Image inspection device for electronic component, image inspection method, and image inspection program

Abstract

This image inspection device 100 for an electronic component 1 comprises: a position detecting means 10 that detects a product position PP on the basis of a product image P; an image correcting means 20 that generates a corrected product image Pc obtained by correcting the product image P on the basis of the product position PP; a first inspecting means 30 that executes a first inspection T1 having a first inspection logic L1 on the basis of the corrected product image Pc and outputs a first inspection result T1R; a relief inspection means 40 that executes a relief inspection Tk having a relief inspection logic Lk different from the first inspection logic L1, when the first inspection result T1R satisfies a specific condition in the first inspection logic L1, and outputs a relief inspection result TkR; and a comprehensive determining means 50 that performs a comprehensive determination of the electronic component 1 on the basis of the first inspection result T1R and the relief inspection result TkR.

Description

Image inspection device, image inspection method, and image inspection program for electronic components

[0001] The present invention relates to an image inspection device, an image inspection method, and an image inspection program for electronic components.

[0002] Conventionally, there have been devices that perform visual inspections in which an inspection A is performed to photograph the product and detect the product position, followed by inspections B and C that are specialized for detecting various visual defects, and only if the product is judged to be non-defective in either inspection is there an overall judgment that the product is non-defective (for example, Patent Document 1 and Patent Document 2).

[0003] In conventional visual inspections, each inspection has its own unique image inspection standard, and the inspection standard is set so that products with specific types of visual defects are always judged as failing (NG). Furthermore, conventional image inspection standards could only set simple inspection standards that produce one inspection result per inspection. However, there are also defective products that are not significantly different from good products in terms of image quality. Even such defective products must be rejected in visual inspection. However, each inspection has overly strict inspection standards, which can result in some products that should be considered good being mistakenly judged as defective.

[0004] Japanese Patent No. 6999150 Japanese Patent No. 7298825

[0005] The present invention aims to provide an image inspection device, an image inspection method, and an image inspection program for electronic components that can prevent products that are intended to be non-defective from being determined to be defective.

[0006] (1) An image inspection device for electronic components according to one aspect of the present invention includes: a position detection unit that detects a product position based on a product image of an electronic component; an image correction unit that corrects the product image based on the product position to generate a corrected product image; a first inspection unit that performs a first inspection having a first inspection logic based on the corrected product image and outputs a first inspection result; a repair inspection unit that performs a repair inspection having a repair inspection logic different from the first inspection logic when the first inspection logic is a pass / fail judgment and the first inspection result is a fail, and outputs a repair inspection result; and a comprehensive judgment unit that makes an overall judgment on the electronic component based on the first inspection result and the repair inspection result, wherein the first inspection logic is a function that judges pass or fail based on a threshold value set for a parameter for evaluating a defect, and the repair inspection logic is generated by artificial intelligence that learns based on a relationship between information on the product image or the corrected product image and the inspection result. (2) In the above (1), the first inspection may be an inspection to detect dirt or foreign matter on an element body among defects of the electronic component. (3) In (1) or (2) above, the comprehensive judgment means may make a comprehensive judgment as a good product when the first inspection result is pass or the repair inspection result is pass, and may make a comprehensive judgment as a defective product when the first inspection result is fail and the repair inspection result is fail. (4) In (1) or (2) above, a second inspection means may be provided that executes a second inspection having a second inspection logic different from the first inspection logic and outputs the second inspection result, and the comprehensive judgment means may make the comprehensive judgment based on the first inspection result, the repair inspection result, and the second inspection result. (5) In (4) above, the comprehensive judgment means may make a comprehensive judgment as a good product when the first inspection result is pass or the repair inspection result is pass and the second inspection result is pass, and may make a comprehensive judgment as a defective product when the first inspection result is fail and the repair inspection result is fail, or when the second inspection result is fail. (6) In the above (1) or (2), the first inspection may be an inspection to detect defects in the electronic component. (7) In the above (5), the second inspection may be an inspection to measure dimensions of the electronic component.(8) An image inspection method for electronic components according to one aspect of the present invention includes a position detection step for detecting a product position based on a product image of an electronic component, an image correction step for generating a corrected product image by correcting the product image based on the product position, a first inspection step for performing a first inspection having a first inspection logic based on the corrected product image and outputting a first inspection result, a repair inspection step for performing a repair inspection having a repair inspection logic different from the first inspection logic when the first inspection logic is a pass / fail judgment and the first inspection result is a fail, and outputting a repair inspection result, and a comprehensive judgment step for making an overall judgment on the electronic component based on the first inspection result and the repair inspection result, wherein the first inspection logic is a function for determining pass or fail based on a threshold value set for a parameter for evaluating a defect, and the repair inspection logic is generated by artificial intelligence that learns based on the relationship between information in the product image or the corrected product image and the inspection result. (9) An image inspection program for electronic components according to one aspect of the present invention causes a computer to execute a position detection function that detects a product position based on a product image of an electronic component, an image correction function that generates a corrected product image by correcting the product image based on the product position, a first inspection function that performs a first inspection having a first inspection logic based on the corrected product image and outputs a first inspection result, a repair inspection function that performs a repair inspection having a repair inspection logic different from the first inspection logic when the first inspection logic is a pass / fail judgment and the first inspection result is a fail, and outputs a repair inspection result, and a comprehensive judgment function that makes a comprehensive judgment on the electronic component based on the first inspection result and the repair inspection result, wherein the first inspection logic is a function that judges pass or fail based on a threshold value set for a parameter for evaluating a defect, and the repair inspection logic is generated by artificial intelligence that learns based on the relationship between information in the product image or the corrected product image and the inspection result.

[0007] According to the present invention, it is possible to provide an image inspection device, an image inspection method, and an image inspection program for electronic components that can prevent products that are intended to be non-defective from being determined to be defective.

[0008] It is an explanatory diagram for explaining an outline of an image inspection device for electronic parts, and is a diagram showing a flow of an image inspection method by the image inspection device for electronic parts.

[0009] (Embodiments) Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. Fig. 1 is an explanatory diagram illustrating an overview of an image inspection device 100 for an electronic component 1. Fig. 2 is a diagram showing the flow of an image inspection method using the image inspection device 100 for an electronic component 1. Note that, hereinafter, parts having common functions may be assigned the same reference numerals or symbols.

[0010] The image inspection device 100 for the electronic component 1 according to the embodiment is used to inspect the appearance of the electronic component 1 .

[0011] The electronic component 1 is a product that is the subject of inspection to determine whether it is a good product (OK) or a defective product (NG). The electronic component 1 is a component or element that constitutes an electronic device. The electronic component 1 is, for example, a resistor, a capacitor, an inductor, a transistor, a diode, an integrated circuit, etc. The electronic component 1 is used to control current or voltage in an electronic circuit, or to process and transfer information. The electronic component 1 is used in electronic devices such as mobile phones, computers, automobiles, and home appliances. The electronic component 1 has a three-dimensional shape, such as a hexahedron.

[0012] The image inspection device 100 includes an imaging means C and a control means S. The image inspection device 100 performs an appearance inspection by, for example, imaging an electronic component 1, which is a workpiece that is placed on the top surface of a rotating, disc-shaped, transparent glass conveying table and is then sucked and conveyed, from six sides (front, back, top, bottom, left side, and right side) using the imaging means C, such as a camera, and determining pass / fail of the captured product image P using the control means S.

[0013] The imaging means C appropriately includes a front camera that images the front side of the electronic component 1, a rear camera that images the rear side of the electronic component 1, a plane camera that images the plane of the electronic component 1, a bottom camera that images the bottom side of the electronic component 1, a right side camera that images the right side of the electronic component 1 (the inner surface facing the rotation center of the conveying table), and a left side camera that images the left side of the electronic component 1 (the outer surface facing away from the rotation center of the conveying table). The imaging means C generates a product image P by imaging the electronic component 1. The product image P is sent to the control means S.

[0014] The control means S is a computer system including a CPU, memories (storage media) such as RAM, ROM, and auxiliary storage devices, an external connection interface with input / output devices, a network interface, and a bus. The memory stores an image inspection program for causing the control means S (computer system) to execute the image inspection method according to this embodiment.

[0015] The CPU is an arithmetic circuit that controls the control means S. The CPU reads out programs stored in the ROM or auxiliary storage device into the RAM. The CPU executes various processes in the programs read out into the RAM. The ROM stores system programs and the like used to control the control means S. The auxiliary storage device stores application programs and the like that execute various processes. The auxiliary storage device is, for example, an HDD, SSD, etc. The external connection interface is an interface for connecting various devices to the control means S. The external connection interface connects, for example, an imaging means C, a display, a keyboard, etc. to the control means S. The network interface functions to communicate via a network based on the control of the CPU. The bus connects the above-mentioned functional units that make up the control means S so that they can communicate with each other.

[0016] The image inspection device 100 captures an image of the electronic component 1 using an imaging means C, and determines whether the electronic component 1 is pass / fail (whether the electronic component 1 is a good product or a defective product) using a control means S based on the captured product image P.

[0017] In detail, as shown in FIG. 2, the image inspection device 100 for the electronic component 1 according to the embodiment (specifically, the control means S of the image inspection device 100) includes a position detection means 10 that detects a product position PP based on a product image P obtained by capturing an image of the electronic component 1, an image correction means 20 that generates a corrected product image Pc by correcting the product image P based on the product position PP, a first inspection means 30 that executes a first inspection T1 having a first inspection logic L1 based on the corrected product image Pc and outputs a first inspection result T1R, a repair inspection means 40 that executes a repair inspection Tk having a repair inspection logic Lk different from the first inspection logic L1 when the first inspection result T1R satisfies a specific condition in the first inspection logic L1 and outputs a repair inspection result TkR, and a comprehensive judgment means 50 that makes a comprehensive judgment on the electronic component 1 based on the first inspection result T1R and the repair inspection result TkR. As a result, even if an electronic component 1 that should be judged as good (OK) is judged as failing by the first inspection means 30 (hereinafter, this phenomenon may be referred to as overkill), if it is judged as passing by the relief inspection means 40, it will be judged as good by the overall judgment means 50. Therefore, the image inspection device 100 for electronic components 1 can prevent products that should be good from being judged as defective.

[0018] The image inspection device 100 may include an inspection logic correction means. The inspection logic correction means is provided between the position detection means 10 or the image correction means 20 and the first inspection means 30 in the information transmission path. The inspection logic correction means corrects parameters of the first inspection logic L1, the second inspection logic L2, the third inspection logic L3, or the repair inspection logic Lk based on the product image P or the corrected product image Pc. The parameters of each of the inspection logics of the first inspection logic L1, the second inspection logic L2, the third inspection logic L3, or the repair inspection logic Lk may be, for example, a function that determines pass or fail based on a threshold value set for a parameter that evaluates defects. The parameters of each of the inspection logics of the first inspection logic L1, the second inspection logic L2, the third inspection logic L3, or the repair inspection logic Lk may be generated by, for example, an artificial intelligence (AI) that learns (or has already learned) based on the relationship between information on the product image P or the corrected product image Pc and the inspection results. This allows the inspection logic to be changed according to the information in the product image P or the corrected product image Pc, thereby enabling accurate inspection results (pass / fail judgments) in each inspection means. Furthermore, for example, the first inspection logic L1 may be a function that judges pass or fail based on a threshold value set for a parameter for evaluating defects, and the repair inspection logic Lk may be generated by artificial intelligence AI that learns based on the relationship between the information in the product image P or the corrected product image Pc and the inspection results. As a result, the first inspection T1 does not use a learning artificial intelligence AI, but the repair inspection Tk does use a learning artificial intelligence AI, so that electronic components 1 (product images P or corrected product images Pc) that are difficult to determine whether they are good or bad are deliberately judged (overkilled) as failing, which is a safe result, by the first inspection T1 (first inspection means 30) which does not use a learning artificial intelligence AI that acts as a filter in the first step (by appropriately adjusting the threshold value so that it is more likely to fail), and these failed electronic components 1 are accurately judged by the repair inspection Tk (rescue inspection means 40) which uses a learning artificial intelligence AI that acts as a filter in the second step, and an appropriate overall judgment can be made by the overall judgment means 50 using these two-step filters.

[0019] The image inspection device 100 may optionally include a preliminary inspection means for performing a preliminary inspection before the first inspection T1 by the first inspection means 30 .

[0020] The position detection means 10 detects the product position PP. The position detection means 10 detects the product position PP, for example, by two-dimensional coordinates set on the top surface of the conveying table based on the rotation axis of the conveying table, and by the distance and direction (angle) based on the rotation axis of the conveying table. The position detection means 10 may also detect the product position PP as two-dimensional coordinates based on one corner of a rectangular area that can be imaged by the imaging means C, for example. Information on the product position PP detected by the position detection means 10 is used to correct the product image P by the image correction means 20.

[0021] The image correcting means 20 generates a corrected product image Pc by correcting the product image P based on the product position PP. For example, if the front surface of the electronic component 1 is not perpendicular to the direction of the front camera, the image correcting means 20 corrects the product image P based on the product position PP to generate a corrected product image Pc that corresponds to an image of the front surface of the electronic component 1 when the front surface of the electronic component 1 is perpendicular to the direction of the front camera. The corrected product image Pc generated by the image correcting means 20 is used in the first inspection T1 performed by the first inspection means 30.

[0022] The first inspection means 30 executes a first inspection T1 having a first inspection logic L1 based on the corrected product image Pc and outputs a first inspection result T1R. The corrected product image Pc is provided from the image correction means 20. The first inspection logic L1 is logic for making a pass / fail judgment in the first inspection T1. The first inspection logic L1 is logic for making a fail judgment when a predefined defect is found in the corrected product image Pc and making a pass judgment when no defect is found. The first inspection logic L1 may be, for example, a function that judges pass or fail based on a threshold value set for a parameter for evaluating the defect. The function may be a model (algorithm) generated by artificial intelligence (AI) that has trained using samples of pass and fail product images as training data.

[0023] In this way, the first inspection T1 is an inspection to detect defects in the electronic component 1. Here, the term "defect" refers to a concept that includes what is called a defect in a general appearance inspection, such as dimensional defects (including dimensions outside the tolerance range or outside the set standard), foreign matter contamination, defects including chips where a corner or other part is missing, and damage where the component is completely broken in appearance. In this way, the first inspection T1 is an inspection to detect defects in the electronic component 1, and therefore can reliably determine that an electronic component 1 that would otherwise be defective is defective.

[0024] The first inspection T1 is preferably an inspection for detecting contamination or foreign matter on the element body among defects in the electronic component 1. More preferably, the first inspection T1 is an inspection for detecting contamination on the element body among defects in the electronic component 1. The element body refers to the remaining portion of the electronic component 1 when the electronic component 1 is divided into the electrode portion and the remaining portion. It has been found that the inspection for contamination or foreign matter on the element body is more likely to result in overkill, in which an electronic component 1 that would otherwise be a good product is judged to be a failure, compared to other inspections such as dimensional inspections in defect inspections. Therefore, by setting the first inspection T1 to an inspection for detecting contamination or foreign matter on the element body that is more likely to result in overkill, electronic components 1 in which contamination or foreign matter on the element body is detected can be initially rejected in the first inspection T1 to be on the safe side, and a precise pass / fail judgment regarding contamination or foreign matter on the element body can be made in the repair inspection Tk that follows in series after the first inspection T1.

[0025] When the first test result T1R satisfies a specific condition in the first test logic L1, the repair test means 40 executes a repair test Tk having a repair test logic Lk different from the first test logic L1, and outputs a repair test result TkR.

[0026] Specifically, the specific condition is when the first inspection logic L1 is a pass / fail judgment and the first inspection result T1R is a fail. As a result, even if the first inspection result T1R is a fail, the electronic component 1 that would otherwise be a non-defective product can be judged as a pass in the repair inspection Tk by the repair inspection means 40 and repaired.

[0027] The repair inspection logic Lk is logic for making a pass / fail judgment in the repair inspection Tk. Unlike the first inspection logic L1, the repair inspection logic Lk is set specifically for inspecting a corrected product image Pc whose first inspection result T1R meets a specific condition (determined as fail) in the first inspection logic L1. The repair inspection logic Lk is logic that fails a corrected product image Pc whose first inspection result T1R meets a specific condition (determined as fail) in the first inspection logic L1 if a predefined defect different from the defect defined in the first inspection logic L1 is found, and passes the corrected product image Pc if no defect is found. The defect defined in the first inspection logic L1 and the defect defined in the repair inspection logic Lk may be the same type, but the pass / fail judgment criteria of the first inspection logic L1 and the repair inspection logic Lk may be different. As such, the image inspection device 100 is equipped with the repair inspection means 40. Therefore, even if the electronic component 1 being inspected is determined to be non-defective in the first inspection T1, it can be determined to be pass in the repair inspection Tk.

[0028] The comprehensive judgment means 50 makes a comprehensive judgment on the electronic component 1 based on the first inspection result T1R and the repair inspection result TkR. Specifically, the comprehensive judgment means 50 makes a comprehensive judgment as a non-defective product when the first inspection result T1R is pass or when the repair inspection result TkR is pass, and makes a comprehensive judgment as a defective product when the first inspection result T1R is fail and the repair inspection result TkR is fail. In other words, the comprehensive judgment means 50 makes a comprehensive judgment as pass when the first inspection result T1R is pass, makes a comprehensive judgment as pass when the first inspection result T1R is fail and the repair inspection result TkR is pass, and makes a comprehensive judgment as fail when the first inspection result T1R is fail and the repair inspection result TkR is fail. As a result, even if the first inspection result T1R is fail, the comprehensive judgment is not immediately made as fail, and even if the first inspection result T1R is fail, the comprehensive judgment is made as pass when the repair inspection result TkR is pass. Therefore, the image inspection device 100 can prevent products that should be considered non-defective from being determined to be defective.

[0029] The image inspection device 100 may include a second inspection means 60 that executes a second inspection T2 having a second inspection logic L2 different from the first inspection logic L1 and outputs a second inspection result T2R. This allows the second inspection T2 to be executed in parallel on the electronic component 1 from a different perspective than the first inspection T1. Therefore, even if an electronic component 1 that would otherwise be defective passes the first inspection T1, it can be comprehensively determined to be defective if it fails the first inspection T2. ​​This prevents products that should otherwise be defective from being determined to be non-defective, thereby improving the inspection accuracy of the image inspection device 100.

[0030] The comprehensive judgment means 50 makes a comprehensive judgment based on the first inspection result T1R, the repair inspection result TkR, and the second inspection result T2R. As a result, even if an electronic component 1 that would otherwise be defective passes the first inspection T1, if it fails the second inspection T2, it can be comprehensively judged as defective. This makes it possible to prevent products that should otherwise be considered good from being judged as defective, and to prevent products that should otherwise be considered defective from being judged as good, thereby improving the inspection accuracy of the image inspection device 100.

[0031] Specifically, the second inspection T2 is preferably an inspection that measures the dimensions of the electronic component 1. It has been found that an inspection that measures dimensions is less likely to result in overkill, in which an electronic component 1 that is actually a good product is judged to be a failure, compared to other inspections such as inspections for dirt or foreign matter on the element body in defect inspections. Therefore, it is preferable to perform the second inspection T2 that measures the dimensions of the electronic component 1, which is an inspection that is less likely to result in overkill, in parallel with the first inspection T1 that is prior to the repair inspection Tk, independently of the first inspection T1.

[0032] The image inspection device 100 may be equipped with a second repair inspection means, similar to the repair inspection means 40, that executes a second repair inspection Tk2 having a second repair inspection logic Lk2 different from the second inspection logic L2 when the second inspection result T2R satisfies a specific condition in the second inspection logic L2, and outputs a second repair inspection result TkR2. That is, the second inspection T2 may be refrained from, similar to the repair inspection Tk for the first inspection T1. This further prevents products that should be considered good from being determined to be defective.

[0033] The image inspection device 100 may include a third inspection means 70 that performs a third inspection T3 having a third inspection logic L3 that is different from the first inspection logic L1 of the first inspection T1 and the second inspection logic L2 of the second inspection T2. ​​The third inspection means 70 is provided in parallel and independent of the first inspection means 30 and the second inspection means 60. This allows the electronic component 1 to be inspected from more perspectives, thereby improving the accuracy of inspecting defective products.

[0034] The technical scope of the present invention is not limited to the above-described embodiments, and various modifications can be made without departing from the spirit of the present invention.

[0035] In addition, the components in the above-described embodiments may be replaced with known components as appropriate without departing from the spirit of the present invention. Furthermore, the above-described modifications may be combined as appropriate without departing from the spirit of the present invention.

[0036] As described above, the image inspection device 100 for electronic components 1 according to the embodiment includes a position detection means 10 that detects a product position PP based on a product image P obtained by capturing an image of the electronic component 1, an image correction means 20 that generates a corrected product image Pc by correcting the product image P based on the product position PP, a first inspection means 30 that executes a first inspection T1 having a first inspection logic L1 based on the corrected product image Pc and outputs a first inspection result T1R, a repair inspection means 40 that executes a repair inspection Tk having a repair inspection logic Lk different from the first inspection logic L1 when the first inspection result T1R satisfies a specific condition in the first inspection logic L1 and outputs a repair inspection result TkR, and a comprehensive judgment means 50 that makes a comprehensive judgment on the electronic component 1 based on the first inspection result T1R and the repair inspection result TkR. As a result, even if an electronic component 1 that should be judged as good (OK) is judged as a failure by the first inspection means 30 (hereinafter, this phenomenon may be referred to as overkill), if it is judged as good by the relief inspection means 40, it will be judged as good by the overall judgment means 50. Therefore, the image inspection device 100 for electronic components 1 can prevent products that should be good from being judged as defective.

[0037] The image inspection method according to the embodiment includes a product position detection step of detecting a product position PP based on a product image P obtained by capturing an electronic component 1, an image correction step of generating a corrected product image Pc by correcting the product image P based on the product position PP, a first inspection step of executing a first inspection T1 having a first inspection logic L1 based on the corrected product image Pc and outputting a first inspection result T1R, a repair inspection step of executing a repair inspection Tk having a repair inspection logic Lk different from the first inspection logic L1 when the first inspection result T1R satisfies a specific condition in the first inspection logic L1 and outputting a repair inspection result TkR, and a comprehensive judgment step of making an overall judgment on the electronic component 1 based on the first inspection result T1R and the repair inspection result TkR. As a result, even if an electronic component 1 that should have been judged to be good (OK) is judged to be a failure by the first inspection means 30 (hereinafter, this event may also be referred to as overkill), if the electronic component is judged to be good in the repair inspection step, it will be judged to be good in the comprehensive judgment step. Therefore, according to the image inspection method for the electronic component 1, it is possible to prevent a product that is intended to be a non-defective product from being determined to be a defective product.

[0038] An image inspection program according to the embodiment causes a computer to execute the image inspection method described above. That is, the image inspection program according to the embodiment causes a computer to execute a product position detection function that detects a product position PP based on a product image P obtained by capturing an electronic component 1, an image correction function that generates a corrected product image Pc by correcting the product image P based on the product position PP, a first inspection function that executes a first inspection T1 having a first inspection logic L1 based on the corrected product image Pc and outputs a first inspection result T1R, a repair inspection function that executes a repair inspection Tk having a repair inspection logic Lk different from the first inspection logic L1 when the first inspection result T1R satisfies a specific condition in the first inspection logic L1 and outputs a repair inspection result TkR, and an overall judgment function that performs an overall judgment on the electronic component 1 based on the first inspection result T1R and the repair inspection result TkR. As a result, even if an electronic component 1 that should be judged as good (OK) is judged as a failure by the first inspection means 30 (hereinafter, this phenomenon may be referred to as overkill), if it is judged as good in the repair inspection step, it will be judged as good in the overall judgment step. Therefore, according to the image inspection method for electronic components 1, it is possible to prevent products that should be good from being judged as defective.

[0039] 100 Image inspection device, 1 Electronic component, 10 Position detection means, 20 Image correction means, 30 First inspection means, 40 Relief inspection means, 50 Overall judgment means, 60 Second inspection means, 70 Third inspection means, C Imaging means, S Control means, L1 First inspection logic, L2 Second inspection logic, L3 Third inspection logic, P Product image, T1 First inspection, T1R First inspection result, T2 Second inspection, T2R Second inspection result, T3 Third inspection, Tk Relief inspection, TkR Relief inspection result

Claims

1. A position detection means for detecting a product position based on a product image obtained by imaging an electronic component; an image correction means for generating a corrected product image by correcting the product image based on the product position; a first inspection means for executing a first inspection having a first inspection logic based on the corrected product image and outputting a first inspection result; a relief inspection means for executing a relief inspection having a relief inspection logic different from the first inspection logic when the first inspection result is non-conforming in the case where the first inspection logic is a pass / fail determination and outputting a relief inspection result; a comprehensive determination means for making a comprehensive determination on the electronic component based on the first inspection result and the relief inspection result, wherein the first inspection logic is a function for determining pass or fail based on a threshold value set for a parameter for evaluating a defect, and the relief inspection logic is generated by an artificial intelligence that learns based on the relationship between the information of the product image or the corrected product image and the inspection result. An image inspection apparatus for electronic components.

2. The image inspection apparatus for electronic components according to claim 1, wherein the first inspection is an inspection for detecting dirt or foreign matter on the body among the defects of the electronic component.

3. The image inspection apparatus for electronic components according to claim 1 or claim 2, wherein the comprehensive determination means makes a comprehensive determination as a non-defective product when the first inspection result is pass or the relief inspection result is pass, and makes a comprehensive determination as a defective product when the first inspection result is non-conforming and the relief inspection result is non-conforming.

4. The image inspection apparatus for electronic components according to claim 1 or claim 2, further comprising a second inspection means for executing a second inspection having a second inspection logic different from the first inspection logic and outputting a second inspection result, wherein the comprehensive determination means makes the comprehensive determination based on the first inspection result, the relief inspection result and the second inspection result.

5. The image inspection apparatus for electronic components according to claim 4, wherein the comprehensive determination means makes a comprehensive determination as a non-defective product when the first inspection result is pass or the relief inspection result is pass, and the second inspection result is pass, and makes a comprehensive determination as a defective product when the first inspection result is non-conforming and the relief inspection result is non-conforming, or the second inspection result is non-conforming.

6. The image inspection apparatus for electronic components according to claim 1 or claim 2, wherein the first inspection is an inspection for detecting a defect of the electronic component.

7. The image inspection apparatus for electronic components according to claim 5, wherein the second inspection is an inspection for measuring the dimensions of the electronic components.

8. A method for inspecting an image of an electronic component, comprising: a position detection step of detecting a product position based on a product image obtained by imaging the electronic component; an image correction step of generating a corrected product image by correcting the product image based on the product position; a first inspection step of performing a first inspection having a first inspection logic based on the corrected product image and outputting a first inspection result; a relief inspection step of, when the first inspection result is non-conforming in a case where the first inspection logic is a pass / fail determination, performing a relief inspection having a relief inspection logic different from the first inspection logic and outputting a relief inspection result; and a comprehensive determination step of performing a comprehensive determination on the electronic component based on the first inspection result and the relief inspection result, wherein the first inspection logic is a function for determining pass or fail based on a threshold value set for a parameter for evaluating a defect, and the relief inspection logic is generated by an artificial intelligence that learns based on a relationship between information of the product image or the corrected product image and the inspection result.

9. A program for inspecting an image of an electronic component, causing a computer to execute: a position detection function of detecting a product position based on a product image obtained by imaging the electronic component; an image correction function of generating a corrected product image by correcting the product image based on the product position; a first inspection function of performing a first inspection having a first inspection logic based on the corrected product image and outputting a first inspection result; a relief inspection function of, when the first inspection result is non-conforming in a case where the first inspection logic is a pass / fail determination, performing a relief inspection having a relief inspection logic different from the first inspection logic and outputting a relief inspection result; and a comprehensive determination function of performing a comprehensive determination on the electronic component based on the first inspection result and the relief inspection result, wherein the first inspection logic is a function for determining pass or fail based on a threshold value set for a parameter for evaluating a defect, and the relief inspection logic is generated by an artificial intelligence that learns based on a relationship between information of the product image or the corrected product image and the inspection result.

Images