KR20200070909A - System and method for guiding a user of vision inspection for detecting defects of adjunct - Google Patents

Abstract

The present relates to a setting guide system for vision inspection of a component defect, which performs a plurality of provisional inspections by using inspection tools of various methods in advance with respect to a defective type registered by an operator in a vision inspection of whether a specific component is defective in a product, and analyzes the results of the provisional inspections to recommend an inspection tool with a high detection rate to an operator. The setting guide system of the present invention comprises: a good image DB for managing a good image of a product; a bad image DB for managing a bad image of the product; an inspection tool unit for managing at least two inspection tools for determining a component defect in the product through a product image; a defective area registration unit for registering labeling information of a defective area designated by the operator from bad images in the bad image DB; a provisional inspection setting parameter estimating unit for analyzing the defective area labeling information registered in the defective area registration unit to estimate and register a provisional inspection setting parameter for the provisional inspection for each of the inspection tools of the inspection tool unit; an inspection unit for executing the inspection tools of the inspection tool unit with respect to the bad image in the bad image DB and the good image in the good image DB in accordance with each of the provisional inspection setting parameter; a result analysis unit for deriving inspection criteria and detection rates for each of the inspection tools by analyzing the results of the provisional inspection for each of the inspection tools, which is executed by the inspection unit; and a recommendation unit for recommending the inspection tools for a product to be inspected through an analysis result of the result analysis unit.

Description

System and method of setting guide for vision inspection of attached defects{SYSTEM AND METHOD FOR GUIDING A USER OF VISION INSPECTION FOR DETECTING DEFECTS OF ADJUNCT}

The present invention relates to a setting guide system and method for vision inspection of accessory defects, and more particularly, various types of inspection tools for defect types registered by an operator in a vision inspection for the presence or absence of a specific accessory defect in a product By performing a number of provisional inspections using them and analyzing the results of these provisional inspections, an operator who is not proficient in vision inspection can set high-accuracy inspection tools according to the guide by recommending an inspection tool with a high detection rate to the operator. The present invention relates to a setting guide system and method for vision inspection of ancillary defects that enable an engineer to set a reliable inspection tool through accurate data detection rate.

In general, electrical or electronic components that are produced in large quantities in a uniform shape, such as elements or batteries as unit parts constituting an electrical device or a circuit, maintain their normal voltage or resistance before shipping the product. After determining whether a product is defective through a product performance inspection process that checks whether it is present, a process of selecting only a normal product is essential.

In the case of automobiles, home appliances, etc., in which several parts are combined/combined, even if a slight malfunction occurs in one of several parts, it may cause a malfunction in normal operation. It is very important to do it.

Therefore, a process for selecting whether a good product or a bad product is performed through various detailed inspections prior to shipment of parts is performed.

Vision inspection, which is one of such precise inspections, includes the sealer application state, the frame engagement state, the marking state, whether the lead or ball grid is damaged or cracked. It is mainly used to inspect the appearance of parts such as (Crack) or scratch.

In recent years, the production process of parts mostly consists of an automated production process that moves along the process line. Therefore, a vision inspection device is installed in various places on the process line, and the parts (hereinafter referred to as inspection objects) moving along the process line Vision inspection is being carried out.

However, in the conventional vision inspection system, the vision inspection method adopts a method in which an expected defect value is previously assigned to equipment by an experienced vision engineer and judged as a defect when the value is exceeded. Therefore, in order to apply the vision inspection device to the actual production site, it is difficult for small and medium-sized manufacturers who are difficult to equip skilled vision inspection engineers due to the difficulty of setting up a large number of parameters by professional engineers skilled in vision inspection. have.

Therefore, the problem to be solved by the present invention is to execute a number of provisional inspections using various types of inspection tools in advance for the defect type registered by the operator in the vision inspection for the presence or absence of a defect in a specific accessory in the product, and this provisional inspection result By analyzing them and recommending inspection tools with high detection rates to workers, even those who are not proficient in vision inspection can set high-accuracy inspection tools according to the guide, and engineers who are proficient in vision inspection can set up reliable inspection tools through accurate data detection rates. It is to provide a setting guide system and method for vision inspection of ancillary defects to enable the.

According to the present invention, a good image DB that manages a good image of a product; A defect image DB that manages a defect image of a product; An inspection tool unit that manages at least two or more inspection tools capable of determining an accessory defect in the product through a product image; A defect area registration unit for registering defect area labeling information designated by an operator from the defective images in the defective image DB; A provisional inspection setting parameter estimator for analyzing the defect area labeling information registered in the defect area registration unit and estimating and registering provisional inspection setting parameters for provisional inspection for each inspection tool of the inspection tool unit; An inspection unit that executes inspection tools of the inspection tool unit according to respective provisional test setting parameters for the defective image in the defective image DB and the good image in the good image DB; A result analysis unit for deriving inspection criteria and detection rates for each inspection tool by analyzing provisional inspection results for each inspection tool executed by the inspection unit; And a recommendation unit recommending an inspection tool for a product to be inspected through an analysis result of the result analysis unit. It provides a setting guide system for vision inspection of the accessory defect comprising a.

Preferably, an inspection area setting unit that sets an inspection area in the image that requires inspection for provisional inspection performed by the inspection unit; Characterized in that it further comprises.

Preferably, the result analysis unit is used in the provisional inspection, when the inspection and distribution of the inspection values of the good and bad products through the inspection tool do not overlap and the good and bad products should be determined based on one value in the inspection tool. It is characterized by learning the distribution characteristics of a good image and a bad image, and deriving an intermediate value of a margin section between a distribution location of a good image and a distribution location of a bad image as an inspection reference value.

Preferably, the result analysis unit, when the provisional inspection through the inspection tool, the inspection value distribution of good and bad products do not overlap, and when it is necessary to determine good and bad products based on the range between the minimum and maximum values in the inspection tool, the temporary inspection It is characterized by learning the distribution characteristics of good and bad images used in the company and deriving the intermediate values of the margin range between the good image distribution range and the bad image distribution range as an inspection reference value.

Preferably, the result analysis unit, when the inspection value distribution of the good and bad products as a result of provisional inspection through the inspection tool overlaps, and the corresponding inspection tool needs to determine good and bad products based on one value, the good quality used in the temporary inspection It is characterized by learning the distribution characteristics of the image and the defective image, and deriving a value that excludes the inspection value of the defective product from the boundary dividing the distribution location of the defect image and the distribution location of the defective image as the inspection reference value.

Preferably, the result analysis unit, the provisional inspection through the inspection tool, the inspection value distribution of good and bad products overlaps, and if it is necessary to determine good and bad products based on the range between the minimum and maximum values in the inspection tool, the temporary inspection It is characterized by learning the distribution characteristics of the good and bad images used for deriving two boundary values before and after the good image distribution range as the inspection reference value, but deriving the inspection reference value to the extent that the inspection value of the defective product is excluded.

Meanwhile, according to another aspect of the present invention, (a) receiving a good image and a bad image of a product and storing it in a good image DB and a bad image DB; (b) registering defect area labeling information designated by an operator from defective images in the defective image DB; (c) analyzing the defect area labeling information and estimating and registering provisional inspection setting parameters for provisional inspection for each inspection tool; (d) performing at least two inspection tools on the defective image in the defective image DB and the good image in the good image DB according to respective provisional test setting parameters to perform provisional inspection; (e) analyzing the results of provisional tests for each inspection tool to derive inspection criteria and detection rates for each inspection tool; And (f) recommending an inspection tool for a product to be inspected through an analysis result. It provides a setting guide method for vision inspection of the accessory defect, characterized in that it comprises a.

Preferably, between the steps (a) and (b), setting an inspection area in the image that requires inspection for provisional inspection performed by the inspection unit; Characterized in that it further comprises.

Preferably, between the steps (c) and (d), the step of receiving an inspection tool to perform the temporary inspection among the inspection tools for the temporary inspection performed by the inspection unit; Characterized in that it further comprises.

Preferably, in the step (e), when the inspection value distribution of the good and bad products does not overlap as a result of the temporary inspection through the inspection tool, the good and bad products should be determined based on one value in the inspection tool, to the temporary inspection. It is characterized by learning the distribution characteristics of the used good image and the bad image and deriving the intermediate value of the margin section between the distributed position of the good image and the distributed position of the bad image as an inspection reference value.

Preferably, in the step (e), when the provisional inspection through the inspection tool does not overlap the distribution of inspection values of good and bad products and the inspection tool needs to determine good and bad products based on the range between the minimum and maximum values It is characterized by learning the distribution characteristics of good and bad images used for provisional inspection to derive the intermediate values of the margin range between the good image distribution range and the bad image distribution range as the inspection reference value.

Preferably, in the step (e), the inspection value distribution of the good and bad products as a result of the temporary inspection through the inspection tool overlaps, and when it is necessary to determine the good and bad products based on one value in the inspection tool, it is used for the temporary inspection. It is characterized by learning the distribution characteristics of the defective good image and the bad image and deriving a value that excludes the inspection value of the bad product from the boundary dividing the distribution position of the good image and the distribution location of the bad image as the inspection reference value.

Preferably, in the step (e), the inspection result distribution of good and bad products as a result of provisional inspection through the inspection tool overlaps, and when the inspection tool needs to determine good and bad products based on the range between the minimum and maximum values, It is characterized by learning the distribution characteristics of good and bad images used for provisional inspection to derive two boundary values before and after the range of good image distribution as inspection reference values, but deriving inspection reference values to the extent that inspection values of defective products are excluded. .

The present invention executes a number of provisional inspections using various types of inspection tools in advance for a defect type registered by an operator in a vision inspection for the presence or absence of a defect in a specific accessory in a product, and analyzes the provisional inspection results to analyze the high detection rate By recommending a tool to an operator, even an inexperienced operator can set a high-accuracy inspection tool according to the guide.

In addition, by providing an accurate detection rate that is data-specific for each inspection tool to a professional vision engineer who is proficient in vision inspection, it is possible to set a more reliable inspection tool than when judged by experience and subject.

1 is a block diagram illustrating a setting guide system for vision inspection of an accessory defect according to an embodiment of the present invention.
2 is a flowchart illustrating a setting guide method for vision inspection of an accessory defect according to an embodiment of the present invention.
3 to 10 are reference diagrams for explaining a setting guide process for vision inspection of accessory defects according to an embodiment of the present invention.

In order to fully understand the operational advantages of the present invention and the objects achieved by the embodiments, reference should be made to the accompanying drawings that illustrate preferred embodiments of the present invention.

Hereinafter, preferred embodiments of the present invention will be described with reference to the accompanying drawings. The same reference numerals in each drawing denote the same members.

First, FIG. 1 is a block diagram illustrating a setting guide system for vision inspection of accessory defects according to an embodiment of the present invention.

The setting guide system to which the present invention is applied may be included in a vision inspection system that performs quality selection. This product sorting system receives product images from cameras installed in various places on the production line where products are produced, and selects whether products are good or bad based on the captured images and moves them along various process lines. A vision inspection will be performed, and the setting guide system uses a variety of inspection tools for defects of specific parts in the product to execute a number of preliminary inspections in advance and analyzes the preliminary inspection results to use an inspection tool with a high detection rate. It will be recommended to workers.

The selection of good products described below will be performed by vision inspection of accessory defects to determine whether a defect is present in a specific accessory in the product.

Referring to FIG. 1, the setting guide system according to an embodiment of the present invention includes a receiving unit 10 for receiving data input from a process operator, a camera unit 20 for capturing and delivering an image of a product in the process, and a process worker Good image DB (31) for managing good image input from the receiving unit 10 by, bad image DB (32) for managing bad image input from the receiving unit 10 by a process worker, the good image and Inspection area setting unit 50 for setting an inspection area for selecting good products from defective images, inspection tool unit 60 for managing various inspection tools capable of discriminating defects in a product through product images, and the defective image Defect area registration unit 51 that registers defect area labeling information designated by an operator from defective images in DB 32, analyzes defect area labeling information registered in the defect area registration unit 51, and analyzes the inspection tool unit ( The inspection according to the provisional inspection setting parameters estimated by the provisional inspection setting parameter estimator 52, and the provisional inspection setting parameter estimator 52 for estimating and registering the provisional inspection setting parameters for the provisional inspection for each inspection tool of 60) Inspection unit 70 for inspecting the defects in the product from the product image by executing the inspection tools of the tool unit 60, a result analysis unit 80 for analyzing the inspection results of the inspection unit 70, the result analysis unit 80 ) Through the analysis results of the recommendation unit (90) that recommends the optimal inspection tool for the product to be inspected, and controls the operation of the above-mentioned components, and inspects various methods in advance in the vision inspection for the defects of specific parts in the product. It can be made by including a control unit 40 that executes a number of provisional inspections using them and analyzes the provisional inspection results and recommends an inspection tool having a high detection rate to an operator.

The receiving unit 10 receives data input from a process worker. Such data may be a good image or a bad image of a product produced in a corresponding process, and may also be control data input from a process operator.

The camera unit 20 may be a camera focused on a product produced in a corresponding process, and photographs an image (image or video) of the product in the process.

The good image DB 31 stores and manages a good image selected as a good product by a process worker among images input from the receiving unit 10 by a process worker.

The defective image DB 32 stores and manages a defective product image selected as a defective product by a process worker among images input from the receiving unit 10 by a process worker.

Here, a good product means a product produced in a state where all parts required for the product are configured without defects in a predetermined position in a product produced in the corresponding process, and for example, this good product is photographed as shown in (a) of FIG. 3. One image can be stored and managed in the good image DB 31 as a good image.

In contrast, a defective product means a product produced in a state in which a defect is generated in at least one of all parts required for the product in a product produced in the process, for example, as illustrated in (b) and (c) of FIG. 3. As described above, images of such defective products having defects in the parts can be stored and managed in the defective image DB 32 as defective product images.

At this time, the accessory means all accessories that are dependent on the product.

The inspection area setting unit 50 is configured to receive and designate an inspection area that requires inspection from a captured image of the camera unit 20 disposed in the corresponding process. As an example, the inspection area is set as a part of an image of a product produced in a corresponding process (a box indicated by a blue solid line in FIG. 4) as illustrated in FIG. 4, thereby increasing inspection speed and inspection efficiency. The entire area may be set as the inspection area.

The inspection tool unit 60 can determine a defect (error) of a specific accessory by comparing an inspection area of a product image to be inspected with an inspection area of a good product image for an inspection area set by the inspection area setting unit 50 It manages various inspection tools. This inspection tool is a tool that can inspect defects of specific parts in the inspection area with relatively complicated calculations. For example, blob comparison tool, local blob comparison tool, pixel number comparison tool, good quality model learning (defective amount) comparison tool, etc. Various tools can be used for this.

Here, the inspection tools of the inspection tool unit 60 according to the present invention are provisional images DB for inspecting the inspection area of the defective image in the defect image DB 32 for the inspection area set by the inspection area setting unit 50 as a provisional inspection. (31) By comparing with the inspection area of the good quality image in the product, it will be determined whether a specific accessory is defective.

The defect area registration unit 51 receives information about a defect area displayed by an operator directly on an image for various defective images and registers it as defect area labeling information.

Referring to FIG. 5, defects exist in a part of parts in the defective images of (a) and (b), and the defective image is directly processed by an operator to display the defective area (representing the area as a rectangle). . In this way, information on the defect area directly displayed on the defect image by the operator is registered and managed as various types and patterns as shown in FIG. 6 as defect area labeling information.

Here, in the defect area labeling information, information such as size, brightness, number of pixels, and occurrence location may be stored together with the image of the defect area.

The actual defect shape and pattern is best experienced by the field operator, so the process of registering the defect area of the operator will ensure that the final inspection results are tailored to the individual site conditions and environment.

The provisional inspection setting parameter estimator 52 analyzes a plurality of defect area labeling information registered in the defect area register 51 by a field worker, and performs provisional inspection for provisional inspection for each inspection tool of the inspection tool unit 60. The setup parameters are estimated and registered.

That is, the provisional inspection setting parameter for executing the provisional inspection for each inspection tool is registered.

Referring to FIG. 7, for example, the provisional inspection setting parameter estimator 52 analyzes the defect area labeling information to calculate the average and variance for the brightness of the defect area to estimate the brightness ranges 54 to 136. The provisional inspection setting parameter is used to perform the provisional inspection of the relevant inspection tool.

The inspection unit 70 individually executes inspection tools of the inspection tool unit 60 for inspection areas of defective images in the defective image DB 32 and inspection areas of good images in the good image DB 31. do. At this time, the inspection unit 70 may increase the speed and efficiency of the provisional inspection by inspecting only the inspection tools selected from the operator to proceed with the provisional inspection process.

The result analysis unit 80 analyzes provisional inspection results of various inspection tools executed by the inspection unit 70 to derive inspection standards and detection rates for each inspection tool.

8 and 9 are views for explaining the inspection criteria according to the inspection type applied to the present invention. In FIGS. 8 and 9, good products are indicated by a green circle and defective products are indicated by a red circle.

First, FIG. 8 illustrates a case in which a distribution pattern represented by inspection values for each sample of an inspection tool when a provisional inspection is performed with one inspection tool does not overlap with each other for a good product and a defective sample, and a constant margin is shown. .

First, FIG. 8(a) shows a form in which good and bad products are determined based on a specific value of a parameter type of a test value among various test tools. For example, in the case of the brightness comparison tool, when the brightness value inspected based on the brightness value 50 is 50 or less, it may be a defective product in which a specific accessory does not exist, and when the inspected brightness value is 50 or more, a quality product in which a specific accessory exists It could be.

In this case, the result analysis unit 80 learns the distribution characteristics of the good image and the bad image used for the provisional inspection and derives an intermediate value of the margin section between the distribution location of the good image and the distribution location of the bad image as an inspection reference value. In this case, the detection rate of the temporary inspection using this inspection tool will be derived as 100%. That is, the inspection criterion is an inspection criterion value expressed as a single value.

Next, (b) of FIG. 8 is a form in which a good or a bad product is determined based on a range of a specific minimum value and a specific maximum value of a parameter type of an inspection value among various inspection tools. For example, in the case of the standard deviation comparison tool, when the deviation value inspected based on the range of the deviation values 40 and 60 is 20 or 80, it may be a defective product in which a specific accessory does not exist, and the inspected deviation value is 50 In some cases, it will be a form that can be a good product with a specific accessory.

In this case, the result analysis unit 80 learns the distribution characteristics of the good image and the bad image used in the temporary inspection, and the front and rear two boundary margin sections between the good image distribution range and the bad image distribution range located before and after it. We will derive the median values of, as the test reference values, and derive the detection rate of false tests using this test tool as 100%. That is, the inspection criterion is a range of inspection criteria expressed as minimum and maximum inspection reference values.

Next, FIG. 9 illustrates a case in which the distribution pattern represented by the inspection values for each sample of the inspection tool when the temporary inspection is performed by one inspection tool does not show a margin because the distribution overlaps with respect to the good and bad samples.

First, in FIG. 9(a), good and bad products are determined based on a specific value of a parameter type of a test value among various test tools, but the test values of good and bad products partially overlap. For example, in the case of the brightness comparison tool, when the brightness value inspected based on the brightness value 50 is 50 or less, it may be a defective product in which a specific accessory does not exist, and when the inspected brightness value is 50 or more, a quality product in which a specific accessory exists It could be. However, as a result of performing the preliminary inspection using this inspection tool, there may be cases in which the distribution of inspection values is partially overlapped with good and bad products, as shown in the drawing, and the boundary criteria cannot be accurately divided.

In this case, the result analysis unit 80 learns the distribution characteristics of the good image and the bad image used for the provisional inspection, and determines the value at which the inspection value of the bad product is excluded from the boundary dividing the distribution location of the good image and the distribution location of the bad image. It will be derived as the test reference value, and the detection rate of the temporary test using this test tool will be derived as 100%. That is, the inspection criterion is an inspection criterion value expressed as a single value. In particular, in order to perfectly classify all defective products in the quality inspection, it is highly reliable inspection process to derive a value that excludes the inspection value of the defective product from the boundary dividing the distribution location of the quality image and the distribution location of the defect image as described above, as a test standard. It is desirable for construction.

Next, (b) of FIG. 9 is a form in which good and bad products are determined based on a range of a specific minimum value and a specific maximum value among parameter types of inspection values among various inspection tools. For example, in the case of the standard deviation comparison tool, when the deviation value inspected based on the range of the deviation values 40 and 60 is 20 or 80, it may be a defective product in which a specific accessory does not exist, and the inspected deviation value is 50 In some cases, it will be a form that can be a good product with a specific accessory. However, as a result of performing the preliminary inspection using this inspection tool, there may be cases where the distribution of inspection values is partially overlapped with good and bad products as shown in the drawing, and the boundary range cannot be accurately divided.

In this case, the result analysis unit 80 learns the distribution characteristics of the good image and the bad image used for the temporary inspection to derive two boundary values before and after the distribution range of the good image as the inspection reference value, but the inspection value of the defective product is excluded. The reference value of the test will be derived for the range, and the detection rate of the temporary test using this test tool will be derived as 100%. That is, the inspection criterion is a range of inspection criteria expressed as minimum and maximum inspection reference values. In particular, in order to completely classify all defective products in the quality inspection, it is desirable to construct a highly reliable inspection process by deriving a range in which the inspection value of the defective product is excluded from the distribution range of the quality product image as described above.

The recommendation unit 90 presents the test result for each test tool to the operator through the analysis result of the result analysis unit 80, and recommends the test tool having the best provisional test result as the recommended test tool.

10 shows an example of a provisional test result for each test tool and a test tool recommendation screen suggested by the recommendation unit 90.

The recommendation unit 90 presents the detection rate and inspection criteria of the temporary inspection for each inspection tool to the worker, and also provides the time and the individual inspection speed for each inspection tool for each inspection tool, thereby providing the operator with the inspection result at a glance. Make sure to understand the pros and cons of each inspection tool. In addition, the recommendation unit 90 recommends an inspection tool having the best provisional test result according to the provisional test result. Priority criteria for the recommendation of such inspection tools are to be first recommended as the inspection tool with the highest detected defect detection rate, and the next recommendation criterion is to select the inspection tools with a wide range of inspection criteria from the derived inspection criteria as the recommendation target. The next recommendation criterion is to target a test tool with a fast test speed. In other words, the most important thing in the vision inspection will be the defect detection rate, then it is important to determine the most products as good products, and the next criterion will be the process standard.

Therefore, even an inexperienced person in vision inspection can recommend the high-precision inspection tool according to the guide and easily set up the vision inspection process. Also, by providing a professional vision engineer who is skilled in vision inspection, it provides accurate detection rate data for each inspection tool. It will be possible to set a more reliable inspection tool than when judged by experience and subjectivity.

Now, a setting guide method for vision inspection of an accessory defect according to an embodiment of the present invention will be described with reference to FIG. 2.

2 is a flowchart illustrating a setting guide method for vision inspection of an accessory defect according to an embodiment of the present invention.

Referring to FIG. 2, the system first receives a good image and a bad image for a product produced in a corresponding process through the receiver 10 and stores them in the good image DB 31 and the bad image DB 32 ( S10). The good image DB 31 stores and manages a good image selected as a good product by a process worker among images input from the receiving unit 10 by a process worker. The defective image DB 32 stores and manages a defective product image selected as a defective product by a process worker among images input from the receiving unit 10 by a process worker.

Thereafter, the system receives an inspection area that needs to be inspected among product images through the inspection area setting unit 50 and sets it as an inspection area (S12).

Thereafter, in order to proceed with the provisional inspection process, the system first receives information about a defect area displayed by an operator directly on an image of various defective images of the defective image DB 32 and registers it as defect area labeling information ( S14).

In addition, the system analyzes a plurality of defect area labeling information registered in the defect area registration unit 51 by a field operator through the provisional inspection setting parameter estimation unit 52 to perform provisional inspection for each inspection tool of the inspection tool unit 60. The preliminary inspection setting parameter for estimation is estimated and registered (S16).

Thereafter, at least two or more inspection tool selection signals are received from the operator through the reception unit 10 among the inspection tools of the inspection tool unit 60, and an inspection tool to be subjected to provisional inspection is selected (S18).

This inspection tool selection step (S18) is to increase the speed and efficiency of the provisional inspection, and may be omitted since all the inspection tools in the inspection tool unit 60 may be subjected to the provisional inspection.

Thereafter, the system individually executes inspection tools of the inspection tool unit 60 for inspection areas of defective images in the defect image DB 32 and inspection areas of quality images in the good image DB 31 (S20). ).

Thereafter, the system analyzes the provisional inspection results of various inspection tools executed by the inspection unit 70 through the result analysis unit 80 (S22), and derives inspection standards and detection rates for each inspection tool (S24).

Thereafter, the system presents the inspection result for each inspection tool to the operator through the recommendation unit 90, and recommends the inspection tool having the best temporary inspection result to the worker as a recommended inspection tool (S26).

The present invention has been described with reference to some embodiments shown in the drawings, but these are merely exemplary, and those skilled in the art will understand that various modifications and other equivalent embodiments are possible therefrom. Therefore, the true technical protection scope of the present invention should be defined by the technical spirit of the appended claims.

10: receiving unit 20: camera unit
31: good image DB 32: bad image DB
40: control unit 50: inspection area setting unit
51: defect area registration unit 52: inspection setting parameter estimation unit
60: inspection tool unit 70: inspection unit
80: result analysis unit 90: recommendation unit

Claims (13)

A quality image DB that manages quality images of products;
A defect image DB that manages a defect image of a product;
An inspection tool unit that manages at least two or more inspection tools capable of determining an accessory defect in the product through a product image;
A defect area registration unit for registering defect area labeling information designated by an operator from the defective images in the defective image DB;
A provisional inspection setting parameter estimator configured to analyze the defect area labeling information registered in the defect area register and estimate and register provisional inspection setting parameters for provisional inspection for each inspection tool of the inspection tool part;
An inspection unit that executes inspection tools in the inspection tool unit according to respective provisional test setting parameters for the defective image in the defective image DB and the good image in the good image DB;
A result analysis unit for deriving inspection criteria and detection rates for each inspection tool by analyzing provisional inspection results for each inspection tool executed by the inspection unit; And
A recommendation unit recommending an inspection tool for a product to be inspected through an analysis result of the result analysis unit; Setting guide system for vision inspection of the accessory defect, characterized in that it comprises a.
According to claim 1,
An inspection area setting unit that sets an inspection area in the image that requires inspection for the provisional inspection performed by the inspection unit; Setting guide system for the vision inspection of the accessory defect, characterized in that it further comprises.
According to claim 1,
The result analysis unit,
As a result of provisional inspection through inspection tools, if the distribution of inspection values of good and bad products does not overlap, and the good and bad products must be determined based on one value in the inspection tool, the distribution characteristics of good and bad images used for temporary inspection A setting guide system for vision inspection of subsidiary defects characterized by deriving an intermediate value of a margin section between a distribution position of a learned good product image and a distribution position of a defective image as an inspection reference value.
According to claim 1,
The result analysis unit,
As a result of provisional inspection through inspection tools, if the distribution of inspection values of good and bad products does not overlap, and the relevant inspection tool needs to determine good and bad products based on the range between the minimum and maximum values, the good and bad images used for temporary inspection A setting guide system for vision inspection of accessory defects, characterized by deriving intermediate values of the margin range between the distribution range of good and bad images and the distribution range of bad images by learning the distribution characteristics of.
According to claim 1,
The result analysis unit,
As a result of provisional inspection through inspection tools, the distribution of inspection values between good and bad products overlaps, and if the inspection tool needs to determine good and bad products based on one value, learn the distribution characteristics of good and bad images used for temporary inspection A setting guide system for vision inspection of subsidiary defects, characterized in that a value that excludes the inspection value of a defective product is derived as an inspection reference value at a boundary dividing the distribution position of a defective product image and a distribution location of a defective product image.
According to claim 1,
The result analysis unit,
As a result of provisional inspection through the inspection tool, the distribution of inspection values between good and bad products overlaps, and if the inspection tool needs to determine good and bad products based on the range between the minimum and maximum values, the good and bad images used in the temporary inspection A setting guide system for vision inspection of subsidiary defects that is characterized by deriving two boundary values before and after the distribution range of a good product image as an inspection reference value by learning distribution characteristics, but deriving an inspection reference value within a range in which inspection values of defective products are excluded.
(a) receiving a good image and a bad image of the product and storing them in a good image DB and a bad image DB;
(b) registering defect area labeling information designated by an operator from defective images in the defective image DB;
(c) analyzing the defect area labeling information and estimating and registering provisional inspection setting parameters for provisional inspection for each inspection tool;
(d) performing at least two inspection tools on the defective image in the defective image DB and the good image in the good image DB according to respective provisional test setting parameters to perform provisional inspection;
(e) analyzing the results of provisional tests for each inspection tool to derive inspection criteria and detection rates for each inspection tool; And
(f) recommending an inspection tool for a product to be inspected through an analysis result; Setting guide method for vision inspection of the accessory defect, characterized in that it comprises a.
The method of claim 7,
Between steps (a) and (b),
Setting an inspection area in the image that requires inspection for provisional inspection performed by the inspection unit; The setting guide method for vision inspection of the accessory defect, characterized in that it further comprises.
The method of claim 7,
Between steps (c) and (d),
Selecting an inspection tool to perform a provisional inspection among the inspection tools for the provisional inspection performed by the inspection unit; The setting guide method for vision inspection of the accessory defect, characterized in that it further comprises.
The method of claim 7,
In step (e),
As a result of provisional inspection through inspection tools, if the distribution of inspection values of good and bad products does not overlap, and the good and bad products must be determined based on one value in the inspection tool, the distribution characteristics of good and bad images used for temporary inspection A setting guide method for vision inspection of accessory defects, characterized by deriving an intermediate value of a margin section between a distribution position of a learned good product image and a distribution position of a defective image as an inspection reference value.
The method of claim 7,
In step (e),
As a result of provisional inspection through inspection tools, if the distribution of inspection values of good and bad products does not overlap, and the relevant inspection tool needs to determine good and bad products based on the range between the minimum and maximum values, the good and bad images used for temporary inspection A setting guide method for vision inspection of accessory defects, characterized by deriving the intermediate values of the margin range between the distribution range of good and bad images and the distribution range of bad images by learning the distribution characteristics of.
The method of claim 7,
In step (e),
As a result of provisional inspection through inspection tools, the distribution of inspection values between good and bad products overlaps, and if the inspection tool needs to determine good and bad products based on one value, learn the distribution characteristics of good and bad images used for temporary inspection A setting guide method for vision inspection of accessory defects, characterized in that a value that excludes the inspection value of a defective product is derived as an inspection reference value at a boundary dividing the distribution position of a defective product image and a distribution location of a defective product image.
The method of claim 7,
In step (e),
As a result of provisional inspection through the inspection tool, the distribution of inspection values between good and bad products overlaps, and if the inspection tool needs to determine good and bad products based on the range between the minimum and maximum values, the good and bad images used in the temporary inspection A setting guide method for vision inspection of accessory defects, characterized by deriving two boundary values before and after the distribution range of a good product image as an inspection reference value by learning distribution characteristics, but deriving an inspection reference value within a range in which inspection values of defective products are excluded.

Images