different quality defect types is determined. Based on the degree of dispersion of identification processing of cameras with different phases, the defect identification focus camera in the camera is determined. Based on the distribution data of the defect identification focus camera and the defect identification results of different defect identification focus cameras, the control strategy for the production parameters of composite material profiles is determined. Based on the control strategy, adjustment processing is carried out during the change of the production parameters of the composite material profiles. Based on the change of the defect identification data of different defect identification focus cameras during the adjustment processing, the focus defect type of the defect identification focus camera is determined. Based on the defect type data of different defect identification cameras and the control strategy of the production parameters of the composite material profile, a joint quality inspection method for composite material profiles based on image data is determined.

2. The method for joint quality inspection of composite profiles based on image data as described in claim 1, characterized in that, The quality inspection data includes the quality defect detection results of cameras at different locations in the production line of the composite material profile.

3. The method for joint quality inspection of composite profiles based on image data as described in claim 1, characterized in that, The camera defect identification results include the identification results of the camera under different quality defect types.

4. The method for joint quality inspection of composite profiles based on image data as described in claim 1, characterized in that, The method for determining the camera of interest for defect identification in the aforementioned camera is as follows: Based on the defect identification results of the camera, determine the number of times the camera identifies different types of quality defects; Based on the number of recognitions, the quality defect types in the camera that have a recognition count above a preset recognition count threshold are determined and used as associated defect types. Based on the degree of dispersion of associated defect types in different cameras, the defect identification focus camera in the camera is determined.

5. The method for joint quality inspection of composite profiles based on image data as described in claim 4, characterized in that, Based on the degree of dispersion of associated defect types in different cameras, the camera of interest for defect identification is determined, specifically including: If the number of cameras with associated defect types exceeds a preset threshold, then all cameras are identified as cameras of interest for defect identification.

6. The method for joint quality inspection of composite profiles based on image data as described in claim 1, characterized in that, The method for determining the control strategy for the production parameters of the composite material profile is as follows: Based on the distribution data of the defect identification focus camera, determine the proportion of the defect identification focus camera among all cameras, and use the proportion of the defect identification focus camera among all cameras as the focus camera proportion. Based on the defect identification results of different defect identification focus cameras, determine the number of associated defect types in the defect identification focus camera, and determine the proportion of associated defect types of the defect identification focus camera based on the proportion of the number of associated defect types in all defect types identified by the defect identification focus camera. By utilizing the proportion of the focus cameras and the proportion of associated defect types in different defect identification focus cameras, the control strategy for the production parameters of the composite material profile is determined.

7. The method for joint quality inspection of composite material profiles based on image data as described in claim 6, characterized in that, If the proportion of the focus camera is greater than the preset focus camera proportion threshold, then the control strategy for the production parameters of the verification material profile is determined to be a relaxed control strategy. That is, only when the production parameters of the production equipment are within the qualified range and the rate of change of the production parameters from the optimal production parameters is greater than the preset rate of change threshold, is it necessary to adjust the production equipment to the optimal production parameters. In this way, the defect types under various operating parameters can be obtained, and the cameras of different phases can be reliably combined and reconstructed.

8. The method for joint quality inspection of composite profiles based on image data as described in claim 7, characterized in that, The optimal production parameters are those that minimize the occurrence of production quality defects.

9. The method for joint quality inspection of composite material profiles based on image data as described in claim 1, characterized in that, The method for determining the image data-based composite profile quality joint inspection method is as follows: The number of defect types in the defect identification camera is determined based on the defect type data in different defect identification cameras. Using the number of the types of defects of interest, determine the cameras that require identification among the defect identification cameras; Based on the identification requirements of the camera and the control strategy of the production parameters of the composite material profile, the image data-based joint quality inspection method for composite material profiles is determined.

10. The method for joint quality inspection of composite material profiles based on image data as described in claim 9, characterized in that, The camera for identifying the required defects is a camera whose number of defect types of interest is greater than a preset threshold for the number of defect types of interest.

Images