PCB surface defect detection system based on image recognition

Abstract

The invention relates to the technical field of PCB surface defect detection, and particularly discloses a PCB surface defect detection system based on image recognition, which comprises an image quality control module, a surface defect type confirmation module, a surface defect report feedback module, a database and a PCB quality feedback module, according to the method, whether the quality of the images correspondingly collected by the PCBs reaches the standard or not is evaluated, if not, the illumination system is adjusted, if yes, the surface defect types corresponding to the target PCBs are determined, the surface defect reports are generated according to the surface defect types and the surface defect severity levels corresponding to the target PCBs, and the surface defect reports are fed back. The influence of subjective judgment caused by manual visual inspection or a traditional measuring tool is avoided, meanwhile, a large amount of data can be rapidly obtained and processed in batches by combining an automatic detection method of coordinate data, and the detection efficiency and the comprehensiveness and reliability of PCB surface defect recognition are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of PCB board surface defect detection, and in particular to a PCB board surface defect detection system based on image recognition. Background Art

[0002] With the advent of the Industrial 4.0 era, the manufacturing industry is developing towards intelligence, high-end and green directions. As the support of electronic components, the automated detection of surface defects of PCB has become the key to the development of the manufacturing industry. Therefore, in order to be able to economically and efficiently detect various defects on the surface of PCB boards, improve detection accuracy and efficiency, and reduce production costs, it is necessary to develop a PCB surface defect detection system based on image recognition.

[0003] For example, the Chinese patent publication number CN115239727B discloses a PCB board surface defect detection method, including: obtaining a pre-processed surface image of the PCB board to be detected, determining a first wire path edge image and a solder joint area image, and then determining a second wire path edge image; determining the horizontal and vertical edge stability corresponding to each group of preset edge pixel points, and then determining the horizontal and vertical deviation values ​​corresponding to each N groups of preset edge pixel points, thereby determining each initial node in the second wire path edge image; according to the position of each preset edge pixel point corresponding to each initial node, determining the wire edge flatness corresponding to each initial node, and then determining the third wire path edge image, thereby judging whether there are defects in the PCB board to be detected and the type of defects. The present invention uses material testing and analysis technology to perform PCB board surface defect detection, thereby improving the accuracy of defect detection.

[0004] The above existing technologies still have the following problems: 1. When collecting the surface image of the PCB board, the surface image of the PCB board is currently only collected directly by a camera, without considering whether the quality of the surface image of the PCB board meets the standard, that is, it does not reflect from the side whether there is a problem with the surface image quality of the PCB board by combining the uniformity of the illumination area on the surface of the PCB board, which reduces the collection quality of the surface image of the PCB board, fails to lay a good foundation for the subsequent surface defect analysis of the PCB board, and reduces the accuracy of the subsequent surface defect analysis of the PCB board.

[0005] 2. The 3D information model corresponding to the PCB board is not created in combination with the collected images, that is, the surface defect type corresponding to the PCB board is not analyzed in depth and detail in combination with the information obtained from the 3D information model, which seriously affects the detection accuracy and reduces the comprehensiveness of PCB board surface defect identification and the reliability of the detection results.

[0006] 3. Currently, the analysis of hole offset and flatness of PCB boards is not combined with coordinate data for accurate analysis, and often relies on manual visual inspection or traditional measurement tools. These methods are easily affected by subjective judgment, resulting in reduced accuracy. Automated detection methods that are not combined with coordinate data cannot quickly obtain large amounts of data and perform batch processing, which significantly reduces detection efficiency. Summary of the invention

[0007] In view of this, in order to solve the problems raised in the above background technology, a PCB board surface defect detection system based on image recognition is proposed.

[0008] The purpose of the present invention can be achieved through the following technical solutions: The present invention provides a PCB board surface defect detection system based on image recognition, including: an image quality control module, which is used to place each PCB board produced in the current batch of the target factory in an image acquisition area, turn on the lighting system, collect grayscale images of each PCB board under the lighting system, and evaluate whether the quality of the image collected corresponding to each PCB board meets the standard. If not, the lighting system is adjusted; if it meets the standard, a surface defect type confirmation module is executed.

[0009] The surface defect type confirmation module is used to record the PCB boards that meet the image quality standards as target PCB boards, build three-dimensional information models of each target PCB board according to the corresponding collected images of each target PCB board, extract PCB board information from the three-dimensional information model, and confirm the surface defect type corresponding to each target PCB board.

[0010] The surface defect report feedback module is used to calculate the surface defect severity index of each target PCB board, evaluate the surface defect severity level corresponding to each target PCB board, generate the surface defect report corresponding to each target PCB board according to the surface defect type and surface defect severity level corresponding to each target PCB board, and feedback the surface defect report.

[0011] The database is used to store the standard position of each hole in each target PCB board and store the surface defect severity index interval corresponding to each surface defect severity level.

[0012] The PCB board quality feedback module is used to analyze the quality qualification coefficient of the PCB boards produced by the current batch of the target factory. If it is less than the set reference PCB board quality qualification coefficient, it indicates that the quality of the PCB boards produced by the current batch of the target factory is unqualified, and feedback is given.

[0013] Compared with the prior art, the embodiments of the present invention have at least the following advantages or beneficial effects: (1) The present invention evaluates whether the quality of the image collected corresponding to each PCB board meets the standard by combining the grayscale image of each PCB board under the lighting system. If it does not meet the standard, the lighting system is adjusted, thereby improving the collection quality of the PCB board surface image, laying a good foundation for the subsequent PCB board surface defect analysis, and improving the accuracy of the subsequent PCB board surface defect analysis.

[0014] (2) The present invention creates a three-dimensional information model corresponding to the PCB board by combining the collected images, extracts the PCB board information from the three-dimensional information model, confirms the surface defect type corresponding to each target PCB board, improves the detection accuracy, and at the same time improves the comprehensiveness of PCB board surface defect identification and the reliability of the detection results.

[0015] (3) The present invention accurately analyzes the hole position offset and flatness of the PCB board by combining coordinate data, avoiding the influence of subjective judgment caused by manual visual inspection or traditional measuring tools. At the same time, the automated detection method combined with coordinate data can quickly obtain a large amount of data and perform batch processing, which significantly improves the detection efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for describing the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0017] Figure 1 It is a schematic diagram of the system module structure connection of the present invention.

[0018] Figure 2 This is a flow chart of image quality control according to the present invention. DETAILED DESCRIPTION

[0019] 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.

[0020] See also Figure 1 As shown, the present invention provides a PCB board surface defect detection system based on image recognition, including: an image quality control module, a surface defect type confirmation module, a surface defect report feedback module, a database and a PCB board quality feedback module.

[0021] The image quality control module is connected to the surface defect type confirmation module, the surface defect type confirmation module is connected to the surface defect report feedback module, the surface defect report feedback module is connected to the PCB board quality feedback module, and both the surface defect type confirmation module and the surface defect report feedback module are connected to the database.

[0022] See also Figure 2 As shown, the image quality control module is used to place each PCB board produced in the current batch of the target factory in the image acquisition area, turn on the lighting system, collect the grayscale image of each PCB board under the lighting system, and evaluate whether the quality of the image collected corresponding to each PCB board meets the standard. If not, adjust the lighting system. If it meets the standard, execute the surface defect type confirmation module.

[0023] It should be noted that the grayscale image of each PCB board under the lighting system is collected by using a high-definition camera, and the image is imported into image processing software, and the grayscale image of each PCB board under the lighting system is obtained by processing the image processing software.

[0024] It should also be noted that the lighting system is turned on in order to design a suitable lighting solution to ensure that the surface of the PCB board is evenly illuminated, thereby improving the contrast and clarity of the image and thus improving the image acquisition quality.

[0025] In a specific embodiment of the present invention, the specific process of evaluating whether the quality of the image collected corresponding to each PCB board meets the standard is as follows: A1, extracting the grayscale value corresponding to each grayscale area from the grayscale image of each PCB board under the lighting system, and performing mean calculation on the grayscale value to obtain the average grayscale value of each PCB board under the lighting system, and recording it as Wherein, i represents the number of the PCB board, i=1,2,...,n.

[0026] A2. Extract the minimum value of the grayscale values ​​corresponding to each grayscale area from the grayscale image of each PCB board under the lighting system and record it as

[0027] A3. Calculate the image quality coefficient β corresponding to each PCB board i ,

[0028] It should be noted that the grayscale uniformity is obtained by the ratio of the minimum grayscale value to the average grayscale value. The closer the value is to 1, the more uniform the illumination is and the better the image quality is.

[0029] A4. Compare the image quality coefficient corresponding to each PCB board with the set reference image quality coefficient. If the image quality coefficient corresponding to a PCB board is greater than or equal to the set reference image quality coefficient, it indicates that the image quality corresponding to the PCB board meets the standard. Otherwise, it indicates that the image quality corresponding to the PCB board does not meet the standard.

[0030] In a specific embodiment of the present invention, the specific process of adjusting the lighting system is as follows: marking a PCB board whose image quality does not meet the standard as an image-abnormal PCB board, moving the position of the lighting lamp in the lighting system, and when the position of the lighting lamp is moved, re-collecting the grayscale image of each image-abnormal PCB board, and evaluating whether the image quality corresponding to each image-abnormal PCB board meets the standard in the same way as the evaluation method of whether the image quality corresponding to each PCB board meets the standard; if it still does not meet the standard, continue to move the position of the lighting lamp in the lighting system, and repeat the above steps until the image quality corresponding to the image-abnormal PCB board meets the standard.

[0031] The embodiment of the present invention evaluates whether the quality of the image collected corresponding to each PCB board meets the standard by combining the grayscale image of each PCB board under the lighting system. If it does not meet the standard, the lighting system is adjusted, thereby improving the collection quality of the surface image of the PCB board, laying a good foundation for the subsequent surface defect analysis of the PCB board, and improving the accuracy of the subsequent surface defect analysis of the PCB board.

[0032] The surface defect type confirmation module is used to record the PCB boards that meet the image quality standards as target PCB boards, build three-dimensional information models of each target PCB board according to the images collected corresponding to each target PCB board, extract PCB board information from the three-dimensional information model, and confirm the surface defect type corresponding to each target PCB board.

[0033] It should be noted that the construction of the three-dimensional information model of each target PCB board according to the corresponding collected images of each target PCB board is a prior art and will not be described in detail here.

[0034] In a specific embodiment of the present invention, the PCB board information includes the position of each hole, the position of the center point of the PCB board, the position of each feature point, the number of surface cracks, the length of each crack, the number of dirt locations and the dirt area corresponding to each dirt location.

[0035] In a specific embodiment of the present invention, the characteristic points are four corners of the PCB board.

[0036] In a specific embodiment of the present invention, the specific process of confirming the surface defect type corresponding to each target PCB board is as follows: B1, extracting each hole position, the center point position of the PCB board, the position of each feature point, the number of surface cracks, the length of each crack, the number of dirt places and the dirt area corresponding to each dirt place from the PCB board information corresponding to each target PCB board, and calculating the hole position deviation degree ψ of each target PCB board respectively. j , Flatness Cracking degreeω j and dirtiness x j , where j represents the number of the target PCB board, j=1,2,...,m.

[0037] In a specific embodiment of the present invention, the specific process of calculating the hole position deviation of each target PCB board is as follows: C1, extracting the standard position of each hole position in each target PCB board from the database, and bringing it into the corresponding three-dimensional information model to obtain the standard position coordinates of each hole position in each target PCB board, recorded as Among them, g represents the hole position number, g=1,2,.r...

[0038] C2. Bring the hole positions corresponding to each target PCB board into the corresponding three-dimensional information model to obtain the coordinates of the hole positions corresponding to each target PCB board, recorded as (x jg ,y jg ,z jg ).

[0039] C3. Calculate the position deviation of each hole in each target PCB board δ jg , Here, Δx′, Δy′ and Δz′ represent the position offset values ​​corresponding to the x-axis, y-axis and z-axis respectively.

[0040] C4. Calculate the hole position deviation of each target PCB board ψ j , Wherein, r represents the number of holes.

[0041] In a specific embodiment of the present invention, the specific process of calculating the flatness of each target PCB board is as follows: D1, the center point position of each target PCB board corresponding to each target PCB board is brought into the corresponding three-dimensional information model to obtain the y-axis value of the center point of each target PCB board corresponding to each target PCB board, and recorded as

[0042] D2. Bring the positions of the feature points corresponding to each target PCB board into the corresponding three-dimensional information model to obtain the y-axis values ​​of the feature points corresponding to each target PCB board, recorded as y jp , where p represents the number of the feature point, p = 1, 2, ..., q.

[0043] D3. Calculate the flatness of each target PCB board Among them, H′ represents the allowed y-axis height difference, q represents the number of feature points, and y j(p+1) Indicates the y-axis value of the p+1th feature point corresponding to the jth target PCB board.

[0044] The embodiment of the present invention accurately analyzes the hole position offset and flatness of the PCB board by combining coordinate data, avoiding the influence of subjective judgment caused by manual visual inspection or traditional measuring tools. At the same time, the automated detection method combined with coordinate data can quickly obtain a large amount of data and perform batch processing, which significantly improves the detection efficiency.

[0045] It should be noted that the specific process of calculating the cracking degree of each target PCB board is as follows: the number of surface cracks corresponding to each target PCB board is recorded as σ j .

[0046] The maximum value is extracted from each crack length corresponding to each target PCB board and recorded as

[0047] Calculate the cracking degree ω of each target PCB board j , Wherein, σ′ and L′ represent the number of surface cracks and crack length of the set reference, respectively.

[0048] It should be noted that the specific process of calculating the contamination degree of each target PCB board is as follows: the number of contamination points corresponding to each target PCB board is recorded as μ j .

[0049] The dirt areas corresponding to the dirt locations on each target PCB board are accumulated to obtain the total dirt area corresponding to each target PCB board, which is recorded as S j .

[0050] Calculate the contamination degree χ of each target PCB board j , Wherein, μ′ and S′ represent the number of dirt sites and the dirt area of ​​the set reference, respectively.

[0051] B2. Establish a surface defect type diagnosis model corresponding to each target PCB board and output the surface defect type diagnosis result. The surface defect type diagnosis model is expressed as: Among them, P1, P2, P3, P4 and P5 represent the evaluation conditions of the diagnosis results of each surface defect type, P1 represents ψ j ≥ψ′, P2 means P3 represents ω j ≥ω′, P4 represents χ j ≥χ′, P5 indicates Among them, ψ′, ω′ and χ′ represent the hole position deviation, flatness, cracking degree and dirt degree of the set reference, respectively.

[0052] The embodiment of the present invention creates a three-dimensional information model corresponding to the PCB board by combining the collected images, extracts the PCB board information from the three-dimensional information model, confirms the surface defect type corresponding to each target PCB board, improves the detection accuracy, and at the same time improves the comprehensiveness of PCB board surface defect identification and the reliability of the detection results.

[0053] The surface defect report feedback module is used to calculate the surface defect severity index of each target PCB board, evaluate the surface defect severity level corresponding to each target PCB board, generate a surface defect report corresponding to each target PCB board according to the surface defect type and surface defect severity level corresponding to each target PCB board, and feedback the surface defect report.

[0054] In a specific embodiment of the present invention, the formula for calculating the surface defect severity index of each target PCB board is: Among them, θ j represents the surface defect severity index of the j-th target PCB board, a1, a2, a3 and a4 represent the weights of the surface defect severity index assessment corresponding to the set hole position offset, flatness, cracking degree and dirt degree, a1+a2+a3+a4=1, and e represents a natural constant.

[0055] In a specific embodiment of the present invention, the setting value of a1 is 0.25, the setting value of a2 is 0.25, the setting value of a3 is 0.25, and the setting value of a4 is 0.25. For PCB boards that require high-precision positioning and connection, the hole position offset is crucial. Flatness has a direct impact on the welding quality, assembly accuracy and stability of the PCB board. Cracking refers to cracks or fractures that appear on the surface or inside of the PCB board. Cracking may cause structural damage, electrical performance degradation or even failure of the PCB board. In any application scenario, cracking is an unacceptable serious defect. Therefore, when calculating the surface defect severity index of the PCB board, cracking is a factor that must be highly valued. In application scenarios that require high cleanliness or high reliability, dirtiness is also one of the important indicators for evaluating the quality of the PCB board. Therefore, in practical applications, it is necessary to comprehensively consider the importance of these four factors based on specific product requirements and application scenarios.

[0056] In a specific embodiment of the present invention, the method for evaluating the surface defect severity level corresponding to each target PCB board is: comparing the surface defect severity index of each target PCB board with the surface defect severity index interval corresponding to each surface defect severity level stored in the database; if the surface defect severity index of a certain target PCB board is within the surface defect severity index interval corresponding to a certain surface defect severity level, then the surface defect severity level is used as the surface defect severity level corresponding to the target PCB board, thereby obtaining the surface defect severity level corresponding to each target PCB board, wherein the surface defect severity level includes no defect, slight defect, general defect and severe defect.

[0057] The database is used to store the standard position of each hole in each target PCB board and store the surface defect severity index interval corresponding to each surface defect severity level.

[0058] Table 1 Data sources in the database of this embodiment

[0059]

[0060] The PCB board quality feedback module is used to analyze the quality qualification coefficient of the PCB boards produced by the current batch of the target factory. If the quality qualification coefficient of the PCB boards produced by the current batch of the target factory is less than the set reference quality qualification coefficient of the PCB boards, it indicates that the quality of the PCB boards produced by the current batch of the target factory is unqualified, and feedback is given.

[0061] In a specific embodiment of the present invention, the specific process of analyzing the quality qualification coefficient of the PCB boards produced in the current batch of the target factory is as follows: E1. Counting the number of target PCB boards with a surface defect severity level of no defect produced in the current batch of the target factory, and recording it as τ 无 .

[0062] E2. Count the number of target PCB boards with general defects and severe defects in the current batch of target factory production respectively, and add them up to get the number of unqualified PCB boards in the current batch of target factory production, which is recorded as τ. 不 .

[0063] E3. Calculate the quality qualification coefficient ζ of the current batch of PCB boards produced by the target factory. Among them, m represents the number of target PCB boards, K1 and K2 represent the proportion of defect-free PCB boards and the proportion of unqualified PCB boards of the set reference, respectively, a5 and a6 represent the weights of the proportion of defect-free PCB boards and the proportion of unqualified PCB boards corresponding to the proportion of PCB board quality qualification assessment, respectively, a5+a6=1.

[0064] In a specific embodiment of the present invention, a5 is set to a value of 0.5, and a6 is set to a value of 0.5. In practical applications, the proportion of defect-free PCB boards and the proportion of unqualified PCB boards are both important indicators for evaluating the quality qualification coefficient of PCB boards. The proportion of defect-free PCB boards directly reflects the quality of production, while the proportion of unqualified PCB boards helps the factory to find problems and make improvements. Therefore, these two indicators should be comprehensively considered when calculating the quality qualification coefficient.

[0065] The above contents are merely examples and explanations of the concept of the present invention. Those skilled in the art may make various modifications or additions to the specific embodiments described or replace them in a similar manner. As long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, they shall all fall within the protection scope of the present invention.

Claims

1. The PCB surface defect detection system based on image recognition is characterized by: include: The image quality control module is used to place the PCB boards produced in the current batch of the target factory in the image acquisition area, turn on the lighting system, collect the grayscale images of each PCB board under the lighting system, and evaluate whether the quality of the corresponding collected images of each PCB board meets the standard. If not, adjust the lighting system. If it meets the standard, execute the surface defect type confirmation module; A surface defect type confirmation module is used to record the PCB boards that meet the image quality standards as target PCB boards, build a three-dimensional information model of each target PCB board according to the corresponding collected images of each target PCB board, extract PCB board information from the three-dimensional information model, and confirm the surface defect type corresponding to each target PCB board; The surface defect report feedback module is used to calculate the surface defect severity index of each target PCB board, evaluate the surface defect severity level corresponding to each target PCB board, generate the surface defect report corresponding to each target PCB board according to the surface defect type and surface defect severity level corresponding to each target PCB board, and feedback the surface defect report; A database for storing the standard position of each hole in each target PCB board and the surface defect severity index interval corresponding to each surface defect severity level; The PCB board quality feedback module is used to analyze the quality qualification coefficient of the PCB boards produced by the current batch of the target factory. If it is less than the set reference PCB board quality qualification coefficient, it indicates that the quality of the PCB boards produced by the current batch of the target factory is unqualified, and feedback is given.

2. The PCB surface defect detection system based on image recognition according to claim 1, characterized in that: The specific process of evaluating whether the quality of the images collected by each PCB board meets the standards is as follows: A1. Extract the grayscale value corresponding to each grayscale area from the grayscale image of each PCB board under the lighting system, and calculate the average value to obtain the average grayscale value of each PCB board under the lighting system, which is recorded as Wherein, i represents the number of the PCB board, i=1,2,...,n; A2. Extract the minimum value of the grayscale values ​​corresponding to each grayscale area from the grayscale image of each PCB board under the lighting system and record it as A3. Calculate the image quality coefficient β corresponding to each PCB board i , A4. Compare the image quality coefficient corresponding to each PCB board with the set reference image quality coefficient. If the image quality coefficient corresponding to a PCB board is greater than or equal to the set reference image quality coefficient, it indicates that the image quality corresponding to the PCB board meets the standard. Otherwise, it indicates that the image quality corresponding to the PCB board does not meet the standard.

3. The PCB surface defect detection system based on image recognition according to claim 2, characterized in that: The specific process of adjusting the lighting system is as follows: marking a PCB board whose image quality does not meet the standard as a PCB board with abnormal image, moving the position of the lighting lamp in the lighting system, and after the position of the lighting lamp is moved, re-collecting the grayscale image of each PCB board with abnormal image, and evaluating whether the quality of the image collected corresponding to each PCB board with abnormal image meets the standard in the same way as the evaluation method of whether the quality of the image collected corresponding to each PCB board meets the standard; if it still does not meet the standard, continue to move the position of the lighting lamp in the lighting system, and repeat the above steps until the quality of the image collected corresponding to the PCB boards with abnormal image meets the standard.

4. The PCB surface defect detection system based on image recognition according to claim 1, characterized in that: The PCB board information includes the positions of each hole, the position of the center point of the PCB board, the positions of each characteristic point, the number of surface cracks, the length of each crack, the number of dirt locations, and the dirt area corresponding to each dirt location.

5. The PCB surface defect detection system based on image recognition according to claim 4 is characterized in that: The specific process of confirming the surface defect type corresponding to each target PCB board is as follows: B1. Extract the hole positions, PCB center position, feature point positions, surface crack number, crack length, dirt number and dirt area corresponding to each dirt from the PCB information corresponding to each target PCB, and calculate the hole position deviation ψ of each target PCB respectively. j , Flatness Cracking degreeω j and dirtiness x j , where j represents the number of the target PCB board, j = 1, 2, ..., m; B2. Establish a surface defect type diagnosis model corresponding to each target PCB board and output the surface defect type diagnosis result. The surface defect type diagnosis model is expressed as: Among them, P1, P2, P3, P4 and P5 represent the evaluation conditions of the diagnosis results of each surface defect type, P1 represents ψ j ≥ψ′, P2 means P3 represents ω j ≥ω′, P4 represents χ j ≥χ′, P5 indicates Among them, ψ′, ω′ and χ′ represent the hole position deviation, flatness, cracking degree and dirt degree of the set reference, respectively.

6. The PCB surface defect detection system based on image recognition according to claim 5 is characterized in that: The specific process of calculating the hole position deviation of each target PCB board is as follows: C1. Extract the standard position of each hole in each target PCB board from the database, and bring it into the corresponding three-dimensional information model to obtain the standard position coordinates of each hole in each target PCB board, recorded as Wherein, g represents the hole number, g = 1, 2, ..., r; C2. Bring the hole positions corresponding to each target PCB board into the corresponding three-dimensional information model to obtain the coordinates of the hole positions corresponding to each target PCB board, recorded as (x jg ,y jg ,z jg ); C3. Calculate the position deviation of each hole in each target PCB board δ jg , Wherein, Δx′, Δy′ and Δz′ represent the position offset values ​​corresponding to the x-axis, y-axis and z-axis respectively; C4. Calculate the hole position deviation of each target PCB board ψ j , Wherein, r represents the number of holes.

7. The PCB surface defect detection system based on image recognition according to claim 5, characterized in that: The specific process of calculating the flatness of each target PCB board is as follows: D1. Bring the center point position of each target PCB board into the corresponding three-dimensional information model to obtain the y-axis value of the center point of each target PCB board, and record it as D2. Bring the positions of the feature points corresponding to each target PCB board into the corresponding three-dimensional information model to obtain the y-axis values ​​of the feature points corresponding to each target PCB board, recorded as y jp , where p represents the number of the feature point, p = 1, 2, ..., q; D3. Calculate the flatness of each target PCB board Among them, H′ represents the allowed y-axis height difference, q represents the number of feature points, and y j(p+1) Indicates the y-axis value of the p+1th feature point corresponding to the jth target PCB board.

8. The PCB surface defect detection system based on image recognition according to claim 5, characterized in that: The formula for calculating the surface defect severity index of each target PCB board is: Among them, θ j represents the surface defect severity index of the j-th target PCB board, a1, a2, a3 and a4 represent the weights of the surface defect severity index assessment corresponding to the set hole position offset, flatness, cracking degree and dirt degree, a1+a2+a3+a4=1, and e represents a natural constant.

9. The PCB surface defect detection system based on image recognition according to claim 8, characterized in that: The method for evaluating the surface defect severity level corresponding to each target PCB board is as follows: comparing the surface defect severity index of each target PCB board with the surface defect severity index interval corresponding to each surface defect severity level stored in the database; if the surface defect severity index of a certain target PCB board is within the surface defect severity index interval corresponding to a certain surface defect severity level, then the surface defect severity level is used as the surface defect severity level corresponding to the target PCB board, thereby obtaining the surface defect severity level corresponding to each target PCB board, wherein the surface defect severity level includes no defect, slight defect, general defect and severe defect.

10. The PCB surface defect detection system based on image recognition according to claim 9, characterized in that: The specific process of analyzing the quality qualification coefficient of the PCB boards produced in the current batch of the target factory is as follows: E1. Count the number of target PCB boards with no surface defect severity level produced by the target factory in the current batch, and record it as τ 无 ; E2. Count the number of target PCB boards with general defects and severe defects in the current batch of target factory production respectively, and add them up to get the number of unqualified PCB boards in the current batch of target factory production, which is recorded as τ. 不 ; E3. Calculate the quality qualification coefficient ζ of the current batch of PCB boards produced by the target factory. Among them, m represents the number of target PCB boards, K1 and K2 represent the proportion of defect-free PCB boards and the proportion of unqualified PCB boards of the set reference, respectively, a5 and a6 represent the weights of the proportion of defect-free PCB boards and the proportion of unqualified PCB boards corresponding to the proportion of PCB board quality qualification assessment, respectively, a5+a6=1.

Images