A quality monitoring system and method

By integrating 2D vision and 3D sensing technologies and combining multi-frame image feature difference analysis, the problems of inaccurate glue height measurement and unstable edge extraction in traditional 2D vision inspection are solved, realizing comprehensive and high-precision monitoring of glue dispensing quality and reducing the risk of missed detection.

CN123118487APending Publication Date: 2026-11-17BEIJING PURA TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611248323.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-18
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Traditional 2D vision inspection cannot accurately measure the dispensing height, making it difficult to determine whether the amount of adhesive is sufficient. Furthermore, edge extraction is unstable in complex environments, leading to a high risk of missed detections and an inability to identify internal defects in transparent adhesives.

Method used

By combining 2D vision and 3D sensing technologies, height information is acquired through laser displacement sensors, structured light sensors, or binocular vision. Combined with polarized light, infrared imaging, or structured light-assisted imaging, and multi-frame image feature difference analysis, a comprehensive evaluation of dispensing quality can be achieved.

Benefits of technology

It achieves high-precision, all-round monitoring of dispensing quality, avoids missed detections caused by adhesive collapse or backflow, improves the real-time performance and reliability of detection, and adapts to stability in complex environments.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123118487A_ABST
    Figure CN123118487A_ABST
Patent Text Reader

Abstract

The application provides a quality monitoring system and method, the system comprising: an image acquisition device for acquiring a real-time image of a dispensing track on a workpiece; a height measuring device for acquiring height information of a dispensing material; a processing device in communication with the image acquisition device and the height measuring device for detecting dispensing breaks or dispensing gaps, wherein the processing device is configured to extract image features from the real-time image, determine the continuity of the dispensing track, fuse the image features and the height information, determine dispensing quality parameters, and generate dispensing monitoring information according to the dispensing quality parameters. The technical solution of the application can realize comprehensive evaluation of dispensing quality, effectively solve the technical defects that traditional 2D vision cannot detect glue height and cannot determine whether the glue amount is sufficient, and eliminate the risk of missed detection caused by glue collapse or backflow.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of machine vision inspection technology, specifically to a quality monitoring system and method. Background Technology

[0002] In modern precision manufacturing and automated assembly, dispensing processes are widely used in critical stages such as electronic packaging, automotive component sealing, and structural bonding. To ensure product reliability and yield, real-time and accurate online inspection of dispensing quality is crucial. Currently, the mainstream inspection methods in industry primarily rely on 2D camera-based machine vision technology. These systems capture images of the dispensing trajectory, use image processing algorithms to extract the width, area, and breakpoint information of the glue line, and compare it with a preset standard template (Golden Sample) to determine whether defects such as insufficient glue, missing glue, or broken glue exist. However, with the increasing complexity of industrial products, especially in scenarios with complex curved surfaces and high precision requirements, such as automotive components, traditional 2D vision inspection is gradually revealing its inherent limitations.

[0003] First, 2D vision can only acquire planar image information and cannot directly measure the height (Z-axis) and volume of the adhesive application. In actual production, adhesives often have fluid properties, and after application, they are prone to collapse, backflow, or stringing, causing the adhesive morphology to change in a very short time. Simple planar area detection cannot reflect the actual filling height of the adhesive, making it difficult to determine whether the adhesive is sufficient to meet the strength requirements for sealing or bonding. Second, in complex working environments, such as strong reflections from metal surfaces, low contrast from black plastic backgrounds, and oil interference, the edge extraction algorithms of traditional 2D cameras often fail, making it difficult to achieve a clear black-and-white separation between the adhesive line and the background. In addition, for transparent or semi-transparent adhesive materials (such as UV adhesives and conductive silver paste), light penetration causes edge blurring, further increasing the difficulty of 2D inspection.

[0004] Therefore, a technical solution is needed that can achieve a comprehensive evaluation of dispensing quality, effectively solving the technical defects of traditional 2D vision in that it cannot detect the glue height or determine whether the glue amount is sufficient, and eliminating the risk of missed detection caused by glue collapse or backflow. Summary of the Invention

[0005] This application aims to provide a quality monitoring system and method that can achieve a comprehensive evaluation of dispensing quality, effectively solving the technical defects of traditional 2D vision in that it cannot detect the glue height or determine whether the glue amount is sufficient, and eliminating the risk of missed detection caused by glue collapse or backflow.

[0006] According to one aspect of this application, a quality monitoring system is provided, comprising: The image acquisition module is used to acquire real-time images of the glue dispensing trajectory on the workpiece; The height measurement module is used to obtain the height information of the dispensing material; The processing module, which is communicatively connected to the image acquisition module and the height measurement module, is used to detect adhesive breaks or gaps in the adhesive coating. The processing module is configured as follows: Image features are extracted from the real-time image, and the image features include at least one of width, brightness distribution, contrast, or texture features. The continuity state of the dispensing trajectory is determined based on inter-frame feature differences across multiple frames. By fusing the image features and the height information, dispensing quality parameters are determined. These dispensing quality parameters include at least one of width consistency, height consistency, or continuity. Dispensing monitoring information is generated based on the dispensing quality parameters.

[0007] According to some embodiments, the height measurement module includes one or more of a laser displacement sensor, a structured light sensor, and a binocular vision sensor, so that the height measurement module acquires the height data of the dispensing section through laser triangulation or structured light imaging.

[0008] According to some embodiments, the image acquisition module includes a 2D camera, which separates the edge of the adhesive line from the product background using an edge recognition algorithm.

[0009] According to some embodiments, the processing module is configured as follows: Receive a real-time image of the dispensing trajectory; Identify adhesive pathway regions in images using a speckle detection algorithm; The coverage or continuity of the adhesive path area is calculated. If the detected adhesive break length is not less than a preset threshold, or the adhesive gap exceeds the allowable range, it is judged as unqualified.

[0010] According to some embodiments, the processing module is configured as follows: For different areas with uneven thickness in the dispensing trajectory, corresponding acceptance criteria are preset; Different detection procedures and judgment criteria are invoked according to the area number, and different error ranges are used to match the thick and thin adhesive lines.

[0011] According to some embodiments, the processing module is configured as follows: The continuity of multi-frame trajectories is analyzed using a temporal model, which includes an RNN or a temporal convolutional network, to distinguish between broken and fine adhesive based on temporal stability.

[0012] According to some embodiments, the image acquisition module further includes a light source component, which includes at least one of ring light, side light and coaxial light, and dynamically switches the light source mode according to the reflective or semi-transparent properties of the adhesive material.

[0013] According to some embodiments, the processing module is further configured to: Acquire images of the Mark points on the workpiece, and correct for changes in the product position based on these Mark points to perform product positioning, ensuring the accuracy of the detection of the relative position of the adhesive line and the product; and / or It integrates the width / line break detection results from 2D vision with the height / volume consistency detection results from 3D vision, and outputs a comprehensive dispensing quality report including width, height and volume.

[0014] According to another aspect of this application, a quality monitoring method is provided, comprising: Acquire real-time images of the adhesive dispensing trajectory on the workpiece; Obtain the height information of the dispensing material; Detect adhesive breaks or gaps in the adhesive application, extract image features from the real-time image, and determine the continuity status of the adhesive dispensing trajectory based on inter-frame feature differences across multiple frames. By fusing the image features and the height information, dispensing quality parameters are determined, and dispensing monitoring information is generated based on these parameters. The image features include at least one of width, brightness distribution, contrast, or texture features, and the dispensing quality parameters include at least one of width consistency, height consistency, or continuity.

[0015] According to some embodiments, the quality monitoring method further includes: For transparent or translucent colloids, polarized light, infrared imaging, or structured light are used for auxiliary imaging.

[0016] According to another aspect of this application, a computing device is provided, comprising: Processor; and A memory storing a computer program that, when executed by the processor, causes the processor to perform the method described in any of the preceding methods.

[0017] According to another aspect of this application, a non-transitory computer-readable storage medium is provided, having stored thereon computer-readable instructions that, when executed by a processor, cause the processor to perform the method as described in any of the preceding claims.

[0018] According to the embodiments of this application, by integrating 2D vision and 3D sensing technologies and combining intelligent image processing and time-series analysis algorithms, the inherent defects of traditional dispensing detection technology are effectively overcome, and comprehensive and high-precision monitoring of dispensing quality is achieved.

[0019] It should be understood that the above general description and the following detailed description are merely exemplary and do not limit this application. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below.

[0021] Figure 1 A diagram of a quality monitoring system apparatus according to an example embodiment is shown.

[0022] Figure 2 A flowchart of a quality monitoring method according to an example embodiment is shown.

[0023] Figure 3 A block diagram of a computing device according to an exemplary embodiment is shown. Detailed Implementation

[0024] 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 this application will be thorough and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted.

[0025] Furthermore, the described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. Numerous specific details are provided in the following description to give a thorough understanding of embodiments of this application. However, those skilled in the art will recognize that the technical solutions of this application can be practiced without one or more of the specific details, or other methods, components, apparatuses, steps, etc., can be employed. In other instances, well-known methods, apparatuses, implementations, or operations are not shown or described in detail to avoid obscuring various aspects of this application.

[0026] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0027] The flowcharts shown in the accompanying drawings are merely illustrative and do not necessarily include all content and operations / steps, nor do they necessarily need to be performed in the described order. For example, some operations / steps can be broken down, while others can be combined or partially combined; therefore, the actual execution order may change depending on the specific circumstances.

[0028] It should be understood that although the terms first, second, third, etc., may be used herein to describe various components, these components should not be limited by these terms. These terms are used to distinguish one component from another. Therefore, the first component discussed below may be referred to as the second component without departing from the teachings of this application. As used herein, the term "and / or" includes all combinations of any one and more of the associated listed items.

[0029] The user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0030] Those skilled in the art will understand that the accompanying drawings are merely schematic diagrams of exemplary embodiments, and the modules or processes in the drawings are not necessarily necessary for implementing this application, and therefore cannot be used to limit the scope of protection of this application.

[0031] In the fields of automated manufacturing and precision assembly, dispensing technology is widely used in key processes such as electronic packaging, automotive component sealing, and new energy battery assembly. With increasingly stringent product quality requirements, higher demands are being placed on the accuracy, stability, and intelligence level of online dispensing quality monitoring.

[0032] Traditional dispensing quality inspection mainly relies on manual visual inspection or automated vision systems based on 2D image processing. It determines the quality of the coating by analyzing two-dimensional geometric features such as the width, length, and position of the adhesive line. However, existing technologies face several technical bottlenecks in practical applications. First, accurate measurement of the adhesive height (Z-axis) is significantly difficult. Traditional 2D cameras can only acquire planar information, making it difficult to reflect the actual filling thickness of the adhesive. Although 3D imaging technologies such as laser displacement sensors, structured light, or binocular vision are used for height reconstruction, dispensing materials often have strong reflective and semi-transparent properties, and the workpiece surface may have complex curvatures, leading to distortion of 3D measurement data and unstable edge extraction, severely affecting the accuracy of height detection. Second, as a fluid, the adhesive continues to undergo physical changes after dispensing, such as sagging, reflow, and stringing. This means that an adhesive line that has just passed inspection may deform before curing, resulting in substandard final quality. Existing systems often lack the ability to compensate for and predict this dynamic evolution process. Secondly, in high-speed production line scenarios, conveyor belt speeds can reach 0.5–2 m / s, requiring vision systems with extremely high real-time performance, and camera frame rates exceeding 500 fps or even higher. Under these conditions, complex image processing algorithms or deep learning models are prone to computational bottlenecks, leading to processing delays and frame drops, making it difficult to meet the real-time requirements of online inspection. Furthermore, the complex industrial environment, with workpieces often exhibiting highly reflective metallic surfaces, low-contrast areas in black plastic, or oil contamination, renders traditional image segmentation methods based on fixed thresholds ineffective, making it difficult to reliably extract the edges of the adhesive. Simultaneously, the system struggles to effectively distinguish between "broken adhesive" and "extremely fine adhesive lines" in continuity judgments, lacking the ability to dynamically correlate time-series image sequences, leading to misjudgments. Moreover, existing surface vision inspection technologies cannot identify internal defects in the adhesive, such as bubbles, voids, and delamination—hidden quality issues—resulting in the risk of missed detections of "OK surface, defective internal structure." Meanwhile, the adhesives used in different processes vary significantly—such as sealants (high viscosity), UV adhesives (high transparency), and conductive adhesives (silver paste reflective), etc. Their optical properties and rheological behaviors are different, making it difficult for a single detection algorithm to be compatible with multiple dispensing processes. This results in poor versatility and requires frequent debugging or even redevelopment.

[0033] In summary, existing dispensing quality inspection technologies still have significant shortcomings in terms of high measurement accuracy, fluid dynamic adaptability, high-speed real-time processing, robustness against complex backgrounds, internal defect identification, and multi-process compatibility. As the complexity of industrial products increases, especially for scenarios with complex curved surfaces and high precision requirements such as automotive parts, traditional 2D vision inspection is gradually revealing its inherent limitations.

[0034] To address these issues, introducing 3D measurement technologies (such as laser displacement sensors or structured light) to acquire height information has become a trend. However, in practical applications, the reflective and semi-transparent properties of dispensing materials make laser stripe extraction difficult, and surface curvature leads to large measurement errors and unstable edge recognition. Simultaneously, high-speed production line operation (speeds of 0.5–2 m / s) demands extremely high real-time performance from the detection system (frame rate > 500 fps), posing a significant challenge to data processing capabilities. Therefore, how to integrate the planar features of 2D vision with the height information from 3D measurement, and combine this with time-series analysis to overcome fluid deformation and complex background interference, to achieve comprehensive and high-precision monitoring of the width, height, volume, and continuity of the dispensing trajectory, is a pressing technical challenge that needs to be solved.

[0035] To address this, this application proposes a quality monitoring system and method that enables comprehensive evaluation of dispensing quality. This effectively solves the technical shortcomings of traditional 2D vision, which cannot detect adhesive height or determine sufficiency, and eliminates the risk of missed detections due to adhesive collapse or backflow. According to embodiments, this invention employs multimodal fusion technology combining real-time images and height information, integrating two-dimensional visual features with three-dimensional morphology data. This allows for a comprehensive evaluation of the planar morphology and three-dimensional filling state of the adhesive line, thereby accurately determining dispensing quality parameters. Furthermore, by dynamically analyzing the continuity of the dispensing trajectory using inter-frame feature difference technology across multiple frames, it can identify defects such as adhesive breaks and gaps in the dispensing process in real time. This achieves high-sensitivity detection of the adhesive path continuity, effectively avoiding the problem of easily missing tiny breakpoints in traditional static image comparison, and improving the real-time performance and reliability of defect detection.

[0036] According to some embodiments, by combining polarized light, infrared imaging, or structured light-assisted imaging technologies with conventional industrial cameras, it is possible to effectively suppress strong reflections on the surface of transparent or translucent colloids and enhance edge visibility in low-contrast areas. This solves the technical problems of blurred imaging and unstable edge recognition in traditional vision systems when processing transparent colloids, ensuring the image acquisition quality under complex materials. It forms a complete technology chain from high-quality imaging → continuous detection → multi-dimensional parameter fusion analysis, which not only realizes comprehensive and high-precision online monitoring of the dispensing process, but also provides reliable data support for closed-loop control of the dispensing process, and has good industrial application value and promotion prospects.

[0037] The following description, in conjunction with the accompanying drawings, illustrates exemplary embodiments of this application.

[0038] Figure 1 A diagram of a quality monitoring system apparatus according to an example embodiment is shown.

[0039] See Figure 1The figure shows a quality monitoring system, including: an image acquisition module 01, a height measurement module 02, and a processing module 03. By using the fusion processing of 2D and 3D data, it can not only calculate the planar area of ​​the adhesive line, but also directly obtain the height and volume parameters of the adhesive. This solves the technical problem that traditional 2D vision cannot detect the height of the adhesive and it is difficult to judge whether the amount of adhesive is sufficient. It effectively avoids the risk of missed detection caused by adhesive collapse, backflow or internal voids, and ensures the reliability of dispensing quality.

[0040] According to some embodiments, the image acquisition module 01 is used to acquire real-time images of the dispensing trajectory on the workpiece. As the basic sensing unit of the system, the image acquisition module 01 is mainly used to acquire real-time image information of the dispensing trajectory on the workpiece surface. Using a high-resolution industrial camera in conjunction with a specialized light source system, continuous images are taken of the glue line during or after the dispensing operation to capture the morphological characteristics of the glue on a two-dimensional plane. The acquired image data not only includes the contour, width, and continuity information of the glue line, but also reflects the grayscale or color differences between the glue and the workpiece background, providing original visual evidence for subsequent image processing and defect identification.

[0041] According to some embodiments, by optimizing optical imaging parameters and synchronous triggering mechanisms, this module can stably acquire clear, distortion-free images in dynamic production environments, ensuring the real-time performance and accuracy of the detection system.

[0042] According to some embodiments, the image acquisition module 01 includes a 2D camera, which separates the adhesive line edge from the product background through an edge recognition algorithm. The core component of the image acquisition module 01 is a 2D industrial camera, which performs high-frame-rate imaging of the adhesive application area on the workpiece surface through an industrial lens to obtain a two-dimensional grayscale or color image of the adhesive line trajectory on a horizontal plane. After acquiring the raw image data, the processing module 03 runs a preset edge recognition algorithm (such as the gradient-based Sobel operator, Canny algorithm, or template matching algorithm) to extract features and segment the image. Specifically, the algorithm accurately locates the adhesive line edge pixels by calculating the gradient changes of pixel grayscale values ​​in local areas of the image, thereby separating the continuously changing grayscale adhesive line entity from the complex workpiece background and forming a clear outline line. This effectively solves the problem that traditional manual visual inspection or simple threshold segmentation is easily interfered with by background texture and uneven lighting, providing an accurate geometric reference for subsequent judgment of adhesive line width, trajectory offset, and adhesive breakage connectivity.

[0043] According to some embodiments, the height measurement module 02 is used to acquire the height information of the dispensing material. The height measurement module 02 is specifically designed to acquire the height information of the dispensing material in the vertical direction in real time to address dispensing quality defects caused by changes in the surface morphology of the workpiece (such as substrate warping or deformation). Through non-contact sensing technology, it accurately captures the displacement data of the adhesive surface relative to a reference plane, ensuring high-precision measurement stability even in complex production environments.

[0044] According to some embodiments, the height measurement module 02 includes one or more of a laser displacement sensor, a structured light sensor, and a binocular vision sensor, thereby acquiring the height data of the dispensing cross-section through laser triangulation or structured light imaging. In a specific embodiment, the height measurement module 02 preferably uses a laser displacement sensor, which operates based on the principle of laser triangulation. The laser emitting unit inside the sensor emits a highly focused laser beam toward the workpiece surface, and the reflected light is converged onto a photodetector (such as a CCD or PSD) array via an optical lens system. When the height of the adhesive changes, the position of the reflected light spot on the photosensitive surface of the detector will shift linearly. Through a pre-calibrated geometric triangulation relationship, the light spot displacement can be converted into a precise height value. Compared with traditional local height measurement methods, this system can perform continuous or dense sampling measurements along the entire dispensing path, effectively identifying minute height differences caused by substrate warping (such as deformation of up to 0.5 mm in aluminum alloy substrates during processing), thereby avoiding quality problems such as needle collisions, adhesive breakage, or scratches on the coating. Furthermore, to accommodate the measurement needs of colloids made of different materials (such as highly reflective, light-absorbing, or transparent materials), structured light sensors or binocular vision sensors can also be integrated. Structured light sensors reconstruct the three-dimensional morphology by projecting a specific coded pattern and analyzing its deformation on the colloid surface; binocular vision, on the other hand, calculates depth information using the principle of parallax. These sensors work in conjunction with the image acquisition module 01, precisely matching the acquired height data with the 2D image through spatiotemporal synchronization and coordinate system calibration. This provides crucial quantitative data for the subsequent processing module 03 to calculate the colloid volume, cross-sectional area, and determine the consistency of the colloid shape, thereby achieving comprehensive and highly reliable monitoring of dispensing quality.

[0045] According to some embodiments, the image acquisition module 01 further includes a light source component, which includes at least one of ring light, side light, and coaxial light, and dynamically switches the light source mode according to the reflective or translucent characteristics of the adhesive material. The image acquisition module 01 further integrates a highly adaptable light source component, designed to overcome the influence of the diverse optical properties of the adhesive material on image quality by optimizing the illumination strategy. It can dynamically switch the most suitable light source mode according to the specific reflective characteristics (such as specular reflection, diffuse reflection) or translucent characteristics of the adhesive material, thereby ensuring the acquisition of clear images with high contrast and low noise. In specific application scenarios, when the object being inspected is a highly reflective material (such as glossy adhesive on a metal substrate), the system can dynamically switch to sidelight mode, using low-angle illumination to highlight the contour shadows of the adhesive line edges and enhance the grayscale difference between the adhesive and the background. For semi-transparent or transparent adhesives (such as some silicone or UV adhesives), coaxial illumination can be enabled. Through the design of vertical incident and vertical reflection light paths, surface glare is effectively eliminated and the transmission clarity of the internal structure is improved. In conventional inspection scenarios, ring light provides uniform diffuse reflection illumination, ensuring the uniform presentation of the overall shape of the adhesive line. This dynamic switching mechanism of the light source based on material characteristics enables the system to adapt to different production environments and adhesive types, significantly improving the stability of image acquisition and the accuracy of subsequent edge recognition algorithms.

[0046] According to some embodiments, the processing module 03 is communicatively connected to the image acquisition module 01 and the height measurement module 02, and is used to detect adhesive breakage or coating gaps. The processing module 03 establishes a high-speed communication connection with the image acquisition module 01 and the height measurement module 02, constituting the core computing unit for dispensing quality detection. Its main function is to accurately identify adhesive breakage defects or coating gaps during the dispensing process through multi-dimensional data fusion analysis.

[0047] According to some embodiments, the processing module 03 is configured to: extract image features from the real-time image, the image features including at least one of width, brightness distribution, contrast, or texture features; determine the continuity state of the dispensing trajectory based on inter-frame feature differences across multiple frames; fuse the image features and the height information to determine dispensing quality parameters, the dispensing quality parameters including at least one of width consistency, height consistency, or continuity; and generate dispensing monitoring information based on the dispensing quality parameters. In specific implementation, the processing module 03 first receives a real-time image stream from the image acquisition module 01, and uses image processing algorithms to extract key image features from it, the image features including at least the width of the adhesive line, brightness distribution, contrast, and texture features. By analyzing these features, the system can preliminarily determine the appearance state of the adhesive.

[0048] To further improve detection accuracy, processing module 03 introduces a time-series analysis mechanism. By performing inter-frame feature difference operations on multiple consecutive images, it analyzes the dynamic changes of the dispensing trajectory during motion, thereby accurately determining the continuity of the dispensing line. For example, when the inter-frame difference result abruptly exceeds a preset threshold, it is determined that there may be a risk of dispensing breakage. More importantly, processing module 03 does not analyze visual data in isolation, but rather fuses the extracted image features with the dispensing height information synchronously fed back by height measurement module 02. By establishing a mathematical model, it comprehensively calculates key quality parameters such as the width consistency, height consistency, and trajectory continuity of the dispensing. Finally, the system generates detailed dispensing monitoring information based on these quantitative parameters, achieving comprehensive closed-loop monitoring of dispensing quality.

[0049] According to some embodiments, the processing module 03 is configured to: receive a real-time image of the dispensing trajectory; identify the adhesive path region in the image using a blob detection algorithm; calculate the coverage or continuity of the adhesive path region; and determine it as unqualified when the detected adhesive break length is not less than a preset threshold, or when the adhesive gap exceeds the allowable range. Specifically, the processing module 03 first acquires a real-time image of the dispensing trajectory captured by the image acquisition module 01. In order to accurately segment the adhesive path morphology from a complex background, the processing module 03 uses a blob detection algorithm to process the image. By performing binarization and connected component analysis on the image, the adhesive path region in the image can be effectively identified and marked, forming a clear adhesive outline.

[0050] After identifying the adhesive path area, processing module 03 performs further quantitative analysis. By calculating the pixel coverage of the adhesive path area in the image or analyzing the continuity of its topological structure, the system can assess the filling status of the adhesive. Processing module 03 has a built-in preset judgment threshold. When the detected adhesive break length is not less than the preset threshold, or the size of the adhesive gap exceeds the allowable range, processing module 03 immediately determines that the current dispensing process is unqualified and generates a corresponding alarm signal or quality mark, thereby achieving rapid response and interception of dispensing defects.

[0051] According to some embodiments, the processing module 03 presets corresponding acceptance criteria for different regions with uneven thickness in the dispensing trajectory; it calls different detection programs and acceptance criteria based on the region number, and uses different error ranges to match thick and thin glue lines. In specific implementation, the system pre-divides the dispensing trajectory into multiple detection regions with different geometric features, and assigns a unique region number and corresponding acceptance criteria to each region. The processing module 03 dynamically calls the matching detection program and acceptance threshold based on the currently analyzed region number. For example, for thick glue line regions, the system uses a more lenient width error range for matching; while for thin glue line regions, it switches to a high-precision detection algorithm and uses a strict small error range for matching, thereby ensuring that glue lines of different specifications can obtain accurate and impartial quality assessment, effectively avoiding misjudgments caused by a single standard detection.

[0052] According to some embodiments, the system further includes a synchronous acquisition mechanism, wherein the image acquisition module 01 and the height measurement module 02 are configured to acquire data coaxially or synchronously, thereby aligning 2D image features with 3D height information in space and time to ensure accurate fusion of multi-dimensional data. Specifically, the image acquisition module 01 and the height measurement module 02 adopt a coaxial optical path design in hardware configuration or implement strict time-series synchronous acquisition in the control system, thereby ensuring that the pixels in each frame of the 2D image can be accurately aligned with the spatial points in the 3D height information in time and spatial coordinates. Through this alignment method, the processing module 03 can perform correlation analysis on the 2D features such as texture and brightness of the adhesive path with the 3D morphological features such as the actual height and volume of the adhesive in the same coordinate system, thereby eliminating data deviations caused by time delays or viewing angle differences, and providing a reliable data foundation for subsequent high-precision determination of dispensing quality.

[0053] According to some embodiments, the processing module 03 utilizes a temporal model to analyze the continuity of multi-frame trajectories. This temporal model includes an RNN or a temporal convolutional network, used to distinguish between broken and thin adhesive based on temporal stability, thus addressing the problem of subtle defects that are difficult to distinguish in traditional static image analysis. Specifically, the processing module 03 inputs the acquired multi-frame continuous images or extracted feature sequences into a pre-trained temporal model, which includes a recurrent neural network (RNN) or a temporal convolutional network (TCN).

[0054] By learning the dynamic changes of the adhesive path over time, the system establishes a recognition of the temporal stability characteristics of the normal adhesive application process. During detection, the model can analyze the evolution trend of the adhesive path morphology between adjacent frames: for normal thin adhesive lines, it exhibits continuous and stable feature output over time; while for adhesive breakage defects, the model captures abrupt changes or interruptions in the feature signal. Through this time-stability-based analysis mechanism, the system can effectively distinguish between low-contrast areas caused by excessively thin adhesive and genuine adhesive breakage areas, significantly reducing the false positive rate caused by excessively thin adhesive lines and improving the robustness of detection.

[0055] According to some embodiments, the processing module 03 also acquires Mark point images on the workpiece and corrects for changes in product position based on the Mark points, thereby performing product positioning and ensuring the detection accuracy of the relative position of the adhesive line and the product; and / or integrates the width / break line detection results of 2D vision with the height / volume consistency detection results of 3D vision, outputting a comprehensive dispensing quality report including width, height, and volume. The processing module 03 is further configured with a high-precision positioning correction function to eliminate positional deviations generated during workpiece transport or clamping. Specifically, the processing module 03 first acquires preset Mark point images on the workpiece surface and accurately calculates the offset and rotation angle between the actual coordinates and theoretical coordinates of the Mark points using an image recognition algorithm. Based on this offset data, the processing module 03 performs real-time correction of product position changes, thereby completing product positioning, ensuring the subsequent detection accuracy of the relative position of the adhesive line and the product, and avoiding misjudgments of the adhesive path position caused by overall workpiece displacement.

[0056] Furthermore, the processing module 03 also possesses multimodal data fusion and analysis capabilities, enabling it to output a comprehensive dispensing quality report. The processing module 03 deeply integrates the width data and line breakage detection results obtained from 2D visual inspection with the height data and volume consistency detection results obtained from 3D visual inspection. By comprehensively analyzing these two types of data, the system can generate a complete quality report containing key indicators such as glue line width, height, and volume, thereby achieving comprehensive quantitative evaluation and traceability of the dispensing process.

[0057] Figure 2 A flowchart of a quality monitoring method according to an example embodiment is shown.

[0058] See Figure 2 The figure shows a quality monitoring method that aims to achieve high-precision online detection of dispensing process quality through multi-dimensional information fusion and dynamic trajectory analysis. The specific steps are as follows.

[0059] In S101, a real-time image of the dispensing trajectory on the workpiece is acquired.

[0060] According to some embodiments, real-time images of the dispensing trajectory on the workpiece are acquired. Specifically, an industrial camera continuously captures visual images of the adhesive path on the workpiece surface during the dispensing process. The industrial camera can employ a CCD or CMOS sensor, coupled with a fixed-focus lens and a dedicated illumination source, and is mounted near the dispensing head or fixed to a motion module, synchronously capturing the forming state of the adhesive on the workpiece during the dispensing process. The acquired real-time images should have sufficient resolution and contrast to clearly present the edge contours and surface features of the adhesive line, providing basic visual data for subsequent image processing. This image sequence is acquired continuously in chronological order, covering the entire dispensing path, ensuring that the dynamic forming process of the adhesive path can be reflected, and supporting real-time analysis of features such as width, shape, and continuity.

[0061] In S103, obtain the height information of the dispensing material.

[0062] According to some embodiments, the height information of the dispensing material can be obtained, optionally by scanning the three-dimensional morphology of the applied area using a non-contact height measurement device. The height measurement device can be a laser displacement sensor, a confocal microscope, or a structured light 3D scanner, mounted on the side of the dispensing head or on an independent motion axis, working in conjunction with an industrial camera to acquire the height distribution data of the adhesive along the Z-axis immediately during or after dispensing. The acquired height information includes the cross-sectional height of the adhesive line, volume accumulation characteristics, and surface undulation changes, reflecting the actual filling thickness and molding consistency of the adhesive. In some embodiments, a three-dimensional height contour map of the adhesive path can be constructed through multi-point sampling or continuous scanning for subsequent fusion analysis with image features, improving the ability to identify three-dimensional defects such as insufficient adhesive, adhesive buildup, and collapse. The height information is synchronized with the real-time image in space and time, ensuring the accuracy and reliability of multimodal data fusion.

[0063] In S105, a break in the glue or a gap in the glue application is detected, image features are extracted from the real-time image, and the continuity state of the glue application trajectory is determined based on the inter-frame feature difference across multiple frames.

[0064] According to some embodiments, detecting adhesive breaks or gaps involves extracting image features from the real-time images and determining the continuity of the dispensing trajectory based on inter-frame feature differences across multiple frames. The image features include width, brightness distribution, contrast or texture features, edge contours of the adhesive line, grayscale distribution, and geometric shape features, which are identified and located in a continuously acquired image sequence using a preset feature extraction algorithm. Further, differential operations are performed on corresponding features between adjacent or multiple frames to calculate the continuity change of the adhesive line in the direction of movement. If the inter-frame feature difference result exceeds a preset threshold, it is determined that an adhesive break or gap exists at that location. By dynamically analyzing the continuity of inter-frame features, defects such as discontinuities, missed coatings, or insufficient adhesive during the dispensing process can be effectively identified, enabling real-time detection and feedback of the dispensing trajectory continuity, thus improving the controllability and reliability of dispensing quality.

[0065] In S107, the image features and the height information are fused to determine the dispensing quality parameters, and dispensing monitoring information is generated based on the dispensing quality parameters.

[0066] According to some embodiments, the image features and height information are fused to determine dispensing quality parameters. Specifically, this includes spatially registering and associating two-dimensional image features such as glue line edge, grayscale, and continuity extracted in S101 with three-dimensional height information such as glue cross-sectional height and volume distribution obtained in S103. A preset fusion algorithm is then used to calculate parameters that comprehensively characterize the dispensing quality. These dispensing quality parameters include glue consistency, glue path formation stability, and the severity of local defects. Further, based on the comparison results between the dispensing quality parameters and preset standard thresholds, dispensing monitoring information containing defect location, type, and level is generated and output through a visual interface or communication interface. This allows the control system to adjust dispensing process parameters in real time or trigger an alarm and glue replenishment mechanism, achieving closed-loop monitoring and continuous optimization of dispensing quality.

[0067] According to some embodiments, for transparent or translucent colloid, polarized light, infrared imaging or structured light is adopted for auxiliary imaging, so as to solve the problem of unstable edge recognition caused by material reflection or low contrast in traditional visual inspection. Specifically in the implementation process, polarized light, infrared imaging or structured light can be used for auxiliary imaging. In the process of image acquisition, by installing a polarizing filter in front of the lens of the industrial camera and using a polarized light source in cooperation, the specular reflected light on the colloid surface can be effectively suppressed, the contrast between the colloid line and the background can be enhanced, so that the edge contour of the colloid can be clearly presented. In addition, infrared imaging technology can be adopted to improve image contrast and realize stable imaging of colloid morphology by utilizing the difference in absorption or scattering characteristics of transparent colloid in a specific infrared band. In some implementations, a structured light projection device can also be introduced to project coded light stripes onto the surface of the colloid, and the three-dimensional morphology of the colloid can be restored by analyzing the deformed stripes, so as to further assist edge positioning and height measurement. The above auxiliary imaging methods can be used alone or in combination, which significantly improves the image acquisition quality of transparent or translucent colloids, and provides a reliable visual basis for subsequent image feature extraction and defect detection.

[0068] Specifically in the implementation scenario, according to some embodiments, the dispensing quality inspection system may be equipped with a two-dimensional industrial camera to vertically collect grayscale images of the dispensing area in view of the high-contrast characteristic between the colloid (e.g., white) and the product substrate (e.g., black). The height data of the dispensing section is collected by one or more of a laser displacement sensor, a structured light sensor, and a binocular vision sensor. The colloid region is extracted through an image segmentation algorithm, its pixel area is calculated, and converted into an actual physical area or equivalent width in combination with a preset pixel equivalent. The measured value is then compared with the standard value and error tolerance preset by the customer. If the measured value is within the tolerance range, it is judged as qualified; otherwise, it is judged as unqualified. After the inspection is completed, it is automatically determined whether glue replenishment is required. If glue replenishment is required, the glue replenishment operation is performed and the inspection is carried out again; if no glue replenishment is required, the product is discharged, and the inspection data is uploaded to the MES system at the same time, before the product is manually taken down.

[0069] According to some embodiments, the technical solution of the present invention adopts the multi-modal fusion technology of real-time images and height information, combines two-dimensional visual features with three-dimensional morphology data, can comprehensively evaluate the planar morphology and three-dimensional filling state of the glue line, so as to accurately determine the dispensing quality parameters, solves the limitation that single visual inspection is easily interfered by ambient light and cannot reflect the colloid thickness, and significantly improves the comprehensiveness and accuracy of dispensing quality judgment.

[0070] Further, dynamic analysis of the continuity of the dispensing trajectory is carried out through the inter-frame feature difference technology based on multiple frames, which can identify defects such as broken glue and gluing gaps in real time during the dispensing process, realize high-sensitivity detection of the continuity state of the glue path, effectively avoid the problem that traditional static image comparison is prone to miss tiny breakpoints, and improve the real-time performance and reliability of defect detection.

[0071] Furthermore, by combining polarized light, infrared imaging, or structured light-assisted imaging technologies with conventional industrial cameras, it can effectively suppress strong reflections on transparent or translucent colloid surfaces and enhance edge visibility in low-contrast areas. This solves the technical problems of blurred imaging and unstable edge recognition in traditional vision systems when processing transparent colloids, ensuring image acquisition quality under complex materials. Thus, a complete technology chain is formed from high-quality imaging → continuous detection → multi-dimensional parameter fusion analysis. It not only realizes comprehensive and high-precision online monitoring of the dispensing process, but also provides reliable data support for closed-loop control of the dispensing process, demonstrating good industrial application value and promising prospects for promotion.

[0072] Figure 3 A block diagram of a computing device according to an example embodiment of this application is shown.

[0073] like Figure 3 As shown, the computing device 30 includes a processor 12 and a memory 14. The computing device 30 may also include a bus 22, a network interface 16, and an I / O interface 18. The processor 12, memory 14, network interface 16, and I / O interface 18 can communicate with each other via the bus 22.

[0074] Processor 12 may include one or more general-purpose CPUs (Central Processing Units), microprocessors, or application-specific integrated circuits, for executing relevant program instructions. According to some embodiments, computing device 30 may also include a high-performance display adapter (GPU) 20 for accelerating processor 12.

[0075] Memory 14 may include a machine-readable medium in the form of volatile memory, such as random access memory (RAM), read-only memory (ROM), and / or cache memory. Memory 14 is used to store one or more programs containing instructions, as well as data. Processor 12 may read the instructions stored in memory 14 to perform the methods described above according to embodiments of this application.

[0076] The computing device 30 can also communicate with one or more networks via the network interface 16. The network interface 16 can be a wireless network interface.

[0077] Bus 22 can include address bus, data bus, control bus, etc. Bus 22 provides a path for exchanging information between components.

[0078] It should be noted that, in specific implementations, the computing device 30 may also include other components necessary for normal operation. Furthermore, those skilled in the art will understand that the device described above may only include the components necessary for implementing the embodiments of this specification, and not necessarily all the components shown in the figures.

[0079] This application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method. The computer-readable storage medium may include, but is not limited to, any type of disk, including floppy disks, optical disks, DVDs, CD-ROMs, microdrives, as well as magneto-optical disks, ROMs, RAMs, EPROMs, EEPROMs, DRAMs, VRAMs, flash memory devices, magnetic cards or optical cards, nanosystems (including molecular memory ICs), network storage devices, cloud storage devices, or any type of medium or device suitable for storing instructions and / or data.

[0080] This application also provides a computer program product including a non-transitory computer-readable storage medium storing a computer program operable to cause a computer to perform some or all of the steps of any of the methods described in the above method embodiments.

[0081] Those skilled in the art will clearly understand that the technical solutions of this application can be implemented using software and / or hardware. In this specification, "unit" and "module" refer to software and / or hardware capable of independently performing or cooperating with other components to perform a specific function, where the hardware may be, for example, a field-programmable gate array (FPGA), integrated circuit, etc.

[0082] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0083] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0084] In the several embodiments provided in this application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some service interface; the indirect coupling or communication connection between apparatuses or units may be electrical or other forms.

[0085] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0086] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0087] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage device. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application.

[0088] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0089] The exemplary embodiments of this application have been specifically shown and described above. It should be understood that this application is not limited to the detailed structures, arrangements, or implementation methods described herein; rather, this application is intended to cover various modifications and equivalent arrangements contained within the spirit and scope of the appended provisions.

Claims

1. A quality monitoring system, characterized in that, include: The image acquisition module is used to acquire real-time images of the glue dispensing trajectory on the workpiece; The height measurement module is used to obtain the height information of the dispensing material; The processing module, which is communicatively connected to the image acquisition module and the height measurement module, is used to detect adhesive breaks or gaps in the adhesive coating. The processing module is configured as follows: Image features are extracted from the real-time image, and the image features include at least one of width, brightness distribution, contrast, or texture features. The continuity state of the dispensing trajectory is determined based on inter-frame feature differences across multiple frames. By fusing the image features and the height information, dispensing quality parameters are determined. These dispensing quality parameters include at least one of width consistency, height consistency, or continuity. Dispensing monitoring information is generated based on the dispensing quality parameters.

2. The quality monitoring system according to claim 1, characterized in that, The height measurement module includes one or more of a laser displacement sensor, a structured light sensor, and a binocular vision sensor, thereby acquiring the height data of the dispensing cross-section through laser triangulation or structured light imaging.

3. The quality monitoring system according to claim 1, characterized in that, The image acquisition module includes a 2D camera, which uses an edge recognition algorithm to separate the edge of the adhesive line from the product background.

4. The quality monitoring system according to claim 1, characterized in that, The processing module is configured as follows: Receive a real-time image of the dispensing trajectory; Identify adhesive path regions in images using a speckle detection algorithm; The coverage or continuity of the adhesive application area is calculated. If the detected adhesive break length is not less than a preset threshold, or the adhesive gap exceeds the allowable range, it is judged as unqualified.

5. The quality monitoring system according to claim 1, characterized in that, The processing module is configured as follows: For different areas with uneven thickness in the dispensing trajectory, corresponding acceptance criteria are preset; Different detection procedures and judgment criteria are invoked according to the area number, and different error ranges are used to match the thick and thin adhesive lines.

6. The quality monitoring system according to claim 1, characterized in that, The processing module is configured as follows: The continuity of multi-frame trajectories is analyzed using a temporal model, which includes an RNN or a temporal convolutional network, to distinguish between broken and fine adhesive based on temporal stability.

7. The quality monitoring system according to claim 1, characterized in that, The image acquisition module further includes a light source component, which includes at least one of ring light, side light and coaxial light, and dynamically switches the light source mode according to the reflective or semi-transparent properties of the adhesive material.

8. The quality monitoring system according to claim 1, characterized in that, The processing module is further configured to: Acquire images of the Mark points on the workpiece, and correct for changes in the product position based on these Mark points to perform product positioning, ensuring the accuracy of the detection of the relative position of the adhesive line and the product; and / or It integrates the width / line break detection results from 2D vision with the height / volume consistency detection results from 3D vision, and outputs a comprehensive dispensing quality report including width, height and volume.

9. A quality monitoring method, characterized in that, include: Acquire real-time images of the adhesive dispensing trajectory on the workpiece; Obtain the height information of the dispensing material; Detect adhesive breaks or gaps in the adhesive application, extract image features from the real-time image, and determine the continuity status of the adhesive dispensing trajectory based on inter-frame feature differences across multiple frames. By fusing the image features and the height information, dispensing quality parameters are determined, and dispensing monitoring information is generated based on these parameters. The image features include at least one of width, brightness distribution, contrast, or texture features, and the dispensing quality parameters include at least one of width consistency, height consistency, or continuity.

10. The quality monitoring method according to claim 9, characterized in that, The method further includes: For transparent or translucent colloids, polarized light, infrared imaging, or structured light are used for auxiliary imaging.