Circuit board production inspection method, system, equipment and medium based on machine vision

Abstract

The present invention discloses a circuit board production detection method, system, equipment and medium based on machine vision, which relates to the technical field of circuit board production. Through an image acquisition module with multi-angle and multi-light source illumination, combined with the spatial precision alignment function of an image preprocessing module, and the 3D model construction and defect positioning capabilities of a reconstruction and adjustment module, the accuracy and efficiency of defect positioning of multi-layer circuit boards are significantly improved; the system can automatically and quickly detect and evaluate appearance and functional defects, and timely judge the qualification of circuit boards, thereby greatly shortening the detection time and improving the overall efficiency of the production line; the modular design reduces the dependence on additional equipment, and reduces the need for manual intervention through automated defect analysis, further reducing equipment and maintenance costs; this efficient, accurate and economical solution is particularly suitable for rapid detection and quality control of multi-layer circuit boards in large-scale production environments.

Description

Technical Field

[0001] The present invention relates to the technical field of circuit board production, and specifically to a circuit board production detection method, system, equipment and medium based on machine vision. Background Art

[0002] Circuit boards are the core components of modern electronic devices, mainly used to support and connect electronic components. According to their structure and purpose, they can be divided into single-sided boards, double-sided boards, and multi-layer boards. They are widely used in various electronic products. However, with the increase in circuit integration and complexity, traditional manufacturing and testing technologies face challenges in handling high-density, tiny components and ensuring product reliability.

[0003] Especially for multilayer boards used in high-density, high-performance equipment, such as communication equipment, servers, and medical electronic equipment, multilayer boards have three or more conductive layers; but due to the complex structure of multilayer boards, there are the following technical disadvantages during production and testing:

[0004] 1. Difficulty in defect location: The internal structure of a multi-layer circuit board is complex. Defects such as short circuits and open circuits may be hidden between multiple conductive layers. Traditional detection methods are difficult to accurately locate problem areas.

[0005] 2. Slow detection speed: The complexity of multi-layer circuit boards requires detection equipment to scan and analyze layer by layer, resulting in a decrease in the overall detection speed and affecting production efficiency. Summary of the invention

[0006] Technical issues solved

[0007] In view of the deficiencies in the prior art, the present invention provides a circuit board production inspection method, system, equipment and medium based on machine vision, which solves the technical problems of high difficulty in defect positioning and slow inspection speed in the background technology.

[0008] Technical Solution

[0009] To achieve the above objectives, the present invention is implemented through the following technical solutions: a circuit board production inspection system based on machine vision, including an image acquisition module, an image preprocessing module, an appearance defect recognition module, a functional defect recognition module and a reconstruction adjustment module;

[0010] The image acquisition module is used to acquire circuit board image data from multiple angles; by deploying a multi-light source lighting device, images of the surface layer, conductive layer and internal layer of the multi-layer circuit board are respectively acquired;

[0011] The image preprocessing module is used to preprocess the image data acquired by the image acquisition module; and is further configured to perform image registration operations for precise spatial alignment of the images;

[0012] The appearance defect recognition module is used to analyze the preprocessed image data and identify the appearance defects of the multilayer circuit board, including the scratch area, stain area, number of disconnections and number of marking errors; and calculate and evaluate the appearance defect index Wgqx of the multilayer circuit board;

[0013] The functional defect recognition module is used to analyze the preprocessed image data again to identify the functional defects in the circuit board; by calculating and evaluating the functional defect index Gnqx, and then further analyzing the functional defects of the multilayer circuit board according to the evaluation results;

[0014] The reconstruction and adjustment module is used to construct a 3D model based on the collected multi-layer image data, and then three-dimensionally locate the detected defects and mark the level and position of the defects; at the same time, it records and stores the relevant data of the appearance defect index Wgqx and the functional defect index Gnqx, and feeds back to the production line.

[0015] Preferably, the image acquisition module includes a device deployment unit and an image acquisition unit; the device deployment unit is used to deploy a high-resolution camera, and is provided with a plurality of light source illumination devices, and is deployed at different positions within the detection area; in the process of collecting images, by adjusting the intensity, angle and irradiation direction of the light source, the images of the surface layer and the internal conductive layer of the multi-layer circuit board are collected respectively;

[0016] The image acquisition unit acquires image data of the surface layer of the circuit board by vertical illumination, and then acquires images of the multi-layer surface layer and the internal conductive layer by multi-angle illumination, including oblique illumination of 30 degrees, 45 degrees and 60 degrees; by oblique illumination, edge images of the electrical layer, hole images and detailed feature images of the connection points between layers are collected; in addition, the image acquisition unit also uses transmitted light source technology to perform penetrating illumination on the circuit board to acquire image data of its internal conductive layer.

[0017] Preferably, the image preprocessing module is used to perform denoising and contrast enhancement on the image, and then use an edge detection algorithm to extract edge information of the structure in the multilayer circuit board, including wires, solder joints and apertures, and then perform geometric correction on the image;

[0018] After completing the above processing, the same feature points in different images are identified through the feature point matching algorithm, and the images of the multi-layer circuit board are accurately aligned in space using the affine transformation and perspective transformation methods. The image registration process includes aligning the conductive paths and welding points in each layer of the image, which is used to consistently superimpose images of different levels in the same coordinate system.

[0019] Preferably, the appearance defect recognition module includes an appearance recognition unit and an appearance evaluation unit; the appearance recognition unit first processes the circuit board surface image by edge detection and morphological analysis technology, recognizes the outline of the scratch, and then uses an area calculation algorithm to obtain and record the scratch area Gmj;

[0020] Then, the stain area on the circuit board surface is detected through color space conversion and cluster analysis technology. The region growing algorithm is used to distinguish the color difference between normal materials and abnormal stains, and the stain area Wmj is calculated;

[0021] Secondly, the connectivity analysis and path tracing algorithm are applied to identify the breakpoints of the conductive path on the circuit board, and after statistics, the number of disconnections Dcs is obtained;

[0022] Finally, the optical character recognition technology is used to identify and proofread the logo content, detect printing errors and misalignments, and obtain the number of logo errors Bcw after statistics;

[0023] The appearance evaluation unit calculates and evaluates the appearance defect index Wgqx by extracting the scratch area Gmj, the stain area Wmj, the number of wire breaks Dcs and the number of errors Bcw and performing dimensionless processing. The specific calculation formula and evaluation content are as follows:

[0024] ;

[0025] By comparing the preset appearance defect threshold Q with the appearance defect index Wgqx, the following evaluation results are generated:

[0026] If the appearance defect threshold Q> the appearance defect index Wgqx, the first appearance result is generated, indicating that the appearance quality of the circuit board is qualified, and the next step of testing can be continued or the finished product stage can be directly entered;

[0027] If the appearance defect threshold Q ≤ the appearance defect index Wgqx, it means that the appearance quality of the circuit board is unqualified. At this time, it is necessary to perform local repair and cleaning treatment, or rework and scrap treatment depending on the defect situation.

[0028] Preferably, the functional defect recognition module includes a functional recognition unit, a functional evaluation unit and a functional defect analysis unit; the functional recognition unit scans the internal functional image data of the circuit board through image segmentation and pattern recognition algorithms to identify the existing functional defects; after the functional defects are identified, they are marked, and after the corresponding functional defect feature vectors are generated, dimensionless processing is performed to construct a functional defect feature vector value set;

[0029] Preferably, the function evaluation unit calculates and obtains the function defect index Gnqx by extracting a set of functional defect feature vector values. The specific calculation formula is as follows:

[0030] ;

[0031] ;

[0032] ;

[0033] ;

[0034] ;

[0035] Where Lcj represents the inter-layer alignment deviation vector value, and Represents the coordinates of the reference point of the i-th layer in the x and y directions; and Represents the coordinates of the corresponding points in the jth layer in the x and y directions;

[0036] Lkt represents the hole conductivity vector value, Represents the resistance value of the through hole; Ldq represents the electrical connection value vector, represents the resistance value of the electrical connection path; Ljy represents the insulation layer integrity vector value, Indicates the leakage current of the insulation layer;

[0037] , , and They represent the absolute values ​​of the inter-layer alignment deviation vector value Lcj, the hole conductivity vector value Lkt, the electrical connectivity vector value Ldq and the insulation layer integrity vector value Ljy respectively;

[0038] The function evaluation unit compares the preset function defect threshold W with the function defect index Gnqx to generate the following evaluation results:

[0039] If the functional defect threshold W> the functional defect index Gnqx, a first functional result is generated, indicating that the functionality of the circuit board is qualified;

[0040] If the functional defect threshold W ≤ the functional defect index Gnqx, a second functional result is generated, indicating that the functionality of the circuit board is unqualified, and further analysis and processing are performed.

[0041] Preferably, the functional defect analysis unit is used to perform defect analysis and take corresponding measures when the evaluation result of the functional defect index Gnqx is the second functional result;

[0042] By arranging the inter-layer alignment deviation index Lcj, the through-hole conductivity vector value Lkt, the electrical connectivity vector value Ldq and the insulation layer integrity vector value Ljy from high to low, the primary cause of the functional defects of the circuit board is found according to the arrangement order, including inter-layer alignment deviation, electrical connectivity problems and insufficient insulation performance; then combining the image data and the 3D model, the layer and specific location of the defect are accurately located, including a specific conductive path, through hole or insulation layer;

[0043] Finally, the defect analysis results are fed back to the production line, and corresponding corrective measures are proposed based on the type and severity of the defects, including adjusting the production process, increasing inspection efforts and rework.

[0044] Preferably, the reconstruction and adjustment module includes a defect positioning unit and a recording and feedback unit; the defect positioning unit first integrates the image data of each layer through image stitching and reconstruction algorithms to construct a three-dimensional structural model of the circuit board; secondly, based on the relevant defect information detected in the appearance defect recognition module and the functional defect recognition module, the spatial coordinate mapping technology is applied to accurately locate each defect point to a specific level and position in the three-dimensional model, and for each defect, the level where the defect is located is automatically marked, including the surface layer, the conductive layer and the internal layer, as well as the specific position coordinates (x, y, z) of the defect in the three-dimensional space. In addition, multiple defects are aggregated and analyzed to identify existing systematic defects and make special markings;

[0045] The circuit board production detection method based on machine vision includes the following steps:

[0046] Step 1: Obtain multi-angle circuit board image data through an image acquisition module; deploy a multi-light source lighting device to respectively acquire images of the surface layer, conductive layer and internal layer of the multi-layer circuit board;

[0047] Step 2: preprocessing the image data acquired by the image acquisition module through the image preprocessing module; and further configuring the image registration operation for precise alignment of the image in space;

[0048] Step 3: Analyze the preprocessed image data through the appearance defect recognition module, and identify the appearance defects of the multi-layer circuit board, including the scratch area, stain area, number of disconnections and number of marking errors; and calculate and evaluate the appearance defect index Wgqx of the multi-layer circuit board;

[0049] Step 4: The pre-processed image data is analyzed again by the functional defect recognition module to identify the functional defects in the circuit board; the functional defect index Gnqx is calculated and evaluated, and then the functional defects of the multi-layer circuit board are further analyzed according to the evaluation results;

[0050] Step 5. Through the reconstruction and adjustment module, a 3D model is constructed based on the collected multi-layer image data. Then, the detected defects are located in three dimensions, and the layers and positions of the defects are marked. At the same time, the relevant data of the appearance defect index Wgqx and the functional defect index Gnqx are recorded and stored, and fed back to the production line.

[0051] A circuit board production inspection device and medium based on machine vision include a memory and a processor, wherein the memory stores a computer program, the processor is configured to run the computer program to execute the above-mentioned system instructions, and the computer program is configured to execute the above-mentioned circuit board production inspection method based on machine vision when running.

[0052] Beneficial Effects

[0053] The present invention provides a communication network fault early warning method and system based on artificial intelligence, which has the following beneficial effects:

[0054] (1) The circuit board production inspection method, system, equipment and medium based on machine vision can comprehensively obtain the image data of the surface layer and internal conductive layer of the circuit board through the image acquisition module with multi-angle and multi-light source illumination, and realize the spatial precise alignment of multi-layer images by combining the image registration function of the image preprocessing module. Through the 3D model constructed by the reconstruction adjustment module, the system can accurately locate the level and specific position of defects in three-dimensional space; this multi-level and multi-dimensional positioning method significantly improves the accuracy and efficiency of circuit board defect positioning, especially for multi-layer circuit boards with complex structures, which can effectively solve the problem of difficult precise positioning in traditional inspection methods.

[0055] (2) The circuit board production inspection method, system, equipment and medium based on machine vision, the system realizes the rapid detection and evaluation of appearance defects and functional defects through automated image processing and defect recognition technology; using the defect feature vector after dimensionless processing, it can quickly calculate the appearance defect index Wgqx and the functional defect index Gnqx, and compare and evaluate them with the preset threshold value to timely judge the qualification of the circuit board. This automated and efficient inspection method greatly reduces the inspection time in the production process, avoids the delay caused by manual inspection, and further improves the overall efficiency of the production line.

[0056] (3) The machine vision-based circuit board production inspection method, system, equipment and medium reduce the dependence on additional equipment by integrating multifunctional inspection modules. At the same time, the system can automatically handle most defects in the defect analysis and inspection result feedback links, reducing the need for human intervention, reducing the complexity of manual operation and the risk of misoperation. By centrally managing and analyzing inspection data, the system can effectively reduce the maintenance cost and operating cost of the equipment, especially in a large-scale production environment, and can significantly reduce the overall production cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 It is a schematic diagram of the framework structure of the circuit board production and detection system based on machine vision of the present invention;

[0058] Figure 2 The figure is a schematic flow chart of the steps of the circuit board production detection method based on machine vision of the present invention. DETAILED DESCRIPTION

[0059] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0060] Example 1

[0061] See also Figure 1 , the present invention provides a circuit board production inspection system based on machine vision, including an image acquisition module, an image preprocessing module, an appearance defect recognition module, a functional defect recognition module and a reconstruction adjustment module;

[0062] The image acquisition module is used to acquire circuit board image data from multiple angles; by deploying a multi-light source lighting device, images of the surface layer, conductive layer and internal layer of the multi-layer circuit board are respectively acquired;

[0063] The image preprocessing module is used to preprocess the image data acquired by the image acquisition module; and is further configured to perform image registration operations for precise spatial alignment of the images;

[0064] The appearance defect recognition module is used to analyze the preprocessed image data and identify the appearance defects of the multilayer circuit board, including the scratch area, stain area, number of disconnections and number of marking errors; and calculate and evaluate the appearance defect index Wgqx of the multilayer circuit board;

[0065] The functional defect recognition module is used to analyze the preprocessed image data again to identify the functional defects in the circuit board; by calculating and evaluating the functional defect index Gnqx, and then further analyzing the functional defects of the multilayer circuit board according to the evaluation results;

[0066] The reconstruction and adjustment module is used to construct a 3D model based on the collected multi-layer image data, and then three-dimensionally locate the detected defects and mark the level and position of the defects; at the same time, it records and stores the relevant data of the appearance defect index Wgqx and the functional defect index Gnqx, and feeds back to the production line.

[0067] In this embodiment, the multi-angle and multi-light source illumination technology of the image acquisition module improves the imaging quality of the complex structure of the circuit board, and can simultaneously obtain detailed images of the surface layer and the internal conductive layer. Secondly, the image preprocessing module ensures the accurate spatial alignment of multi-layer images through precise image registration operations, laying the foundation for subsequent defect detection. The appearance defect recognition module and the functional defect recognition module conduct in-depth analysis of the appearance and functional defects of the circuit board respectively, and provide quantitative evaluation methods by calculating the appearance defect index Wgqx and the functional defect index Gnqx, which greatly improves the reliability and accuracy of detection. Finally, the reconstruction and adjustment module realizes the accurate marking of multi-layer circuit board defects by constructing a 3D model and three-dimensionally locating defects, and can promptly feedback the detection results to the production line, ensuring closed-loop control of the production process.

[0068] Example 2

[0069] See also Figure 1 The image acquisition module includes a device deployment unit and an image acquisition unit; the device deployment unit is used to deploy a high-resolution camera, and is provided with a plurality of light source illumination devices, and is deployed at different positions within the detection area; in the process of collecting images, by adjusting the intensity, angle and irradiation direction of the light source, the images of the surface layer and the internal conductive layer of the multi-layer circuit board are collected respectively;

[0070] The image acquisition unit acquires image data of the surface layer of the circuit board by vertical illumination, and then acquires images of the multi-layer surface layer and the internal conductive layer by multi-angle illumination, including oblique illumination of 30 degrees, 45 degrees and 60 degrees; by oblique illumination, edge images of the electrical layer, hole images and detailed feature images of the connection points between layers are collected; in addition, the image acquisition unit also uses transmitted light source technology to perform penetrating illumination on the circuit board to acquire image data of its internal conductive layer.

[0071] In this embodiment, the device deployment unit uses a high-resolution camera combined with a multi-light source lighting device. By adjusting the intensity, angle and irradiation direction of the light source, it can accurately capture images of the surface layer and the internal conductive layer, thereby greatly improving the imaging effect of the complex circuit board structure. Secondly, the image acquisition unit can capture detailed images of different levels, including key details of surface features and internal layers, such as the edges, holes and interlayer connection points of the electrical layer, through vertical irradiation and multi-angle oblique irradiation technology. In addition, the application of transmitted light source technology makes the image acquisition of the internal conductive layer more comprehensive and accurate. This improvement not only improves the comprehensiveness and accuracy of image acquisition, but also effectively improves the quality and speed of defect detection, especially for the detection of complex, multi-layer circuit boards. It has significant advantages.

[0072] Example 3

[0073] The image preprocessing module is used to perform denoising and contrast enhancement on the image, and then use the edge detection algorithm to extract the edge information of the structure in the multilayer circuit board, including the wires, solder joints and apertures, and then perform geometric correction on the image;

[0074] After completing the above processing, the same feature points in different images are identified through the feature point matching algorithm, and the images of the multi-layer circuit board are accurately aligned in space using the affine transformation and perspective transformation methods. The image registration process includes aligning the conductive paths and welding points in each layer of the image, which is used to consistently superimpose images of different levels in the same coordinate system.

[0075] In this embodiment, the denoising and contrast enhancement techniques are used to significantly improve the quality of the image, ensuring that the key structures of the circuit board can be more clearly identified in subsequent processing. Secondly, the edge detection algorithm is used to accurately extract important edge information such as wires, solder joints and apertures, laying the foundation for subsequent geometric correction and registration. Then, the distortion in the image is eliminated through geometric correction, making the image more consistent with the actual physical structure. In addition, the feature point matching algorithm is used, combined with affine transformation and perspective transformation technology, to achieve precise spatial alignment of multi-layer images, so that the conductive paths and solder joints of each layer can be consistently superimposed in the same coordinate system. This improvement not only improves the spatial consistency of the image, but also greatly enhances the comprehensive utilization efficiency of multi-layer image data, ensuring the accuracy and reliability of subsequent analysis and defect identification.

[0076] Example 4

[0077] The appearance defect recognition module includes an appearance recognition unit and an appearance evaluation unit; the appearance recognition unit first processes the circuit board surface image through edge detection and morphological analysis technology, recognizes the outline of the scratch, and then uses the area calculation algorithm to obtain and record the scratch area Gmj;

[0078] Then, the stain area on the circuit board surface is detected through color space conversion and cluster analysis technology. The region growing algorithm is used to distinguish the color difference between normal materials and abnormal stains, and the stain area Wmj is calculated;

[0079] Secondly, the connectivity analysis and path tracing algorithm are applied to identify the breakpoints of the conductive path on the circuit board, and the number of disconnections Dcs is obtained after statistics.

[0080] Finally, the optical character recognition technology is used to identify and proofread the logo content, detect printing errors and misalignments, and obtain the number of logo errors Bcw after statistics;

[0081] The appearance evaluation unit calculates and evaluates the appearance defect index Wgqx by extracting the scratch area Gmj, the stain area Wmj, the number of wire breaks Dcs and the number of errors Bcw and performing dimensionless processing. The specific calculation formula and evaluation content are as follows:

[0082] ;

[0083] By comparing the preset appearance defect threshold Q with the appearance defect index Wgqx, the following evaluation results are generated:

[0084] If the appearance defect threshold Q> the appearance defect index Wgqx, the first appearance result is generated, indicating that the appearance quality of the circuit board is qualified, and the next step of testing can be continued or the finished product stage can be directly entered;

[0085] If the appearance defect threshold Q ≤ the appearance defect index Wgqx, it means that the appearance quality of the circuit board is unqualified. At this time, it is necessary to perform local repair and cleaning treatment, or rework and scrap treatment depending on the defect situation.

[0086] In this embodiment, the purpose of collecting the scratch area Gmj is that scratches on the surface of the circuit board may expose or damage the internal conductive path, affecting the electrical performance of the circuit board. Collecting the scratch area helps to evaluate the degree of physical damage on the surface, thereby determining its impact on the overall quality of the circuit board;

[0087] The purpose of collecting the stain area Wmj is that surface stains may come from pollutants or material unevenness during the production process, which may affect the welding quality or cause electrical short circuits. By measuring the stain area, the severity of the contamination can be evaluated to determine whether cleaning or repair is needed;

[0088] The purpose of collecting the number of disconnections Dcs is that the disconnection of the conductive path directly affects the conduction of current and may cause the circuit board to fail. Counting the number of disconnections can help identify major defects in production and assess their threats to the functionality and reliability of the circuit board;

[0089] The purpose of collecting the number of errors Bcw is that incorrect identification or character printing may cause erroneous operation of the circuit board during installation or use. Recording the number of errors helps to detect printing quality and ensure the accuracy and readability of product identification.

[0090] Example 5

[0091] The functional defect recognition module includes a functional recognition unit, a functional evaluation unit and a functional defect analysis unit; the functional recognition unit scans the internal functional image data of the circuit board through image segmentation and pattern recognition algorithms to identify the existing functional defects; after the functional defects are identified, they are marked, and after the corresponding functional defect feature vectors are generated, dimensionless processing is performed to construct a functional defect feature vector value set;

[0092] The function evaluation unit calculates and obtains the function defect index Gnqx by extracting the functional defect feature vector value set. The specific calculation formula is as follows:

[0093] ;

[0094] ;

[0095] ;

[0096] ;

[0097] ;

[0098] Where Lcj represents the inter-layer alignment deviation vector value, and Represents the coordinates of the reference point of the i-th layer in the x and y directions; and Represents the coordinates of the corresponding points in the jth layer in the x and y directions;

[0099] Lkt represents the hole conductivity vector value, Represents the resistance value of the through hole; Ldq represents the electrical connection value vector, represents the resistance value of the electrical connection path; Ljy represents the insulation layer integrity vector value, Indicates the leakage current of the insulation layer;

[0100] , , and They represent the absolute values ​​of the inter-layer alignment deviation vector value Lcj, the hole conductivity vector value Lkt, the electrical connectivity vector value Ldq and the insulation layer integrity vector value Ljy respectively;

[0101] The function evaluation unit compares the preset function defect threshold W with the function defect index Gnqx to generate the following evaluation results:

[0102] If the functional defect threshold W> the functional defect index Gnqx, a first functional result is generated, indicating that the functionality of the circuit board is qualified;

[0103] If the functional defect threshold W ≤ the functional defect index Gnqx, a second functional result is generated, indicating that the functionality of the circuit board is unqualified, and further analysis and processing are performed.

[0104] The functional defect analysis unit is used to perform defect analysis and take corresponding measures when the evaluation result of the functional defect index Gnqx is the second functional result;

[0105] By arranging the inter-layer alignment deviation index Lcj, the through-hole conductivity vector value Lkt, the electrical connectivity vector value Ldq and the insulation layer integrity vector value Ljy from high to low, the primary cause of the functional defects of the circuit board is found according to the arrangement order, including inter-layer alignment deviation, electrical connectivity problems and insufficient insulation performance; then combining the image data and the 3D model, the layer and specific location of the defect are accurately located, including a specific conductive path, through hole or insulation layer;

[0106] Finally, the defect analysis results are fed back to the production line, and corresponding corrective measures are proposed based on the type and severity of the defects, including adjusting the production process, increasing inspection efforts and rework.

[0107] In this embodiment, the interlayer alignment deviation index Lcj reflects the precise alignment of the conductive paths of each layer of the multilayer circuit board. If there is a deviation in the interlayer alignment, it may cause a short circuit, a break or signal interference in the conductive path, thereby affecting the electrical performance of the circuit board; collecting the interlayer alignment deviation index Lcj helps to detect the interlayer alignment accuracy and ensure that the position of each layer of the conductive path is accurate.

[0108] The through-hole conductivity vector value Lkt indicates whether the through-hole in the circuit board can effectively conduct current. The through-hole connects the conductive paths of different layers and is an important structure to ensure the transmission of electrical signals. The through-hole conductivity vector value Lkt helps to determine whether the electrical performance of the through-hole meets the standard, avoiding poor signal transmission or circuit failure caused by excessive through-hole resistance.

[0109] The electrical connectivity vector value Ldq reflects the overall resistance and connection status of each conductive path in the circuit board. Good electrical connectivity is the basis for ensuring the normal flow of current and the normal function of the circuit board; collecting the electrical connectivity vector value Ldq can evaluate the possible poor connection, open circuit or high resistance problems in the circuit to ensure the reliability of electrical performance.

[0110] The insulation layer integrity vector value Ljy is directly related to the insulation performance of the circuit board, preventing current leakage or short circuit; the insulation layer integrity vector value Ljy is collected to evaluate whether there are cracks, holes or other defects in the insulation layer, thereby ensuring the safety and stability of the circuit board in a high voltage environment.

[0111] Example 6

[0112] The reconstruction and adjustment module includes a defect location unit and a recording and feedback unit; the defect location unit first integrates the image data of each layer through image stitching and reconstruction algorithms to construct a three-dimensional structural model of the circuit board; secondly, based on the relevant defect information detected in the appearance defect recognition module and the functional defect recognition module, the spatial coordinate mapping technology is applied to accurately locate each defect point to a specific level and position in the three-dimensional model, and for each defect, the level where the defect is located, including the surface layer, the conductive layer and the internal layer, and the specific position coordinates (x, y, z) of the defect in the three-dimensional space are automatically marked. In addition, multiple defects are aggregated and analyzed to identify existing systematic defects and make special markings;

[0113] In this embodiment, the multi-layer image data is integrated into an overall three-dimensional structural model through image stitching and reconstruction algorithms. This method makes it possible to accurately locate each defect point in three-dimensional space, greatly improving the detection accuracy of multi-layer circuit board defects; multiple defects are aggregated and analyzed to identify possible systemic defects and make special annotations. This not only helps to discover single problems, but more importantly, it can identify the potential connections between multiple defects, thereby effectively improving the accuracy and efficiency of overall quality control.

[0114] Example 7

[0115] See also Figure 2 , a circuit board production detection method based on machine vision includes the following steps:

[0116] Step 1: Obtain multi-angle circuit board image data through an image acquisition module; deploy a multi-light source lighting device to respectively acquire images of the surface layer, conductive layer and internal layer of the multi-layer circuit board;

[0117] Step 2: preprocessing the image data acquired by the image acquisition module through the image preprocessing module; and further configuring the image registration operation for precise alignment of the image in space;

[0118] Step 3: Analyze the preprocessed image data through the appearance defect recognition module, and identify the appearance defects of the multi-layer circuit board, including the scratch area, stain area, number of disconnections and number of marking errors; and calculate and evaluate the appearance defect index Wgqx of the multi-layer circuit board;

[0119] Step 4: The pre-processed image data is analyzed again by the functional defect recognition module to identify the functional defects in the circuit board; the functional defect index Gnqx is calculated and evaluated, and then the functional defects of the multi-layer circuit board are further analyzed according to the evaluation results;

[0120] Step 5. Through the reconstruction and adjustment module, a 3D model is constructed based on the collected multi-layer image data. Then, the detected defects are located in three dimensions, and the layers and positions of the defects are marked. At the same time, the relevant data of the appearance defect index Wgqx and the functional defect index Gnqx are recorded and stored, and fed back to the production line.

[0121] In this embodiment, a comprehensive inspection and analysis of multi-layer circuit boards is achieved through a series of precise and systematic steps. First, multi-angle image acquisition and multi-light source illumination technology are used to ensure that detailed image data of each layer of the circuit board is obtained. Subsequently, image preprocessing and registration operations improve the precise alignment of images in space, laying the foundation for subsequent defect identification. In the appearance defect recognition module, by calculating the appearance defect index Wgqx, the system can accurately identify and evaluate the surface defects of the circuit board; while the functional defect recognition module further analyzes the functional defects and generates the functional defect index Gnqx to ensure the functional integrity of the circuit board. Finally, through the reconstruction and adjustment module, a three-dimensional model of the circuit board is constructed, and the defects are accurately located and marked to achieve efficient feedback and production adjustment of defects. This method improves the degree of automation of detection, enhances the accuracy and comprehensiveness of defect identification, and provides reliable data support and analysis basis for the optimization of production processes and the improvement of product quality.

[0122] Example 8

[0123] A circuit board production inspection device and medium based on machine vision include a memory and a processor, wherein the memory stores a computer program, the processor is configured to run the computer program to execute the above-mentioned system instructions, and the computer program is configured to execute the above-mentioned circuit board production inspection method based on machine vision when running.

[0124] The specific contents are as follows:

[0125] Image data acquisition: The image acquisition module is used to obtain multi-angle circuit board image data. During the acquisition process, a multi-light source lighting device is used to ensure that the surface and internal conductive layer images of the multi-layer circuit board are captured.

[0126] Image preprocessing: In the image preprocessing module, the collected image data is preprocessed, including noise removal, image enhancement, and image registration technology is used to achieve precise spatial alignment of the image.

[0127] Appearance defect recognition: Analyze the pre-processed image data, identify and quantify the appearance defects of the circuit board in the appearance defect recognition module, calculate and evaluate the appearance defect index Wgqx, and generate an appearance quality report

[0128] Functional defect identification: Use the functional defect identification module to analyze and identify functional defects in image data, calculate the functional defect index Gnqx, and conduct in-depth analysis based on the results to generate a functional quality report.

[0129] In this embodiment, the image acquisition module is combined with a multi-light source lighting device to achieve comprehensive image acquisition of the multi-layer circuit board, ensuring clear imaging of the surface and internal conductive layer. The image preprocessing module further optimizes the image quality through noise removal and image enhancement technology, combined with precise image registration, laying the foundation for subsequent defect identification. The appearance defect recognition module can effectively identify and quantify the appearance defects of the circuit board, calculate the appearance defect index Wgqx, and generate a detailed quality report. The functional defect recognition module focuses on the analysis of functional defects, calculates the functional defect index Gnqx, and generates a corresponding functional quality report to ensure the functional integrity of the circuit board. Finally, the reconstruction and adjustment module achieves accurate labeling and real-time feedback through three-dimensional model construction and three-dimensional defect positioning, thereby optimizing the adjustment and improvement of the production line. This method improves the automation level and accuracy of overall detection, and improves production efficiency and product quality.

[0130] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. The circuit board production inspection system based on machine vision is characterized by: It includes an image acquisition module, an image preprocessing module, an appearance defect recognition module, a functional defect recognition module and a reconstruction adjustment module; The image acquisition module is used to acquire circuit board image data from multiple angles; by deploying a multi-light source lighting device, images of the surface layer, conductive layer and internal layer of the multi-layer circuit board are respectively acquired; The image acquisition module includes a device deployment unit and an image acquisition unit; the device deployment unit is used to deploy a high-resolution camera, and is also provided with a plurality of light source illumination devices, which are deployed at different positions within the detection area; in the process of collecting images, the images of the surface layer and the internal conductive layer of the multi-layer circuit board are collected respectively by adjusting the intensity, angle and irradiation direction of the light source; The image acquisition unit acquires image data of the surface layer of the circuit board by vertical illumination, and then acquires images of the multi-layer surface layer and the internal conductive layer by multi-angle illumination, including oblique illumination of 30 degrees, 45 degrees and 60 degrees; the edge image of the electrical layer, the hole image and the detail feature image of the interlayer connection point are collected by oblique illumination; in addition, the image acquisition unit also uses the transmission light source technology to perform penetrating illumination on the circuit board to acquire image data of its internal conductive layer; The image preprocessing module is used to preprocess the image data acquired by the image acquisition module; and is further configured to perform image registration operations for precise spatial alignment of the images; The appearance defect recognition module is used to analyze the preprocessed image data and identify the appearance defects of the multilayer circuit board, including the scratch area, stain area, number of disconnections and number of marking errors; and calculate and evaluate the appearance defect index Wgqx of the multilayer circuit board; The functional defect recognition module is used to analyze the preprocessed image data again to identify the functional defects in the circuit board; by calculating and evaluating the functional defect index Gnqx, and then further analyzing the functional defects of the multilayer circuit board according to the evaluation results; The reconstruction and adjustment module is used to construct a 3D model based on the collected multi-layer image data, and then three-dimensionally locate the detected defects and mark the level and position of the defects; at the same time, it records and stores the relevant data of the appearance defect index Wgqx and the functional defect index Gnqx, and feeds back to the production line.

2. The circuit board production inspection system based on machine vision according to claim 1, characterized in that: The image preprocessing module is used to perform denoising and contrast enhancement on the image, and then use the edge detection algorithm to extract the edge information of the structure in the multilayer circuit board, including the wires, solder joints and apertures, and then perform geometric correction on the image; After completing the above processing, the same feature points in different images are identified through the feature point matching algorithm, and the images of the multi-layer circuit board are accurately aligned in space using the affine transformation and perspective transformation methods. The image registration process includes aligning the conductive paths and welding points in each layer of the image, which is used to consistently superimpose images of different levels in the same coordinate system.

3. The circuit board production inspection system based on machine vision according to claim 1, characterized in that: The appearance defect recognition module includes an appearance recognition unit and an appearance evaluation unit; the appearance recognition unit first processes the circuit board surface image through edge detection and morphological analysis technology, recognizes the outline of the scratch, and then uses the area calculation algorithm to obtain and record the scratch area Gmj; Then, the stain area on the surface of the circuit board is detected through color space conversion and cluster analysis technology, and the color difference between normal materials and abnormal stains is distinguished by the region growing algorithm to calculate the stain area Wmj; Secondly, the connectivity analysis and path tracing algorithm are applied to identify the breakpoints of the conductive path on the circuit board, and the number of disconnections Dcs is obtained after statistics. Finally, the optical character recognition technology is used to identify and proofread the logo content, detect printing errors and misalignments, and obtain the number of logo errors Bcw after statistics; The appearance evaluation unit calculates and evaluates the appearance defect index Wgqx by extracting the scratch area Gmj, the stain area Wmj, the number of wire breaks Dcs and the number of errors Bcw and performing dimensionless processing. The specific calculation formula and evaluation content are as follows: ； By comparing the preset appearance defect threshold Q with the appearance defect index Wgqx, the following evaluation results are generated: If the appearance defect threshold Q> the appearance defect index Wgqx, the first appearance result is generated, indicating that the appearance quality of the circuit board is qualified, and the next step of testing is continued or directly entered into the finished product stage; If the appearance defect threshold Q ≤ the appearance defect index Wgqx, it means that the appearance quality of the circuit board is unqualified. At this time, it is necessary to perform local repair and cleaning treatment, or rework and scrap treatment depending on the defect situation.

4. The circuit board production inspection system based on machine vision according to claim 1, characterized in that: The functional defect recognition module includes a functional recognition unit, a functional evaluation unit and a functional defect analysis unit; the functional recognition unit scans the internal functional image data of the circuit board through image segmentation and pattern recognition algorithms to identify the existing functional defects; after the functional defects are identified, they are marked, and after the corresponding functional defect feature vectors are generated, dimensionless processing is performed to construct a functional defect feature vector value set.

5. The circuit board production inspection system based on machine vision according to claim 4 is characterized in that: The function evaluation unit calculates and obtains the function defect index Gnqx by extracting the functional defect feature vector value set. The specific calculation formula is as follows: ； ； ； ； ； Where Lcj represents the inter-layer alignment deviation vector value, and Represents the coordinates of the reference point of the i-th layer in the x and y directions; and Represents the coordinates of the corresponding points in the jth layer in the x and y directions; Lkt represents the hole conductivity vector value, Represents the resistance value of the through hole; Ldq represents the electrical connection value vector, represents the resistance value of the electrical connection path; Ljy represents the insulation layer integrity vector value, Indicates the leakage current of the insulation layer; , , and They represent the absolute values ​​of the inter-layer alignment deviation vector value Lcj, the hole conductivity vector value Lkt, the electrical connectivity vector value Ldq and the insulation layer integrity vector value Ljy respectively; The function evaluation unit compares the preset function defect threshold W with the function defect index Gnqx to generate the following evaluation results: If the functional defect threshold W> the functional defect index Gnqx, a first functional result is generated, indicating that the functionality of the circuit board is qualified; If the functional defect threshold W ≤ the functional defect index Gnqx, a second functional result is generated, indicating that the functionality of the circuit board is unqualified, and further analysis and processing are performed.

6. The circuit board production inspection system based on machine vision according to claim 5, characterized in that: The functional defect analysis unit is used to perform defect analysis and take corresponding measures when the evaluation result of the functional defect index Gnqx is the second functional result; By arranging the inter-layer alignment deviation value Lcj, the through-hole conductivity vector value Lkt, the electrical connectivity vector value Ldq and the insulation layer integrity vector value Ljy from high to low, the primary cause of the functional defects of the circuit board is found according to the arrangement order, including inter-layer alignment deviation, electrical connectivity problems and insufficient insulation performance; then combining the image data and the 3D model, the layer and specific position of the defect are accurately located, including a specific conductive path, through hole or insulation layer; Finally, the defect analysis results are fed back to the production line, and corresponding corrective measures are proposed based on the type and severity of the defects, including adjusting the production process, increasing inspection efforts and rework.

7. The circuit board production inspection system based on machine vision according to claim 1, characterized in that: The reconstruction and adjustment module includes a defect positioning unit and a recording and feedback unit; the defect positioning unit first integrates the image data of each layer through image stitching and reconstruction algorithms to construct a three-dimensional structural model of the circuit board; secondly, based on the relevant defect information detected in the appearance defect recognition module and the functional defect recognition module, the spatial coordinate mapping technology is applied to accurately locate each defect point to a specific level and position in the three-dimensional model, and for each defect, the level where the defect is located is automatically marked, including the surface layer, the conductive layer and the internal layer, as well as the specific position coordinates (x, y, z) of the defect in the three-dimensional space. In addition, multiple defects are aggregated and analyzed to identify existing systematic defects and make special markings.

8. A circuit board production inspection method based on machine vision, applied to a circuit board production inspection system based on machine vision according to any one of claims 1 to 7, characterized in that: The following steps are involved: Step 1: Obtain multi-angle circuit board image data through an image acquisition module; deploy a multi-light source lighting device to respectively acquire images of the surface layer, conductive layer and internal layer of the multi-layer circuit board; Step 2: preprocessing the image data acquired by the image acquisition module through the image preprocessing module; and further configuring the image registration operation for precise alignment of the image in space; Step 3: Analyze the preprocessed image data through the appearance defect recognition module, and identify the appearance defects of the multi-layer circuit board, including the scratch area, stain area, number of disconnections and number of marking errors; and calculate and evaluate the appearance defect index Wgqx of the multi-layer circuit board; Step 4: The pre-processed image data is analyzed again by the functional defect recognition module to identify the functional defects in the circuit board; the functional defect index Gnqx is calculated and evaluated, and then the functional defects of the multi-layer circuit board are further analyzed according to the evaluation results; Step 5. Through the reconstruction and adjustment module, a 3D model is constructed based on the collected multi-layer image data. Then, the detected defects are located in three dimensions, and the layers and positions of the defects are marked. At the same time, the relevant data of the appearance defect index Wgqx and the functional defect index Gnqx are recorded and stored, and fed back to the production line.

9. A computer storage medium, comprising a memory, characterized in that: The memory stores a computer program, and the computer program is configured to execute the circuit board production inspection method based on machine vision in claim 8 when running.

Images