Glass bracket production line full-process visual monitoring and intelligent management method
By constructing a full-process image data sequence and dimensional defect detection on the glass bracket production line, the problems of insufficient monitoring and low detection accuracy in the existing technology are solved, and efficient and accurate quality control of the glass bracket production line is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-17
- Publication Date
- 2026-04-14
AI Technical Summary
Existing technologies lack continuous monitoring throughout the glass bracket production process, resulting in low detection accuracy. They are prone to misjudgment due to human error and environmental factors, cannot accurately calculate pixel area difference rates, and easily miss small or hidden defects during defect detection, leading to quality fluctuations and resource waste.
By deploying image acquisition units on the glass support production line, a full-process image data sequence is constructed. Combined with a database of qualified glass support size images, size detection and defect detection are performed, generating a visual quality report. Automated judgment is made using pixel region overlap rate and preset difference rate thresholds.
It enables comprehensive visual monitoring of the glass bracket production line, improves the accuracy and efficiency of inspection, reduces manual inspection errors and production costs, and ensures product quality consistency and production efficiency.
Smart Images
Figure CN121860918A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of industrial quality control technology, specifically, it relates to a method for full-process visual monitoring and intelligent management of glass bracket production lines. Background Technology
[0002] With the rapid development of the global economy, the manufacturing industry has increasingly higher requirements for production efficiency and product quality. In the field of photovoltaic glass bracket production, traditional production monitoring and quality inspection methods are no longer sufficient to meet the needs of modern production.
[0003] Existing technologies typically rely on discrete inspection steps or manual sampling, lacking continuous and systematic monitoring of the entire production process. This leads to difficulties in problem localization and low efficiency. Specifically, existing technologies struggle to track the complete state changes of each glass bracket from start to finish in real time. Secondly, in terms of dimensional inspection, they largely rely on manual measurement or comparison with fixed templates, lacking dynamic correlation with a database of qualified glass bracket size images. This makes it impossible to accurately calculate pixel area difference rates, and misjudgments are easily caused by human error or environmental factors. In terms of defect detection, existing technologies typically employ basic image processing or visual inspection, which can easily miss minute or hidden defects. There are significant deficiencies in the comprehensiveness, accuracy, and automation of production monitoring, which can easily lead to quality fluctuations and resource waste.
[0004] To address the aforementioned issues, this invention proposes a method for full-process visual monitoring and intelligent management of glass bracket production lines. Summary of the Invention
[0005] To address the shortcomings of existing technologies, this invention provides a method for full-process visual monitoring and intelligent management of glass bracket production lines, solving the problems of insufficient full-process monitoring and low detection accuracy in existing technologies.
[0006] The objective of this invention can be achieved through the following technical solutions: A method for full-process visual monitoring and intelligent management of a glass support production line, the method comprising: Step 1: Obtain the glass bracket production line and monitor the entire process of the glass bracket production line using a pre-built image acquisition unit. Continuously acquire image data of the glass brackets at any station in the glass bracket production line and sort them according to the entire process of the glass bracket production line to construct the full process image data sequence associated with the corresponding glass bracket. Step 2: Extract image data of the glass bracket at any workstation based on the full-process image data sequence, and perform size detection in conjunction with the pre-constructed qualified glass bracket size image database to determine whether the size status of the glass bracket belongs to an abnormal size status. Step 3: Extract image data of the glass bracket in normal size at the corresponding workstation. Based on the pixel value of the image data, evaluate whether there is a defect area in the image data. Based on the defect area, perform defect detection on the glass bracket to determine whether the defect state of the glass bracket belongs to an abnormal defect state. Step four involves summarizing the glass supports assessed as having abnormal dimensions or defects, along with their associated workstations, defect detection results, dimension detection results, and the time of occurrence, to generate a visual quality report.
[0007] As a further aspect of the present invention, in step one, the image acquisition unit includes a visible light imaging component and a data transmission component, covering all workstations of the glass bracket production line; The visible light imaging component is deployed at the workstation for inspecting the appearance of the glass bracket surface, and is used to acquire image data under preset lighting conditions. The data transmission component is used to transmit image data to the computing processing unit; The pixel size of the glass bracket captured in the image data is in a 1:1 ratio with the actual object, and the pixel size of the image data is uniform. The pixel values of the background captured by the visible light imaging component are significantly different from those of the glass support.
[0008] As a further aspect of the present invention, the specific method for constructing the full-process image data sequence associated with the corresponding glass support in step one is as follows: Obtain all workstations in the glass support production line according to the order of the glass support production line, and arrange them as workstation sequence G1, G2, ..., Gj, where j is the total number of workstations; Obtain any glass bracket, denoted as BZ; The process of glass support BZ from the first station G1 to the last station Gj in the station sequence G1, G2, ..., Gj is determined. The j images captured by the image acquisition unit are sorted according to the station sequence G1, G2, ..., Gj to obtain the full process image data sequence V1, V2, ..., Vj, where V1 to Vj represent the first to j-th image data, and Gi corresponds to Vi, where i is the counting index with a value range of 1 to j.
[0009] As a further aspect of the present invention, in step one, the total number of image data in the full-process image data sequence of the glass bracket BZ is j only if and only if the glass bracket BZ has undergone dimensional inspection and defect inspection throughout the entire glass bracket production line process and has been determined to be in a non-abnormal defect state and a normal dimensional state; Conversely, the total number of image data in the entire process image data sequence of the glass bracket BZ is equal to the value of the station index where the glass bracket BZ is determined to be in an abnormal defect state or an abnormal size state.
[0010] As a further aspect of the present invention, the specific method for determining whether the dimensional state of the glass bracket belongs to an abnormal dimensional state in step two is as follows: Obtain any image data Vi and its corresponding workstation Gi from the full-process image data sequence of the glass bracket BZ; Extract the qualified glass bracket size image data B_Vi associated with the corresponding workstation Gi from the pre-built qualified glass bracket size image database; Based on background segmentation, the pixel portion of the glass support in the qualified glass support size image data B_Vi is extracted and denoted as B_XVi. Similarly, the pixel portion of the glass support in the image data Vi is extracted and denoted as XVi. A two-dimensional coordinate system is constructed using the bottom row of pixels in the qualified glass bracket size image data B_Vi as the horizontal axis and the leftmost column of pixels as the vertical axis. Label the pixel portion B_XVi in a two-dimensional coordinate system and determine the geometric center of the pixel portion B_XVi, denoted as B_OXVi; Determine the geometric center OXVi of the pixel portion XVi; Let the geometric center OXVi coincide with the geometric center B_OXVi, and plot the pixel part XVi in the two-dimensional coordinate system. Let the pixel part XVi rotate about the geometric center OXVi as the axis. The rotation is stopped if and only if the pixel region overlap rate between pixel region XVi and pixel region B_XVi is determined to be the maximum, and the pixel region difference rate between pixel region XVi and pixel region B_XVi at this time is extracted, where the pixel region difference rate = 100% - pixel region overlap rate. Compare the pixel region difference rate with the preset pixel region difference rate threshold; If the pixel area difference rate is greater than the pixel area difference rate threshold, the size status of the glass bracket BZ after processing by station Gi is determined to be an abnormal size status, and the glass bracket BZ is rejected and not included in the subsequent process. Conversely, if the glass bracket BZ is found to be of normal size, it will continue to participate in subsequent processes.
[0011] As a further aspect of the present invention, in step two, the qualified glass bracket size image database is pre-constructed by the operator; Any record in the qualified glass bracket size image database includes the corresponding workstation and the qualified glass bracket size image data associated with that workstation.
[0012] As a further aspect of the present invention, the specific method for determining whether the defect state of the glass bracket belongs to an abnormal defect state in step three is as follows: Obtain image data Vi of the glass bracket BZ, which is determined to be of normal size, at station Gi, and extract the pixel portion XVi of image data Vi; Compare the pixel values of pixel part XVi and pixel part B_XVi at the same coordinate position, determine the pixel value difference rate, and mark all pixels in pixel part XVi whose absolute value of pixel value difference exceeds the preset first difference threshold as initial difference pixels. Adjacent pixels in the initial difference pixels are aggregated into an independent candidate defect region. The physical area of each candidate defect region is calculated, where the physical area is equal to the total number of pixels in the candidate defect region multiplied by the actual physical area of a single pixel. If the actual physical area of the candidate defect region is greater than the preset first area threshold, then it is determined that there is a defect region in the image data Vi, and the defect state of the glass bracket BZ corresponding to the image data Vi is determined to be an abnormal defect state. Conversely, if the image data Vi does not contain a defective area, it is determined that the glass bracket BZ does not belong to an abnormal defect state.
[0013] As a further aspect of the present invention, the specific method for generating the visualized quality report in step four is as follows: If the glass bracket BZ at any workstation Gi is determined to be in an abnormal size state, the glass bracket BZ, workstation Gi, the time of occurrence of the size detection, and the result of the size detection are extracted and combined with the image data Vi of the glass bracket BZ at workstation Gi to generate a visual quality report. In this report, the size detection result is equal to the abnormal size state of the glass bracket BZ. If a glass bracket BZ at any workstation Gi is determined to be in an abnormal defect state, the glass bracket BZ, workstation Gi, defect detection result, and defect detection occurrence time are extracted and combined with the image data Vi of the glass bracket BZ at workstation Gi to generate a visual quality report. In this report, the defect detection result is equal to the abnormal defect state of the glass bracket BZ.
[0014] The beneficial effects of this invention are: This invention achieves comprehensive visualization and refined management of glass bracket production lines through full-process image monitoring and intelligent analysis. Its core advantage lies in constructing a coherent sequence of image data using image acquisition units covering all workstations, ensuring data consistency and traceability; quickly identifying dimensional anomalies by comparing dimensional inspection with a qualified database, and improving the accuracy and efficiency of inspection by combining pixel value analysis of defect areas; and finally generating a visualized quality report that summarizes anomaly information, facilitating real-time decision-making and problem localization, thereby reducing manual inspection errors and production costs, and improving product quality and production efficiency. This invention constructs a full-process image data sequence according to the workstation sequence of the glass bracket production line, ensuring comprehensive visual monitoring of the production process. Its core advantage lies in that a complete image sequence is only recorded when a product passes size and defect inspection and is determined to be normal; otherwise, the sequence terminates prematurely at the abnormal workstation. This improves the accuracy and efficiency of quality control, quickly locates defective workstations, simplifies the fault diagnosis process, and avoids redundant image acquisition and processing of abnormal products, thereby saving storage and computing resources. Furthermore, through serialized tracking, the reliability and traceability of the production line are enhanced, providing efficient, economical, and automated quality assurance for glass bracket manufacturing. This invention achieves automated detection of the dimensional status of glass supports by constructing a database of qualified glass support dimensions and utilizing background segmentation and geometric center alignment. Its core advantage lies in employing a pixel region overlap rate comparison mechanism, which can accurately identify products with abnormal dimensions, effectively avoiding the subjective errors and inefficiencies of manual inspection. Furthermore, by using a preset difference rate threshold and automatic rotation comparison, rejection decisions can be made quickly, improving the accuracy and consistency of the production process. This ensures product quality consistency while optimizing production line efficiency and reducing subsequent process risks caused by dimensional deviations. Attached Figure Description
[0015] The invention will now be further described with reference to the accompanying drawings.
[0016] Figure 1 This is a flowchart illustrating the method described in this invention; Figure 2 This is a flowchart illustrating the method described in Embodiment 2 of the present invention; Figure 3 This is a flowchart illustrating the method described in Embodiment 3 of the present invention; Figure 4 This is a flowchart illustrating the method described in Embodiment 4 of the present invention. Detailed Implementation
[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0018] Example 1 A method for full-process visual monitoring and intelligent management of glass bracket production lines, such as... Figure 1 As shown, this method includes the following: This method: A full-process visual monitoring and intelligent management method for photovoltaic glass bracket production lines integrates computer vision, image processing, data analysis and visualization technologies. It aims to realize real-time monitoring, quality inspection and intelligent management of photovoltaic glass brackets in the production process, thereby improving the production efficiency and product quality of photovoltaic glass brackets.
[0019] The first step is to acquire any glass support production line and monitor the entire process of the glass support production line using a pre-built image acquisition unit. Image data of glass supports at any workstation in the production line is continuously acquired and sorted according to the entire production line process to construct a sequence of full-process image data associated with each glass support. Specifically: The image acquisition unit includes a visible light imaging component (e.g., a high-resolution industrial camera) and a data transmission component (e.g., an Ethernet or 5G module). The visible light imaging component and the data transmission component cover all workstations on the glass bracket production line and acquire image data of the glass brackets (semi-finished glass brackets, hereinafter referred to as glass brackets) after the operation is completed at each workstation. The visible light imaging component acquires image data under lighting conditions preset by the operator, and the data transmission component transmits the image data to the computing and processing unit in real time via wired or wireless network. All subsequent analysis and calculation operations are completed in the computing and processing unit.
[0020] The pixel size of the glass bracket in the acquired image data is 1:1 with the actual object, meaning that each pixel corresponds to a fixed unit of the actual size. This is achieved by the operator through camera calibration and lens parameter pre-setting. Secondly, the pixel size of all image data is uniform, and the pixel value of the shooting background is uniform and has a significant difference from the pixel value of the glass bracket, which facilitates the use of background segmentation, such as the Otsu algorithm or the Canny operator algorithm.
[0021] After receiving the image data, the computing and processing unit sorts it according to the timestamp and workstation order, and constructs a full-process image data sequence for each glass bracket. The full-process image data sequence records the entire process of the glass bracket from entering the production line to the finished glass bracket, thus forming a traceable data chain.
[0022] The second step involves extracting image data of the glass support at any workstation based on the full-process image data sequence, and then performing dimensional detection in conjunction with a pre-built database of qualified glass support dimensions to determine whether the dimensional status of the glass support is abnormal. Specifically: Images of glass supports at specific workstations are extracted from the full-process image data sequence and compared with qualified glass support size image data at the corresponding workstations in the pre-built qualified glass support size image database. By comparing with the standard last shift, it is determined whether the size status of the glass support is abnormal. The qualified glass support size image database is pre-built by the operators, and the qualified glass support size image data in the qualified glass support size image database are all reviewed and marked by the operators. If the dimensions of the glass bracket are determined to be abnormal, the glass bracket will be rejected from the current workstation and considered a defective product, and will not proceed to the next step of the process.
[0023] The third step involves extracting image data of the glass bracket in its normal-sized state at the corresponding workstation. Based on the pixel values of the image data, the presence of defective areas is assessed. Defect detection is then performed on the glass bracket based on these defective areas to determine whether the defective state is an abnormal defect. Specifically: For glass brackets of normal size, further analysis of the pixel values in their image data is conducted to assess whether defective areas exist. If defective areas are found, defect detection continues to determine whether the defective state of the glass bracket is an abnormal defect. This is because sometimes, minor scratches can be considered blemishes rather than defects and do not affect the normal use of the glass bracket. Therefore, it is necessary to determine whether the defective state of the glass bracket is an abnormal defect. If the defective state of the glass bracket is determined to be an abnormal defect, the glass bracket is directly removed from the current workstation and regarded as a defective product, and will not proceed to the subsequent operation process.
[0024] The fourth step involves summarizing the glass supports assessed as having abnormal dimensions or defects, along with their associated workstations, defect detection results, dimensional inspection results, and the time of occurrence, to generate a visual quality report. Specifically: Information on glass supports that are assessed as having abnormal dimensions or defects is compiled, including workstation location, defect detection results, dimensional detection results, image data of the glass supports, and the time of occurrence. A visual quality report is generated and stored first, and then the operators are notified.
[0025] Example 2 This embodiment, based on embodiment 1, further discloses a method for constructing a complete image data sequence associated with the corresponding glass support, such as... Figure 2 As shown, it specifically includes the following: Based on the content described in Example 1, obtain any glass bracket production line, and extract each workstation in the entire glass bracket production line according to the production sequence of the glass bracket production line, and arrange them into a workstation sequence G1, G2, ..., Gj, where j represents the total number of workstations in the entire glass bracket production line, and the workstation sequence G1, G2, ..., Gj represents the logical flow of the glass bracket production process.
[0026] Next, obtain any glass bracket to be produced and denote it as BZ. As described in Example 1, glass bracket BZ in the glass bracket production line refers to the semi-finished glass bracket.
[0027] When the glass bracket BZ begins production, it will pass through each station in the sequence G1, G2, ..., Gj. During the process from the first station G1 to the last station Gj, after the glass bracket BZ completes processing at each station, the image acquisition unit takes a picture of it, acquiring image data of the glass bracket BZ at the corresponding station. This results in j images. These j images are then sorted according to the station sequence G1, G2, ..., Gj to obtain the full-process image data sequence V1, V2, ..., Vj associated with the glass bracket BZ. Here, V1 to Vj represent the first to the j-th image data, and station Gi corresponds to image data Vi, where i is a counting index ranging from 1 to j. It needs to be explained that the total number of image data in the full-process image data sequence of glass bracket BZ can be j only if the glass bracket BZ is determined to be in a non-abnormal defect state and a normal size state after dimensional inspection and defect inspection at every station in the glass bracket production line. Otherwise, when the glass bracket BZ is determined to be in an abnormal defect state or an abnormal size state at any station in the entire process of the glass bracket production line, the total number of image data in the entire process image data sequence of the glass bracket BZ is equal to the value of the index of that station. For example, if glass bracket BZ is determined to be in an abnormal defect state at workstation G24, then 24 image data associated with glass bracket BZ can be obtained. Therefore, the total number of image data in the entire process image data sequence of glass bracket BZ is equal to 24. Furthermore, when glass bracket BZ is determined to be in an abnormal defect state or an abnormal size state at workstation G24, glass bracket BZ is directly rejected and regarded as a defective product, and will not enter the subsequent operation process.
[0028] The purpose of this embodiment is to capture images of the glass bracket at each workstation using an image acquisition unit, forming a complete image data sequence for monitoring and detecting the size and defect status of the glass bracket during the production process. After the glass bracket is processed at each workstation, the acquisition unit will capture an image, which will then be arranged in the order of the workstations to form a complete image sequence. If an abnormal size or defect is detected in the glass bracket at a certain workstation, the image acquisition will terminate, retaining only the image data sequence up to that workstation, and the glass bracket will be rejected and will not enter the subsequent process.
[0029] Example 3 This embodiment further discloses a size detection method based on embodiment 2, such as... Figure 3 As shown, it specifically includes the following: This embodiment constructs a high-precision, fully automated visual dimension detection method that is tightly coupled with the production tools of the glass bracket. First, the image data Vi corresponding to any station Gi in the full-process image data sequence of the glass bracket BZ determined in Embodiment 2 is processed as an example. The image data corresponding to the other stations are processed in the same way as the image data Vi.
[0030] Obtain the qualified glass bracket size image database described in Example 1, extract the qualified glass bracket size image data associated with workstation Gi, and record it as qualified glass bracket size image data B_Vi; Then, using the pre-prepared background segmentation algorithm: Otsu's algorithm and the Canny operator, the pixel portion of the glass bracket is directly extracted from the qualified glass bracket size image data B_Vi, and the extracted pixel portion is labeled as B_XVi; Next, repeat the above steps to extract the pixel portion of the glass bracket from the image data Vi, and mark the extracted pixel portion as XVi; Ideally, the pixel portion B_XVi and the pixel portion XVi should match as closely as possible. Based on this principle, the dimensions of the glass bracket are measured through the following steps. Obtain the qualified glass bracket size image data B_Vi again, and use the bottom row of pixels in the qualified glass bracket size image data B_Vi as the horizontal axis of the two-dimensional coordinate system, the leftmost column of pixels as the vertical axis of the two-dimensional coordinate system, and the first pixel in the lower left corner as the origin of the coordinate system to construct a two-dimensional coordinate system. Next, the pixel part B_XVi is marked in the constructed two-dimensional coordinate system, and the geometric center of the pixel part B_XVi is determined and denoted as B_OXVi. Next, determine the geometric center of the pixel portion XVi and mark it as OXVi; Then, place the pixel part XVi in the constructed two-dimensional coordinate system, and make the geometric center OXVi of the pixel part XVi coincide with the geometric center B_OXVi of the pixel part B_XVi, so that the pixel part XVi and the pixel part B_XVi are in the same plane. At this point, pixel part XVi and pixel part B_XVi exist simultaneously in the two-dimensional coordinate system, and the geometric centers of pixel part XVi and pixel part B_XVi overlap. Let pixel part XVi be rotated around the geometric center OXVi as the central axis until it rotates 360 degrees and returns to its original position. During this process, determine the rotation angle at which the pixel region overlap rate between pixel part XVi and pixel part B_XVi is maximized, and adjust pixel part XVi to rotate to this angle. Then extract the pixel region difference rate between pixel part XVi and pixel part B_XVi at this time, where the pixel region difference rate between pixel part XVi and pixel part B_XVi = 100% - pixel region overlap rate.
[0031] In the above steps, by aligning the geometric center OXVi of pixel part XVi with the geometric center B_OXVi of pixel part B_XVi, the comparison error caused by the translation of the pixel part of the glass bracket in the two-dimensional coordinate system is eliminated. Secondly, to prevent comparison errors that may occur on the production line due to shooting errors or positional deviations of the glass holder; Next, the determined pixel region difference rate is compared with the pixel region difference rate threshold preset by the operator. If the pixel area difference rate is greater than the preset pixel area difference rate threshold, the size status of the glass bracket BZ after being processed by station Gi is determined to be an abnormal size status, and the glass bracket BZ is simultaneously rejected. The glass bracket BZ will not participate in the subsequent operation process. If the pixel area difference rate is less than or equal to the preset pixel area difference rate threshold, the glass bracket BZ is determined to be in normal size and can continue to participate in subsequent operation processes.
[0032] The core advantage of the dimension detection method provided in this embodiment lies in its proactive elimination of two common positional errors during image acquisition through geometric center alignment and rotation search. This ensures that the benchmark for dimension comparison is accurate and fair, and the measurement results only reflect the changes in the size and shape of the workpiece itself, rather than the positional changes during the photographing process. This method is more sensitive to the overall deformation of the workpiece and can quickly detect dimensional deviations. It is more comprehensive and reliable than methods that simply measure a few key points or the overall size.
[0033] Example 4 This embodiment, based on embodiment 3, further discloses a method for defect detection of glass supports that have passed dimensional inspection and for generating a visual quality report, such as... Figure 4 As shown, it specifically includes the following: After the glass bracket BZ has undergone the size detection in Example 3 and is determined to be in a normal size state, defect detection is still required. This example provides a defect identification method based on pixel-level difference analysis and regional morphological processing. First, the image data Vi of the glass bracket BZ at station Gi is acquired again, and the pixel part XVi of the glass bracket BZ is extracted from the image data Vi. At this point, the pixel portion XVi and the pixel portion B_XVi of the qualified glass bracket size image data B_Vi are in a state of maximum overlap. Compare the pixel values of the pixel portion XVi and the pixel portion B_XVi at the same coordinate position in the two-dimensional coordinate system, and calculate the pixel value difference rate. In the pixel part of XVi, pixels whose absolute value of all pixel value difference rate exceeds the first difference threshold preset by the operator are marked as initial difference pixels; The pixels that are adjacent in the two-dimensional coordinate system among the initial difference pixels are aggregated into an independent candidate defect region, and the physical area of each candidate defect region is calculated. The physical area of each candidate defect region is equal to the total number of pixels in the corresponding candidate defect region multiplied by the actual physical area of a single pixel. The actual physical area of a single pixel is obtained based on the pixel ratio of the pixel part to the glass bracket of the actual object being 1:1.
[0034] Extract the actual physical area of any candidate defect region. If the actual physical area of the candidate defect region in any pixel part XVi is greater than the first area threshold preset by the operator, then it is determined that there is a defect region in the image data Vi, and the defect state of the glass bracket BZ corresponding to the image data Vi is determined to be an abnormal defect state. If it is determined that there is no candidate defect area in pixel part XVi whose actual physical area is greater than the first area threshold preset by the operator, then the glass bracket BZ is determined not to be an abnormal defect state. Once the glass bracket BZ has passed all the processing stations and finally passed all the dimensional and defect inspections, the glass bracket BZ is considered to have been successfully produced; otherwise, a corresponding visual quality report will be generated. The visualization quality report is divided into two types: one is a visualization quality report for dimensional inspection, and the other is a visualization quality report for defect detection. The specific steps for generating a visual quality report for dimensional inspection are as follows: If the glass bracket BZ at any workstation Gi is determined to be in an abnormal size state, then the glass bracket BZ, workstation Gi, size detection result, and size detection time are extracted and combined with the image data Vi of the glass bracket BZ at workstation Gi to generate a visual quality report associated with the glass bracket BZ. In this report, the size detection result is equal to the abnormal size state of the glass bracket BZ. The specific steps for generating a visual quality report for defect detection are as follows: If a glass bracket BZ at any workstation Gi is determined to be in an abnormal defect state, then the glass bracket BZ, workstation Gi, defect detection result, and defect detection occurrence time are extracted and combined with the image data Vi of the glass bracket BZ at workstation Gi to generate a visual quality report. In this report, the defect detection result is equal to the abnormal defect state of the glass bracket BZ.
[0035] All data in the formulas described above are numerical calculations performed with dimensions removed. Furthermore, any content not described in detail in this specification is existing technology known to those skilled in the art.
[0036] The above description is merely an example and illustration of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the invention or exceed the scope defined in the claims, all of which should fall within the protection scope of the present invention.
[0037] It should be stated that all user data collected in this application was collected with the user's consent and authorization. Furthermore, the uses of user data are legal and compliant, and the use and processing of user data comply with the relevant laws, regulations, and standards of the relevant regions.
Claims
1. A method for full-process visual monitoring and intelligent management of a glass support production line, characterized in that, The method includes: Step 1: Obtain the glass bracket production line and monitor the entire process of the glass bracket production line using a pre-built image acquisition unit. Continuously acquire image data of the glass brackets at any station in the glass bracket production line and sort them according to the entire process of the glass bracket production line to construct the full process image data sequence associated with the corresponding glass bracket. Step 2: Extract image data of the glass bracket at any workstation based on the full-process image data sequence, and perform size detection in conjunction with the pre-constructed qualified glass bracket size image database to determine whether the size status of the glass bracket belongs to an abnormal size status. Step 3: Extract image data of the glass bracket in normal size at the corresponding workstation. Based on the pixel value of the image data, evaluate whether there is a defect area in the image data. Based on the defect area, perform defect detection on the glass bracket to determine whether the defect state of the glass bracket belongs to an abnormal defect state. Step four involves summarizing the glass supports assessed as having abnormal dimensions or defects, along with their associated workstations, defect detection results, dimension detection results, and the time of occurrence, to generate a visual quality report.
2. The method according to claim 1, characterized in that, In step one, the image acquisition unit includes a visible light imaging component and a data transmission component, covering all workstations of the glass bracket production line; The visible light imaging component is deployed at the workstation for inspecting the appearance of the glass bracket surface, and is used to acquire image data under preset lighting conditions. The data transmission component is used to transmit image data to the computing processing unit; The pixel size of the glass bracket captured in the image data is in a 1:1 ratio with the actual object, and the pixel size of the image data is uniform. The pixel values of the background captured by the visible light imaging component are significantly different from those of the glass support.
3. The method according to claim 1, characterized in that, In step one, the specific method for constructing the full-process image data sequence associated with the corresponding glass support is as follows: Obtain all workstations in the glass support production line according to the order of the glass support production line, and arrange them as workstation sequence G1, G2, ..., Gj, where j is the total number of workstations; Obtain any glass bracket, denoted as BZ; The process of glass support BZ from the first station G1 to the last station Gj in the station sequence G1, G2, ..., Gj is determined. The j images captured by the image acquisition unit are sorted according to the station sequence G1, G2, ..., Gj to obtain the full process image data sequence V1, V2, ..., Vj, where V1 to Vj represent the first to j-th image data, and Gi corresponds to Vi, where i is the counting index with a value range of 1 to j.
4. The method according to claim 3, characterized in that, In step one, the total number of image data in the full-process image data sequence of the glass bracket BZ is j if and only if the glass bracket BZ has undergone dimensional inspection and defect inspection throughout the entire glass bracket production line process and has been determined to be in a non-abnormal defect state and a normal dimensional state; Conversely, the total number of image data in the entire process image data sequence of the glass bracket BZ is equal to the value of the station index where the glass bracket BZ is determined to be in an abnormal defect state or an abnormal size state.
5. The method according to claim 4, characterized in that, In step two, the specific method for determining whether the dimensional state of the glass bracket is abnormal is as follows: Obtain any image data Vi and its corresponding workstation Gi from the full-process image data sequence of the glass bracket BZ; Extract the qualified glass bracket size image data B_Vi associated with the corresponding workstation Gi from the pre-built qualified glass bracket size image database; Based on background segmentation, the pixel portion of the glass support in the qualified glass support size image data B_Vi is extracted and denoted as B_XVi. Similarly, the pixel portion of the glass support in the image data Vi is extracted and denoted as XVi. A two-dimensional coordinate system is constructed using the bottom row of pixels in the qualified glass bracket size image data B_Vi as the horizontal axis and the leftmost column of pixels as the vertical axis. Label the pixel portion B_XVi in a two-dimensional coordinate system and determine the geometric center of the pixel portion B_XVi, denoted as B_OXVi; Determine the geometric center OXVi of the pixel portion XVi; Let the geometric center OXVi coincide with the geometric center B_OXVi, and plot the pixel part XVi in the two-dimensional coordinate system. Let the pixel part XVi rotate about the geometric center OXVi as the axis. The rotation is stopped if and only if the pixel region overlap rate between pixel region XVi and pixel region B_XVi is determined to be the maximum, and the pixel region difference rate between pixel region XVi and pixel region B_XVi at this time is extracted, where the pixel region difference rate = 100% - pixel region overlap rate. Compare the pixel region difference rate with the preset pixel region difference rate threshold; If the pixel area difference rate is greater than the pixel area difference rate threshold, the size status of the glass bracket BZ after processing by station Gi is determined to be an abnormal size status, and the glass bracket BZ is rejected and not included in the subsequent process. Conversely, if the glass bracket BZ is found to be of normal size, it will continue to participate in subsequent processes.
6. The method according to claim 5, characterized in that, In step two, the database of images of qualified glass bracket dimensions is pre-built by the operator. Any record in the qualified glass bracket size image database includes the corresponding workstation and the qualified glass bracket size image data associated with that workstation.
7. The method according to claim 5, characterized in that, In step three, the specific method for determining whether the defect state of the glass bracket belongs to an abnormal defect state is as follows: Obtain image data Vi of the glass bracket BZ, which is determined to be of normal size, at station Gi, and extract the pixel portion XVi of image data Vi; Compare the pixel values of pixel part XVi and pixel part B_XVi at the same coordinate position, determine the pixel value difference rate, and mark all pixels in pixel part XVi whose absolute value of pixel value difference exceeds the preset first difference threshold as initial difference pixels. Adjacent pixels in the initial difference pixels are aggregated into an independent candidate defect region. The physical area of each candidate defect region is calculated, where the physical area is equal to the total number of pixels in the candidate defect region multiplied by the actual physical area of a single pixel. If the actual physical area of the candidate defect region is greater than the preset first area threshold, then it is determined that there is a defect region in the image data Vi, and the defect state of the glass bracket BZ corresponding to the image data Vi is determined to be an abnormal defect state. Conversely, if the image data Vi does not contain a defective area, it is determined that the glass bracket BZ does not belong to an abnormal defect state.
8. The method according to claim 7, characterized in that, In step four, the specific method for generating the visualized quality report is as follows: If the glass bracket BZ at any workstation Gi is determined to be in an abnormal size state, the glass bracket BZ, workstation Gi, the time of occurrence of the size detection, and the result of the size detection are extracted and combined with the image data Vi of the glass bracket BZ at workstation Gi to generate a visual quality report. In this report, the size detection result is equal to the abnormal size state of the glass bracket BZ. If a glass bracket BZ at any workstation Gi is determined to be in an abnormal defect state, the glass bracket BZ, workstation Gi, defect detection result, and defect detection occurrence time are extracted and combined with the image data Vi of the glass bracket BZ at workstation Gi to generate a visual quality report. In this report, the defect detection result is equal to the abnormal defect state of the glass bracket BZ.