TWI757825B - System and method for pcb inspection based on false defect detection - Google Patents

Abstract

System and method for PCB inspection based on false defect detection，the system includes automatic optical inspection equipment, database server and inspection equipment; the automatic optical inspection equipment is used to scan a printed circuit board which is to be inspected to obtain a scanned image and compare it with the corresponding standard images loaded by a database server, so as to construct a defect list containing defect coordinate information corresponding to a preliminarily determined defect of the scanned image; the maintenance equipment can load the scanned image and the corresponding defect list through the database server, and check the preliminarily determined defects at each defect coordinate in the defect list of the scanned image one by one; if the re-examined defect is a false defect, the defect will be deleted from the defect list, and the maintenance equipment will repair the printed circuit board according to remaining defects in the corresponding defect list.

Description

PCB maintenance system and maintenance method based on false point defect detection

The invention relates to the field of circuit board inspection and repair, in particular to a PCB inspection system and repair method based on false point defect detection.

In today's highly developed electronic industry era, Printed Circuit Board (PCB) has become one of the indispensable and important components in computers, electronic communications and other products. In the production process of printed circuit boards, there will be defects of short circuit or open circuit, and the quality of printed circuit boards determines the qualification of corresponding electronic device products. Therefore, the quality inspection and maintenance of printed circuit boards are particularly important.

In the prior art, automated optical inspection equipment (Automated Optical Inspection, AOI) is commonly used in the production process of circuit boards. AOI can detect defects on the PCB, and then manually perform maintenance according to the defects detected by AOI. Nowadays, customers not only have requirements for the work efficiency of AOI itself, but also have higher and higher requirements for the work efficiency of AOI inspection and maintenance. At present, the common AOI suppliers in the market can only provide independent AOI equipment to be inspected. After the PCB gets the defect coordinates from the AOI equipment, it is moved to the maintenance equipment. According to the defect coordinates, the defects are manually repaired one by one through the maintenance equipment. a lot of time.

A solution for improving PCB defect detection and maintenance is lacking in the prior art.

In order to solve the problems of the prior art, the present invention provides a PCB inspection system and inspection method based on false point defect detection, which greatly improves the PCB defect inspection and inspection efficiency. The technical scheme is as follows:

In one aspect, the present invention provides a PCB overhaul system based on false point defect detection, including automatic optical inspection equipment, a database server and an overhaul equipment, wherein the overhaul equipment is configured with a defect virtual detection module for verifying false point defects, the The automatic optical detection equipment and the defect virtual detection module are all connected in communication with the database server;

The automatic optical inspection equipment is used to scan the printed circuit board to be inspected to obtain the scanned image, and compare it with the corresponding standard image loaded through the database server, so as to construct a defect list. The defect coordinate information of the defects initially determined in the scanned image;

The database server is used for storing the scanned images and the corresponding defect list output by the automatic optical inspection equipment;

The defect virtual detection module of the maintenance equipment can load the scanned image and the corresponding defect list through the database server, and carry out one-by-one preliminary determination of defects at each defect coordinate of the scanned image in the defect list. Re-inspection, if the re-inspection defect is a false point defect, the defect is deleted from the defect list, and the inspection equipment repairs the defects at the remaining defect coordinates in the corresponding defect list of the printed circuit board.

As a first optional technical solution, re-inspecting the initially determined defect includes: extracting a local image at the defect coordinates corresponding to the initially determined defect, and judging whether the local image satisfies the short-circuit feature or the open-circuit feature, wherein all the The short-circuit feature includes a straight line connecting two cables, and the open-circuit feature includes a gap in the cable. If any one of the characteristics is satisfied, the defect is determined to be a real defect, otherwise, the defect is determined to be a false point defect.

As a second optional technical solution, re-inspecting the initially determined defect includes: extracting a partial image at the defect coordinates corresponding to the initially determined defect, and judging whether the partial image satisfies the following conditions at the same time: non-linear, irregular And for an isolated figure, if the above characteristics are met at the same time, the defect is determined to be a false point defect.

As a third optional technical solution, the re-inspection of the initially determined defects includes:

Load several preset defect template images through the database server, and the defect template images are marked as real defects or false point defects;

Extracting the partial image at the defect coordinates corresponding to the initially determined defect, and comparing the similarity with the defect template image to find the defect template image with the highest similarity;

If the defect template image with the highest similarity is marked as a real defect, the preliminary determined defect is determined as a real defect; if the defect template image with the highest similarity is marked as a false point defect, the preliminary determined defect is determined as a false point defect. The identified defects are false point defects.

As a fourth optional technical solution, re-inspecting the initially determined defect includes: extracting a partial image at the defect coordinate corresponding to the initially determined defect, inputting it into the trained neural network model, according to the neural network model As a result of the output, it is determined whether the defect is a real defect or a false point defect.

Further, scanning the printed circuit board to be inspected includes scanning the PCB with different viewing angles to obtain different viewing angles, and the viewing angles include two-dimensional viewing and three-dimensional viewing.

Further, the maintenance equipment further includes a movable camera device, the camera device can be moved to the remaining defect coordinates in the corresponding defect list of the printed circuit board, and the defects at the defect coordinates can be enlarged and displayed, for manual maintenance.

Further, the number of the database server is one, the number of the automatic optical detection equipment and the maintenance equipment is multiple, and the number of the automatic optical detection equipment and the maintenance equipment is the same or different.

On the other hand, the present invention provides a PCB repair method based on false point defect detection, comprising the following steps:

Scan the printed circuit board to be tested to obtain a scanned image;

Comparing it with the standard image of the printed circuit board, taking the difference as a preliminarily determined defect and constructing a defect list, the defect list including defect coordinate information corresponding to the preliminarily determined defect of the scanned image;

Rechecking the initially determined defects of the scanned image at each defect coordinate in the defect list one by one, if the rechecking defect is a false point defect, delete the defect from the defect list;

Repair the defects at the remaining defect coordinates in the corresponding defect list of the printed circuit board.

Further, re-inspecting each preliminary determined defect includes the following steps:

Extract the partial image at the defect coordinates corresponding to the preliminary judged defect, and judge the partial image according to any one or more of the following methods:

The first way is to determine whether the local image satisfies the short-circuit feature or the open-circuit feature, wherein the short-circuit feature includes a straight line connecting two cables, and the open-circuit feature includes a gap in the cable. feature, the defect is determined to be a real defect, otherwise the defect is determined to be a false point defect;

The second method is to judge whether the local image meets the following conditions at the same time: non-linear, irregular and isolated graphics, if the above characteristics are met at the same time, then the defect is judged to be a false point defect;

The third method is to compare the similarity between the partial image and several preset defect template images marked as real defects or false point defects, and determine the defect according to the calibration of the defect template image with the highest similarity Whether it is a real defect or a false point defect;

The fourth method is to input the partial image into the neural network model after training, and determine whether the defect is a real defect or a false point defect according to the output result of the neural network model.

The present invention has the following beneficial effects:

a. Investigate the false point defects detected by AOI, eliminate the false point defects that do not need maintenance, and then carry out maintenance, which greatly improves the maintenance efficiency;

b. Connect the AOI with the maintenance equipment through the database server to achieve efficient data transmission;

c. The maintenance equipment is equipped with a movable camera device to locate and magnify the real defects after eliminating the false point defects, so as to improve the efficiency of manual maintenance;

d. Multiple AOI devices are configured with one database server to connect with multiple maintenance devices, saving space and cost.

In the following detailed description, numerous specific details are set forth in order to provide a thorough understanding of the present invention. However, one skilled in the art will understand that the present invention may be practiced without these specific details. In other instances, well-known methods, procedures and components have not been described in detail so as not to obscure the present invention.

The subject matter that is regarded as the invention is particularly pointed out and distinctly claimed in the concluding portion of the specification. However, the organization, method of operation, and subject matter, features, and advantages of the present invention are best understood by reference to the following detailed description, when read in conjunction with the accompanying drawings.

It will be appreciated that for simplicity and clarity of illustration, elements shown in the figures have not necessarily been drawn to scale. For example, the dimensions of some elements may be exaggerated relative to other elements for clarity.

Since the illustrative embodiments of the invention may be practiced to a large extent using electronic components and circuits well known to those skilled in the art, as described above, no further detail will be explained beyond what is deemed necessary , in order to understand and appreciate the basic concepts of the present invention, so as not to obscure or distract the teachings of the present invention.

This paper provides a PCB repair system based on false point defect detection, see Figure 1, the PCB inspection system based on false point defect detection includes automatic optical inspection equipment (the following AOI), database server (data base Server) and maintenance equipment , a defect virtual detection module (also called virtual verification and detection module, Virtual Verification and Repair Station, VVR) for verifying false point defects is configured on the maintenance equipment, and the automatic optical detection equipment and the defect virtual detection module are all the same as the above Database server communication connection;

The automatic optical inspection equipment is used to scan the printed circuit board to be inspected to obtain the scanned image, and compare it with the corresponding standard image loaded through the database server. A defect list, the defect list includes defect coordinate information corresponding to the preliminary determined defects of the scanned image;

The database server is used for storing the scanned images and the corresponding defect list output by the automatic optical inspection equipment;

The defect virtual detection module of the maintenance equipment can load the scanned image and the corresponding defect list through the database server, and carry out one-by-one preliminary determination of defects at each defect coordinate of the scanned image in the defect list. Re-inspection, if the re-inspection defect is a false point defect, the defect is deleted from the defect list, and the inspection equipment repairs the defects at the remaining defect coordinates in the corresponding defect list of the printed circuit board.

As shown in Figure 1, the AOI equipment can obtain the overall layout picture of the defect after scanning the PCB board, and can accurately calibrate the coordinates of the corresponding defect point in the picture. In the AOI equipment system, it also has the function of judging the type of defect , such as circuit board missing soldering, excessive soldering and soldering errors, etc. Connected with AOI is a database server with data storage function, the database server can accurately store the information input after AOI scanning, connected with the database server is the VVR system for overhauling equipment, VVR collects the defect information of the corresponding plate in the database server , Through its own intelligent judgment system or artificial picture inspection, it can accurately judge the "false point" information and "false point" coordinate information in the defect information, and then delete the judged "false point" information through operation. After the "false point", move to the corresponding "true point" defect coordinate position through the Video on the VVR device for manual inspection.

Compared with all the defect points scanned by AOI before, all the defect points need to be repaired by a separate maintenance equipment, and the method of manually overhauling the defect points one by one reduces a lot of work wasted in the processing of "false points", which not only improves the work. Efficiency, and avoids manual misjudgment of "false points" in maintenance.

In a preferred embodiment of the present invention, different viewing angles can be used to scan the PCB to obtain different viewing angles, such as a two-dimensional view of a certain contrast, saturation, and hue (such as Figure 5) or a 3D visual image ( For example, Figure 6), especially the 3D scanning vision shown in Figure 6, can accurately determine "false points" and "true points", which improves the accuracy of judgment, and does not delete "false points" by mistake. Moreover, it can also provide image reference for subsequent manual repair, which is more convenient for manual repair.

As a first optional technical solution, using the exclusion method to re-examine the initially determined defect includes: extracting a partial image at the defect coordinates corresponding to the initially determined defect, and judging whether the partial image satisfies the short-circuit feature or the open-circuit feature, Wherein, the short-circuit feature includes a straight line connecting two cables (as shown in Figure 2), and the open-circuit feature includes a gap (not shown) on the cable. If any one of the characteristics is satisfied, it is determined that the The defect is a real defect, otherwise it is determined that the defect is a false point defect. "False point defect" is misjudged as a defect in the preliminary judgment process, and does not need manual maintenance. Therefore, it needs to be deleted from the defect list; "true defect" needs to be manually repaired point by point, as shown in Figure 2. The narrow gap will lead to a short circuit of the PCB, and then the narrow gap needs to be removed manually.

As a second optional technical solution, using the feature correspondence method to re-inspect the initially determined defect includes: extracting a partial image at the defect coordinates corresponding to the initially determined defect, and judging whether the partial image satisfies the following conditions at the same time: For straight, irregular and isolated graphics (as shown in Figure 3), if the above characteristics are met at the same time, the defect is determined to be a false point defect. The "false point defects" can be dust, stains, or fingerprints, etc., which will exist in large quantities in the PCB board, and will be judged as defect points during AOI scanning. On these large numbers of "false point defects", the embodiment of the present invention introduces a VVR system, which can greatly reduce the time spent in this aspect.

As a third optional technical solution, using the similarity matching method to re-examine the initially determined defects includes:

Load several preset defect template images through the database server, and the defect template images are marked as real defects or false point defects;

Extracting the partial image at the defect coordinates corresponding to the initially determined defect, and comparing the similarity with the defect template image to find the defect template image with the highest similarity;

If the defect template image with the highest similarity is marked as a real defect, the preliminary determined defect is determined as a real defect; if the defect template image with the highest similarity is marked as a false point defect, the preliminary determined defect is determined as a false point defect. The identified defects are false point defects.

As a fourth optional technical solution, re-inspecting the initially determined defect includes: extracting a partial image at the defect coordinate corresponding to the initially determined defect, inputting it into the trained neural network model, according to the neural network model As a result of the output, it is determined whether the defect is a real defect or a false point defect. Wherein, the neural network model can be a deep neural network in the prior art, and the neural network can be trained in combination with a back-propagation algorithm and a stochastic gradient descent method. The application of the neural network in the field of image recognition is the prior art, and the specific model of the neural network and the related algorithm itself can be understood as adopting the prior art.

It should be noted that, in order to further improve the detection accuracy of false point defects, the above four technical solutions for re-inspecting the initially determined defects can be used in combination, such as using the combination of the first and third technical solutions, Only when both determine that the current defect is a false point defect, the current defect can be regarded as a false point defect and then deleted from the defect list.

It is even possible to combine the four technical solutions to judge whether the current defect is a real defect or a false point defect. Obviously, if the judgment results of the four technical solutions are all false point defects, the accuracy of the judgment that the current defect is a false point defect is greatly improved; or, at least three (or two) of the judgment results of the four technical solutions are judged as If there is a false point defect, the current defect is regarded as a false point defect and then deleted from the defect list.

In a preferred embodiment of the present invention, the maintenance equipment further includes a movable camera device, the camera device can move to the remaining defect coordinates in the corresponding defect list of the printed circuit board, and monitor the defect coordinates The defects at the place are enlarged and displayed for manual maintenance. In this embodiment, the camera device has two functions. The first is to move to the relative position of the defect to be repaired, so as to prompt the maintenance personnel to repair the PCB at the current relative position of the camera device to avoid missed repairs; The second is that the camera device can magnify and display the current defective area at a high magnification, so that the maintenance personnel can clearly and quickly determine the current defect to be repaired and avoid wrong inspection.

In a preferred embodiment of the present invention, as shown in FIG. 4 , the number of the database server is one, and the number of the automatic optical inspection equipment and the maintenance equipment is multiple. Configure a set of database server for multiple AOI devices to connect with multiple VVR systems. The database server can only use one set at the customer. It can work with multiple AOIs and VVRs at the same time. The database server can not only collect single or multiple Piece PCB defect information, can also collect multiple AOI scanning PCB defect information, so that a system that can work through a database server can save space and cost for customers. Obviously, if the automatic optical detection equipment and/or maintenance equipment is only one set, it also belongs to the protection scope claimed by the present invention, and multiple automatic optical detection equipment and maintenance equipment are only preferred embodiments, and are not necessarily limited. For the case of multiple automatic optical inspection equipment and maintenance equipment, the number of the two can be different, that is, there is not necessarily a corresponding relationship between the two.

In one embodiment of the present invention, a PCB repair method based on false point defect detection is provided, as shown in FIG. 7 , the repair method includes the following steps:

S1. Scan the printed circuit board to be tested to obtain a scanned image;

S2, compare it with the standard image of the printed circuit board, take the difference as a preliminary judged defect and construct a defect list, the defect list contains the defect coordinate information corresponding to the preliminary judged defect of the scanned image;

S3. Start traversing the defect list, such as re-inspecting the initially determined defects at the coordinates of the first defect in sequence;

S4. If the result of the re-inspection is that the defect is a false point defect, execute S5; otherwise, execute S6;

S5, delete the false point defect obtained by rechecking from the defect list;

S6, determine whether to complete the traversal of the defects in the defect list, if completed, execute S7, otherwise, traverse the defect at the next defect coordinate in the defect list and continue to execute S4;

S7, repair the defects at the remaining defect coordinates in the corresponding defect list of the printed circuit board.

As described in the above embodiment, the re-inspection of each initially determined defect includes the following steps:

Extract the partial image at the defect coordinates corresponding to the preliminary judged defect, and judge the partial image according to any one or more of the following methods:

The first way is to determine whether the local image satisfies the short-circuit feature or the open-circuit feature, wherein the short-circuit feature includes a straight line connecting two cables, and the open-circuit feature includes a gap in the cable. feature, the defect is determined to be a real defect, otherwise the defect is determined to be a false point defect;

The second method is to judge whether the local image meets the following conditions at the same time: non-linear, irregular and isolated graphics, if the above characteristics are met at the same time, then the defect is judged to be a false point defect;

The third method is to compare the similarity between the partial image and several preset defect template images marked as real defects or false point defects, and determine the defect according to the calibration of the defect template image with the highest similarity Whether it is a real defect or a false point defect;

The fourth method is to input the partial image into the neural network model after training, and determine whether the defect is a real defect or a false point defect according to the output result of the neural network model.

The above four methods refer to the detailed description of the above-mentioned system embodiments, and the method for determining according to various methods also refers to the specific descriptions that the four technical solutions can be combined to determine in the above-mentioned barrel embodiments, which will not be repeated here.

In the present invention, the defect coordinates and scanned images (preferably, including the prediction type of the defect) after AOI detection are connected to the maintenance equipment through the database server, and the maintenance equipment is also equipped with a high-end defect virtual detection module. The defect virtual detection module automatically filters defect points, the system classifies all gray-scale defect images (defective scanned images) as "true", or "false", "false" defects are deleted, the "false" Defects may be dust, dirt, etc., so they can be deleted from the maintenance defect list; or by manually inspecting the gray-scale defect map provided by AOI, false defects can be deleted from the maintenance defect list. Then the "real" defects will be magnified by the camera and checked manually, which can save the inspection time of the maintenance equipment for the "false" defect points. Realizing the transformation of Industry 4.0 not only saves a lot of manual maintenance time and labor costs, but also reduces the probability of manual misjudgment.

The present invention checks the false point defects detected by the AOI, removes the false point defects that do not need maintenance, and then performs the maintenance, thereby greatly improving the maintenance efficiency; the AOI is connected with the maintenance equipment through the database server to realize efficient data transmission; the maintenance equipment is set The movable camera device locates and magnifies the real defects after eliminating the false point defects, improving the efficiency of manual maintenance; multiple AOI equipments are equipped with a database server to connect with multiple maintenance equipments, saving space and cost.

Furthermore, those skilled in the art will appreciate that the boundaries between the operations described above are exemplary only. Multiple operations may be combined into a single operation, a single operation may be distributed among additional operations, and operations may be performed at at least partially overlapping times. Furthermore, alternative embodiments may include multiple illustrations of particular operations, and the order of operations may vary in various other embodiments.

However, other modifications, changes and substitutions are also possible. Accordingly, the specification and drawings are to be regarded in an illustrative rather than a restrictive sense.

In the claim statement, any reference signs placed between parentheses shall not be construed as limiting the claim. The word "comprising" does not exclude the presence of other elements or steps listed in the claim statement. Also, the terms "a" or "an" as used herein are defined as one or more than one. Unless stated otherwise, terms such as "first" and "second" are used to arbitrarily distinguish between the elements these terms describe. Accordingly, these terms are not necessarily intended to denote the timing or other priority of these elements. The mere fact that certain measures are recited in mutually different claims does not mean that a combination of these measures cannot be used.

While certain features of the invention have been illustrated and described herein, many modifications, substitutions, changes and equivalents will now occur to those of ordinary skill in the art. Therefore, it should be understood that the scope of the appended claims is intended to cover all such modifications and changes as fall within the true spirit of this invention.

without

The subject matter that is regarded as the invention is particularly pointed out and distinctly claimed in the concluding portion of the specification. However, the organization, method of operation, and subject matter, features, and advantages of the present invention are best understood by reference to the following detailed description when read in conjunction with the accompanying drawings, wherein: 1 is a schematic structural diagram of a PCB inspection system based on false point defect detection provided by an embodiment of the present invention; FIG. 2 is a characteristic schematic diagram of a real short-circuit defect scanning image provided by an embodiment of the present invention; 3 is a characteristic schematic diagram of a false point defect scanning image provided by an embodiment of the present invention; 4 is a schematic structural diagram of a single database server corresponding to multiple AOIs and multiple VVRs provided by an embodiment of the present invention; 5 is a two-dimensional perspective view obtained by scanning a PCB with a two-dimensional perspective provided by an embodiment of the present invention; 6 is a three-dimensional perspective view obtained by scanning a PCB with a three-dimensional perspective provided by an embodiment of the present invention; FIG. 7 is a flowchart of a PCB repair method based on false point defect detection according to an embodiment of the present invention.

This representative picture is a flow chart, so there is no symbol for simple description

Claims (2)

A PCB repair system based on false point defect detection is characterized in that it includes automatic optical inspection equipment, a database server and repair equipment, the inspection equipment is equipped with a defect virtual detection module for verifying false point defects, and the automatic optical inspection The equipment and the virtual defect detection module are all connected in communication with the database server, the number of the database server is one, the number of the automatic optical detection equipment and the maintenance equipment is multiple, and the number of the automatic optical detection equipment and the maintenance equipment is The same or different; the automatic optical inspection equipment is used for scanning the printed circuit board to be inspected to obtain a scanned image, including scanning the PCB with different viewing angles to obtain different viewing angles, and the viewing angles include two-dimensional viewing and Three-dimensional perspective view, after the automatic optical inspection equipment scans the printed circuit board, the overall layout picture of the defect can be obtained, and the coordinates of the corresponding defect point can be accurately calibrated in the picture; and the corresponding standard image loaded through the database server The image is compared to construct a defect list, the defect list contains defect coordinate information corresponding to the preliminary determined defects of the scanned image; the database server is used for storing the scanned image output by the automatic optical inspection device and the corresponding defect list; the defect virtual detection module of the maintenance equipment can load the scanned image and the corresponding defect list through the database server, and perform a preliminary analysis of the scanned image at each defect coordinate in the defect list. The determined defects are re-inspected one by one. The maintenance equipment also includes a movable camera device. If the re-inspected defects are false point defects, the The defect is deleted from the defect list, the inspection equipment repairs the defects at the remaining defect coordinates in the defect list corresponding to the printed circuit board, and the camera device can be moved to the defect list corresponding to the printed circuit board. At the remaining defect coordinates, the defects at the said defect coordinates are enlarged and displayed for manual maintenance; wherein, the re-inspection of the initially determined defects includes: combining any two, three or four of the following Judgment method, there are at least two ways to judge the current defect as a false point defect, and then the current defect can be regarded as a false point defect and then deleted from the defect list: the first method is to extract the defect coordinates corresponding to the defect in the preliminary judgment. A partial image to determine whether the partial image satisfies the short-circuit feature or the open-circuit feature, wherein the short-circuit feature includes a straight line connecting two cables, and the open-circuit feature includes a gap in the cable. If any one of the characteristics is satisfied , then it is determined that the defect is a real defect, otherwise it is determined that the defect is a false point defect; the second method is to extract the local image at the defect coordinates corresponding to the defect that is initially determined, and determine whether the local image satisfies the following conditions at the same time : Non-straight, irregular and isolated graphics. If the above characteristics are met at the same time, the defect is determined to be a false point defect; the third method is to load several preset defect template images through the database server. The template image is marked as a real defect or a false point defect; extract the local image at the defect coordinates corresponding to the initially determined defect, and compare it with the defect template image to find the one with the highest similarity. Defect template image; if the defect template image with the highest similarity is marked as a real defect, the initially determined defect is determined as a real defect; if the defect template image with the highest similarity is marked as a false point defect , the defect of the preliminary judgment is determined to be a false point The fourth method is to extract the local image at the defect coordinates corresponding to the initially determined defect, input it into the trained neural network model, and determine whether the defect is a real defect or a real defect according to the output of the neural network model False point defect. A method for overhauling a PCB based on false point defect detection is characterized in that it includes the following steps: scanning a printed circuit board to be detected to obtain a scanned image, including scanning the PCB with different viewing angles to obtain views of different viewing angles, and the viewing angles are The view includes two-dimensional view and three-dimensional view. After scanning the printed circuit board, a picture of the overall layout of the defect can be obtained, and the coordinates of the corresponding defect point can be accurately calibrated in the picture; it is compared with the standard image of the printed circuit board. comparing, taking the difference as a preliminary judged defect and constructing a defect list, the defect list including the defect coordinate information corresponding to the preliminary judged defect of the scanned image; for each of the scanned image in the defect list The preliminary determined defects at the defect coordinates are re-inspected one by one. If the re-inspection defect is a false point defect, the defect is deleted from the defect list; the remaining defect coordinates in the corresponding defect list of the printed circuit board are During the maintenance process, the camera device is used to move to the remaining defect coordinates in the corresponding defect list of the printed circuit board, and the defects at the defect coordinates are enlarged and displayed for manual maintenance; The one-by-one re-inspection of the initially determined defects includes: using any two of the following in combination Or three or four judgment methods, there are at least two methods to determine the current defect as a false point defect, and then the current defect can be regarded as a false point defect and then deleted from the defect list: The first method is to extract the corresponding defects of the preliminary judgment. The local image at the defect coordinates is determined to determine whether the local image satisfies the short-circuit feature or the open-circuit feature, wherein the short-circuit feature includes a straight line connecting two cables, and the open-circuit feature includes a gap in the cable, If any one of the characteristics is satisfied, the defect is determined to be a real defect, otherwise, the defect is determined to be a false point defect; the second method is to extract the local image at the defect coordinates corresponding to the initially determined defect, and determine the local image. Whether the following conditions are met at the same time: non-linear, irregular and isolated graphics, if the above characteristics are met at the same time, the defect is determined to be a false point defect; the third method is to load several preset defect template images through the database server image, the defect template image is marked as a real defect or a false point defect; extract the local image at the defect coordinates corresponding to the defect that is initially determined, and compare it with the defect template image to find the The defect template image with the highest similarity; if the defect template image with the highest similarity is marked as a real defect, the preliminary determined defect is determined as a real defect; if the defect template image with the highest similarity is marked as a real defect If it is marked as a false point defect, it is determined that the preliminary judgment defect is a false point defect; the fourth method is to extract the local image at the defect coordinates corresponding to the preliminary judgment defect, and input it into the neural network model that has completed the training. According to the output result of the neural network model, it is determined whether the defect is a real defect or a false point defect.

Images