Category condition setting support device

Abstract

To provide information useful for setting classification conditions for classifying defect types of an object to be inspected.SOLUTION: A classification condition setting support apparatus comprises: a basic information storage section, which stores basic information including basic imaging information that is imaging information obtained by imaging an object to be inspected and is associated with any of defect types and a basic defect type that is a defect type associated with each piece of the basic imaging information; a classification condition setting section that sets a classification condition for classifying the imaging information into any of the defect types; a basic defect type classification section that classifies the basic imaging information according to the classification condition; a classification result confirmation screen generation section that generates a classification result confirmation screen including the number of pieces of classification basic imaging information that is target basic imaging information classified into a predetermined defect type, the basic defect type associated with the classification basic imaging information, and the number of pieces of correct answer basic imaging information that is the number of pieces of the basic imaging information associated with the predetermine defect type among pieces of the target basic imaging information, by classifying the target basic imaging information according to the classification condition; and a display section.SELECTED DRAWING: Figure 5

Description

Technical Field

[0001] The present invention relates to a classification condition setting support device that supports setting classification conditions for classifying defect types of an inspection object.

Background Art

[0002] Conventionally, in appearance inspection, defects generated in an inspection object have been classified based on imaging information obtained by imaging the inspection object. Classification of such defect types of the inspection object determines feature amounts obtained based on the imaging information and classification conditions including thresholds set for the feature amounts, and determines whether the feature amounts obtained based on the imaging information are greater than or equal to the threshold or less than the threshold according to the classification conditions. The feature amounts and thresholds constituting the classification conditions are determined so that defects appearing in the imaging information of the inspection object can be appropriately classified. Various methods have been proposed for determining such classification conditions (see, for example, Patent Documents 1 to 3).

[0003] In order to optimize the classification conditions, imaging information obtained by imaging an inspection object having a defect in advance is confirmed, and correct answer information is created by associating an appropriate defect type with each imaging information. Then, by comparing the defect types obtained as a result of classifying the imaging information constituting the correct answer information according to the classification conditions with the defect types associated with the imaging information as the correct answer information, the suitability of the classification conditions is determined, and the selection of feature amounts and the setting of thresholds are adjusted to optimize the classification conditions.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Patent Document 2

Patent Document 3

Summary of the Invention

[0005] However, as the number of defects increases, such as during the pre-production inspection stage of the inspected items, the time required to associate each defect with a specific type increases, and errors in the association become more likely. Furthermore, as the number of defect types to be classified increases, the number of features that make up the classification conditions also increases, making the analysis work required to consider effective classification conditions more difficult. Consequently, it also becomes difficult to verify the appropriateness of the classification conditions that have been considered.

[0006] This invention has been made in view of the above-mentioned problems, and aims to provide useful information for setting classification conditions for classifying the types of defects in an inspected object. [Means for solving the problem]

[0007] The present invention, which solves the above problems, A basic information storage unit stores basic information, which includes basic imaging information obtained by imaging an object under inspection, and which includes basic imaging information associated with any of the defect types, and basic defect types that are the defect types associated with each of the basic imaging information. A classification condition setting unit sets classification conditions for classifying imaging information obtained by imaging the object under inspection into one of the aforementioned defect types, A basic defect type classification unit that classifies the basic imaging information according to the classification conditions, By classifying the target basic imaging information, which is at least a part of the aforementioned basic imaging information, according to the classification conditions, the target basic imaging information classified into a predetermined defect type is the classification basic A classification result confirmation screen generation unit generates a classification result confirmation screen that includes the number of imaging information, the basic defect type associated with the classification basic imaging information, and the number of correct basic imaging information, which is the number of basic imaging information from the target basic imaging information associated with the predetermined defect type. A display unit that displays the classification result confirmation screen, It is a classification condition setting support device equipped with the following features.

[0008] According to this, the basic imaging information stored in the basic information storage unit and pre-associated with defect types is classified in the basic defect type classification unit according to the classification conditions set in the classification condition setting unit, and the result of this classification is displayed on the display unit as a classification result confirmation screen. The basic defect type classification unit does not need to classify all of the basic imaging information stored in the basic information storage unit; at least a portion of it needs to be selected as target basic imaging information for classification. The classification result confirmation screen includes the number of classified basic imaging information items that have been classified into a predetermined defect type according to the set classification conditions, the basic defect types that have been pre-associated with these classified basic imaging information items, and the number of correct basic imaging information items, which is the number of basic imaging information items that have been pre-associated with the predetermined defect type among the target basic imaging information items. By displaying such a classification result confirmation screen, the user can clearly recognize the number of target basic imaging information items that have been correctly classified into the predetermined defect type and the breakdown of target basic imaging information items that have been incorrectly classified into the predetermined defect type among the target basic imaging information items that have been classified into the predetermined defect type. Furthermore, the number of correct basic imaging data points is also displayed, allowing the user to recognize whether the target basic imaging data points pre-associated with the specified defect type have been properly classified without being overlooked. In this way, the user can obtain information to consider the appropriateness of the set classification conditions from this classification result confirmation screen, and thus this classification condition setting support device can provide the user with useful information for setting classification conditions.

[0009] Furthermore, in the present invention, The classification result confirmation screen may include classification basic imaging information indicators displayed in a manner corresponding to the number of classification basic imaging information items and in a manner corresponding to the basic defect type associated with the classification basic imaging information.

[0010] In this way, the classification base imaging information index is displayed in a manner corresponding to the number of classification base imaging information items; for example, the length, size, and position of the classification base imaging information index differ depending on the number of classification base imaging information items, making it easy to visually recognize whether the classification conditions are appropriate or not. Furthermore, the classification base imaging information index is displayed in a manner corresponding to the basic defect type pre-associated with the classification base imaging information; for example, the color, pattern, and shape of the classification base imaging information index differ for each basic defect type, making it easy to visually recognize whether the classification conditions are appropriate or not.

[0011] Furthermore, in the present invention, The classification result confirmation screen may include a mark indicating that the number of classification basic imaging information items displayed by the classification basic imaging information index matches the number of correct basic imaging information items.

[0012] In this way, the classification result confirmation screen includes a mark indicating that the number of classification basic imaging information displayed by the classification basic imaging information index, which is displayed in a manner corresponding to the number of classification basic imaging information, matches the number of correct basic imaging information. By comparing the display manner of the classification basic imaging information index with the mark, it is possible to visually and easily recognize whether all the basic imaging information included in the target basic imaging information and pre-associated with a predetermined defect type has been correctly classified.

[0013] Furthermore, in the present invention, The classification result confirmation screen displays the number of classification base imaging information corresponding to a number exceeding the number of correct base imaging information. The display area may also include a display area for displaying image information indicators.

[0014] By doing so, since the classification result confirmation screen includes a display area for displaying a classification basis imaging information index corresponding to the number exceeding the number of correct basis imaging information, even when all the basis imaging information included in the target basis imaging information and associated with a predetermined defect type in advance is correctly classified into the predetermined defect type, the presence or absence of the target basis imaging information misclassified into the predetermined defect type due to oversight can be easily visually recognized by the classification basis imaging information index.

[0015] Also, in the present invention, It may be provided with a basic information generation unit that causes the imaging information obtained by imaging a test object having a defect to be displayed on the display unit and receives an input of the defect type associated with the imaging information.

[0016] By doing so, while the user displays and checks the imaging information on the display unit, basic information can be easily obtained by associating it with the defect type.

[0017] Also, in the present invention, The classification condition setting unit may have at least any one of an importance designation unit that receives a designation of weighting for the defect type to be classified, a target defect designation unit that receives a designation of the defect type to be the target of classification in the basic defect type classification unit, and a feature amount designation unit that receives a designation of a feature amount obtained based on the imaging information and used for classification in the basic defect type classification unit.

[0018] By doing so, in order for the classification condition setting unit to set classification conditions, the user can specify the weighting of the defect type, the defect type to be the classification target, and the feature amount used for classification. Therefore, the classification result according to the classification conditions corresponding to the user's purpose can be confirmed on the classification result confirmation screen, and more useful information can be obtained for setting the classification conditions.

[0019] Also, in the present invention, The classification condition setting unit may set the feature amount and the threshold value of the feature amount based on the designation received through at least any one of the importance designation unit, the target defect designation unit, and the feature amount designation unit.

[0020] In this way, without the user performing the complicated operation of setting the feature amount and the threshold value for classifying defect types according to the purpose, based on the weighting of defect types, the defect types to be classified, and the feature amounts used for classification, the feature amount and the threshold value for classifying defect types based on the feature amount can be set according to the purpose.

Effect of the Invention

[0021] According to the present invention, it is possible to provide useful information for setting classification conditions for classifying defect types of an inspection object.

Brief Description of the Drawings

[0022] [Figure 1] It is a schematic configuration diagram of an appearance inspection management system according to an embodiment of the present invention. [Figure 2] It is a flowchart showing the procedure of classification result confirmation processing according to an embodiment of the present invention. [Figure 3] It is a diagram showing an example of image data according to an embodiment of the present invention. [Figure 4] It is a diagram showing an example of display of an input screen for setting classification conditions according to an embodiment of the present invention. [Figure 5] It is a diagram showing an example of display of a classification result confirmation screen according to an embodiment of the present invention.

Mode for Carrying Out the Invention

[0023] 〔Application Example〕 Hereinafter, application examples of the present invention will be described with reference to the drawings.

[0024] Figure 1 is a schematic diagram of the visual inspection management system 1 to which the present invention is applied. The visual inspection management system 1 mainly comprises a visual inspection device 2 and an inspection management device 3. The visual inspection device 2 is a device for acquiring a visual image of a sheet-like object T to be inspected and detecting defects based on the image. This visual inspection device 2 is connected to the inspection management device 3 via a network so as to be able to communicate bidirectionally.

[0025] In the visual inspection device 2, when classifying the defect type based on the visual image of the inspected object T in which a defect has been detected, appropriate thresholds are set for various feature quantities (e.g., defect length, defect width, etc.) obtained from the visual image. If the classification conditions (also called classification logic), which include such feature quantities and thresholds for defect type classification, are not set appropriately, important defects will be missed, making accurate defect detection difficult. On the other hand, associating the visual image with the defect type shown in the visual image and generating correct answer data for setting the classification logic is not easy when the number of defects and defect types are large. For this reason, the visual inspection management system 1 classifies at least a portion of the correct answer data according to the classification logic set based on the information entered by the user, and provides the user with the classification result as a classification result confirmation screen 6 described later, so that the appropriateness of the set classification logic can be easily judged and evaluated.

[0026] Figure 4 shows an example of the display of input screen 5 for setting classification conditions. The user selects and specifies the defect type to be classified from multiple defect types, each numbered, such as "metal," by checking the checkbox in the "Target Defect" column. In Figure 4, four types are selected for classification: "metal," "pinhole," "foreign object," and "wrinkle." Input screen 5 also has a "critical defect" item. "Critical defects" refer to the defect type that the user considers particularly important among the multiple defect types. In Figure 4, the checkbox in the "metal" row is checked. By specifying "critical defects" in this way, a classification logic is set that ensures no defects are overlooked for the defect type designated as "critical." In addition, on the classification result confirmation screen 6, different display colors are used to distinguish each defect type, so a pull-down menu is provided to select the display color for each defect type. Furthermore, input screen 5 also has checkboxes for selecting and specifying the features to be used.

[0027] Based on the information entered via input screen 5, the results of classifying pre-prepared correct answer data are provided to the user as a classification result confirmation screen 6, as shown in Figure 5. The configuration of the classification result confirmation screen 6 will be explained below using "metals" as an example.

[0028] The classification result confirmation screen 6 includes a defect type column 61, a major defect display column 62, a logic result (number of items) column 63, a number display column 66, 67, and a total column 68. The defect type column 61 displays the defect type name, "metal" 611, and the display color for "metal" (for the sake of notation, in Figure 5, shading is used instead of color for distinction). The major defect display column 62 displays a check mark 621 indicating that "metal" is a major defect. The breakdown of the classification results by the classification logic is displayed in the logic result (number of items) column 63 and the number display columns 66, 67. The number of image data classified as "metal" is indicated by indicators 651, 652 on bars displayed in the horizontal display area 631, and by numbers 662, 672 displayed in the number display columns 66, 67 along with squares 661, 671 indicating the display color. In the center of the display area 631, the number of correct answers is 642, that is, the correct data that was classified and related to "metal". The number of image data items that were assigned is displayed numerically, and an inverted triangle mark 641 is displayed, which marks the rightmost position of indicator 651 when the classification result matches the correct number. As shown in Figure 5, indicator 651, which indicates that the correct type is "metal," extends from the left edge of the display area 631 to the position of the inverted triangle mark 641. In addition, in the count display area 66, eight image data items with the correct type "metal" are correctly classified as "metal," and this matches the "8" in the correct number 642 displayed on the display area 631, indicating that "metal" is being classified without being overlooked by this classification logic. In the "metal" display area 631, indicator 652 with a display color representing "pinhole" is also displayed. This indicates that this classification logic over-represents "metal," causing "pinhole" to be classified as "metal." In this way, the number of "pinhole" image data items that were incorrectly classified as "metal" can be accurately determined by the number "3" in the count display area 67 and the square 671 that displays the display color indicating the defect type.

[0029] As shown in Figure 5, the classification result confirmation screen 6 allows for a clear visual inspection of which defect types are critical defects, whether each defect type, including critical defects, has been correctly classified, and whether there have been any oversights or over-exposures, using bar-shaped indicators 651, the number of correct answers 642, and marks 641 provided in the display area 631. The specific number of cases can be grasped from the display fields 66 and 67. Furthermore, since the indicator 651 showing the number of image data is color-coded according to the correct type, the breakdown of correct types of image data classified into each defect type by the classification logic can be clearly recognized. By displaying classification result confirmation screens from multiple classification logics on the display device 37 in parallel or by switching between them, and comparing the classification results, it is possible to easily consider which classification logic is more appropriate according to the purpose of the inspection.

[0030] [Example 1] The configuration of the visual inspection management system 1 according to Embodiment 1 of the present invention will be described below with reference to the drawings. However, the configuration of the device described in this embodiment should be modified as appropriate depending on various conditions. In other words, the scope of this invention is not intended to be limited to the following embodiment.

[0031] (Visual Inspection Management System) Figure 1 is a schematic diagram showing the overall configuration of the visual inspection management system 1 according to an embodiment of the present invention. The visual inspection management system 1 mainly comprises a visual inspection device 2 and an inspection management device 3.

[0032] (Visual inspection device) The visual inspection device 2 is a device for acquiring visual images of sheet-like articles and detecting defects based on those images. Its main components include a lighting system, a measuring system, a transport mechanism (not shown), and a control terminal 23.

[0033] The object to be inspected T is transported horizontally (in the direction of the arrow) by a transport mechanism (not shown), and during transport, an image of the object's appearance is continuously acquired by a measurement system, and inspection is performed based on this image. The object to be inspected T is formed in a sheet shape, and examples include paper, cloth, and film. Furthermore, it is not limited to a single material, but may be a sheet body having multiple layers, such as packaging paper made by laminating film and nonwoven fabric. It may also be a food product such as dried seaweed.

[0034] The lighting system includes a light source 211 that irradiates the surface of the object T under inspection with visible light (e.g., white light). These light sources may include, for example, LED lighting.

[0035] The measurement system includes a camera 221 that captures light (hereinafter referred to as surface reflected light) that is irradiated from the light source 211 and reflected from the surface of the object T under inspection. The camera is equipped with a light receiving sensor capable of detecting the light it captures, a lens, and a signal output unit, and outputs the light detected by the light receiving sensor as an electrical signal via the lens. For example, a CCD or CMOS sensor can be used as the sensor.

[0036] The camera 221 captures an image of the object under inspection while it is illuminated by light from the light source 211. The control terminal 23 processes the captured image, and by comparing the obtained feature values ​​with a preset inspection threshold, any locations with feature values ​​that deviate from the threshold are determined to be defects.

[0037] The control terminal 23 has the following functional modules: an image acquisition unit 231, a feature calculation unit 232, a defect determination unit 233, and a defect type classification unit 234.

[0038] The image acquisition unit 231 has the function of capturing images from the camera 221, for example, by acquiring an image of the object under inspection when illuminated by illumination light. The feature quantity calculation unit 232 has the function of calculating feature quantities used for visual inspection based on the image of the object under inspection. Note that the feature quantities are not limited to one, for example, defect peak (%), defect width (mm), defect length (mm), defect area (mm) 2), various types of information may be calculated, such as brightness information, circularity, Ferre ratio, roundness, extensibility, average density, density pattern, vertical density change, horizontal density change, plane density change, hue (H) peak, saturation (S) peak, lightness (V) peak, R peak (light), R peak (dark), G peak (light), G peak (dark), B peak (light), B peak (dark), etc.

[0039] The defect determination unit 233 compares the feature quantities calculated by the feature quantity calculation unit 232 with a preset threshold and determines that any location with feature quantities that deviate from the threshold is a defect.

[0040] The defect type classification unit 234, when a defect is detected in the inspected object by the defect determination unit 233, classifies the type of defect based on a predetermined threshold and the image features that indicate the defect. The types of defects to be classified can be arbitrarily set by the user, for example, types such as metal, foreign matter, pinholes, streaks, gel, and resin can be set as appropriate. The defects are not limited to these, and types such as dirt and wrinkles may also be set, or they may be classified into even finer categories.

[0041] (Inspection and control device) The aforementioned visual inspection device 2 is connected to the inspection management device 3 via a network (LAN), and the visual inspection device 2 and the inspection management device 3 communicate information bidirectionally. The inspection management device 3 processes the information received from the visual inspection device 2 and transmits information related to the inspection to the visual inspection device 2. The inspection management device 3 is composed of a general-purpose computer system equipped with a CPU (not shown), a storage device 35, an input device 36, a display device 37, etc. The storage device 35 stores at least defect image data transmitted from the visual inspection device 2. Here, the defect image data includes the type and value of feature quantities obtained from images of the parts of the inspected object that have been determined to be defective, and supplementary information indicating the truth or falsity of the primary inspection determined by visual inspection, the type of defect, etc.

[0042] The inspection management device 3 may be composed of one computer or multiple computers. Alternatively, all or part of the functions of the inspection management device 3 may be implemented on the control terminal 23 of the visual inspection device 2. Alternatively, some of the functions of the inspection management device 3 may be implemented on a server on the network (such as a cloud server).

[0043] The inspection management device 3 according to this embodiment includes, as a functional module, a correct answer data generation unit 31, and The system comprises a classification condition setting unit 32, a correct answer data classification unit 33, and a classification result confirmation screen generation unit 34. Here, the inspection management device 3 corresponds to the classification condition setting support device of the present invention. Furthermore, the correct answer data generation unit 31, the classification condition setting unit 32, the correct answer data classification unit 33, and the classification result confirmation screen generation unit 34 correspond to the basic information generation unit, the basic defect type classification unit, the classification condition setting unit, and the classification result confirmation screen generation unit of the present invention, respectively.

[0044] The correct answer data generation unit 31 displays image data of the object under inspection T stored in the storage device 35, which includes some kind of defect, on the display device 37, and receives input from the user via the input device 36 to associate the defect type with the displayed image. For example, multiple image data 40-49 as shown in Figure 3 are displayed on the display device 37, and the user checks the image data 40-49 and associates each image data 40-49 with a defect type such as metal, foreign matter, pinhole, streak, gel, resin, etc. In this way, correct answer data including the image data and the associated defect type is generated. The image data used in the correct answer data corresponds to the basic imaging information of the present invention, the defect type associated with this image data corresponds to the basic defect type of the present invention, and the correct answer data corresponds to the basic information of the present invention.

[0045] The classification condition setting unit 32 sets classification conditions that allow for appropriate classification of defect types based on the defect type, important defect, and features used, as described below, selected and specified by the user. The classification conditions (classification logic) include the above-mentioned features and a threshold value for those features. Any appropriate method can be used to set the classification logic based on the information selected and specified by the user. The classification conditions (classification logic) correspond to the classification conditions of the present invention.

[0046] The correct answer data classification unit 33 classifies at least a portion (or all) of the correct answer data stored in the storage device 35 into defect types according to the classification logic set by the classification condition setting unit 32. At least a portion of the correct answer data subject to classification by the correct answer data classification unit 33 corresponds to the basic information of the present invention.

[0047] The classification result confirmation screen generation unit 34 generates a classification result confirmation screen 6 that includes the results of classifying the correct answer data by the correct answer data classification unit 33, and displays it on the display device 37. The detailed configuration of the classification result confirmation screen will be described later. The classification result confirmation screen 6 corresponds to the classification result confirmation screen of the present invention.

[0048] Figure 2 shows the procedure for confirming the classification result in the inspection management device 3.

[0049] First, the correct answer data generation unit 31 receives input of image data containing defects and a defect type that the user associates with the image data containing defects (step S1). At this time, image information is acquired based on images obtained by imaging the object to be inspected that has defects using the visual inspection device 2. The correct answer data generation unit 31 displays a plurality of image data 40 to 49 containing defects, as shown in Figure 3, on the display device 37, and similarly accepts the user's specification of the defect type to be associated from a plurality of candidate defect types displayed on the display device 37 by clicking on the input device 36, etc. The image information may also include information other than images. In this way, correct answer data for classifying the type of defect detected from the image of the object to be inspected is prepared. The generated correct answer data is stored in a predetermined area of ​​the storage device 35. Here, the predetermined area of ​​the storage device 35 that stores the correct answer data corresponds to the basic information storage unit of the present invention.

[0050] Next, the classification condition setting unit 32 sets the feature quantities and thresholds that constitute the classification conditions for classifying the defect type (step S2). At this time, the user can select and specify the type of defect, etc., for which they want to check the classification results in the visual inspection management system 1. Figure 4 is an example of the display of the input screen 5 for the user to input information in order to set the classification conditions. At the top of the input screen 5, each row is numbered. The defect type specification field 51, where the defect type is displayed, is shown. The defect type field 511 displays metal, pinhole, foreign matter, wrinkle, dirt, and scratch, but the defect type is not limited to these. The target defect field 512 in the third column of input screen 5 displays checkboxes for selecting and specifying the defect type for which the classification result to be checked as the "target defect". By checking these checkboxes, the user can select and specify the defect type to be checked on the classification result confirmation screen 6 described later. Here, metal, pinhole, foreign matter, and wrinkle are selected as target defects. In addition, the important defect field 513 in the fourth column of input screen 5 displays checkboxes for selecting and specifying important defects. Important defects are the defect types that the user has judged to be particularly important among the defect types to be detected. By selecting and specifying important defects in this way, when setting the classification conditions, the feature quantities and thresholds are set so that the defect types designated as important defects are not overlooked. Here, metal is selected as an important defect. The fifth column of input screen 5 is provided as a color specification field 514. This allows the user to select a display color for each defect type in order to clearly distinguish the selected defect type when displaying the classification result screen described later. The appropriate display color can be selected from a pull-down menu. Here, metal, pinholes, foreign matter, and wrinkles are set to be displayed in blue, green, cyan, and magenta, respectively. Here, whether or not it is a significant defect is the weighting assigned to the defect type, and the significant defect column 513 corresponds to the importance designation section of the present invention. Also, the target defect column 512 corresponds to the target defect designation section of the present invention. Furthermore, a feature specification section 52 for selecting and specifying feature quantities to be used when classifying defect types is displayed at the bottom of the input screen 5. The feature specification section 52 displays checkboxes 521 to 523 for selecting and specifying feature quantities. Here, you can select whether or not to use "defect length," "defect width," and "defect area" as feature quantities by checking the checkboxes. The feature quantities that can be selected and specified are not limited to these. In this case, the feature specification section 52 corresponds to the feature specification section of the present invention. As described above, based on the defect target, critical defect type, and features to be used selected and specified by the user through the input screen 5, the classification condition setting unit 32 searches for classifiable features and thresholds, and sets features and thresholds suitable for classifying the selected target defects and critical defects.

[0051] Next, according to the classification conditions set in step S2, the correct data classification unit 33 classifies the image data included in the correct data prepared in step S1 (step S3). The classification results are stored in a predetermined area of ​​the storage device 35. The image data to be classified by the correct data classification unit 33 may also include data of defective images stored in the storage device 35.

[0052] Next, the classification result confirmation screen generation unit 34 generates the classification result confirmation screen 6 shown in Figure 5 based on the classification result of the correct data in step S3 and displays it on the display device 37 (step S4).

[0053] Figure 5 shows the classification result confirmation screen 6. In Figure 5, symbols are specifically assigned to the "metal" type; however, in the following explanation, similar configurations for other defect types will be described using the same symbols. On the left edge of the classification result confirmation screen 6, the defect type column 61 displays the names of the defect types classified in the column direction. Here, "metal" 611, "pinhole", "foreign object", and "wrinkle" selected on the input screen shown in Figure 3 are displayed. Along with these defect type names, the defect type column 611 displays squares 612 of the color used to display the corresponding defect type as specified on the input screen shown in Figure 3. However, in Figure 5, the defect types are distinguished by the type of shading instead of color. Here, each of the defect types displayed in the defect type column 61 corresponds to a predetermined defect type of the present invention.

[0054] To the right of the defect type column 61, a critical defect display column 62 is provided. (See the input drawing in Figure 3.) For "metals" designated as a significant defect on a surface, a check mark 621 is displayed to indicate that it is a significant defect.

[0055] To the right of the critical defect display area 62, the logic result (number of cases) area 63 displays the number of classification results based on the classification logic used to classify the defect types. Here, the number refers to the number of image data classified for each defect type displayed on the left edge of the classification result confirmation screen 6. The number is represented by bar-shaped indicators 651 to 656 displayed within the horizontal display area 631. Each of these indicators 651 to 656 has a length proportional to the number of cases for the corresponding defect type. In addition, in the center of the display area 631, the number of image data associated with the corresponding defect type (correct count) included in the classified image data is indicated by the number 642 and an inverted triangle mark 641 indicating its position within the display area 631. As for the indicators displayed within the display area 631 (for example, indicator 651), first, indicator 651, which shows the number of image data associated with the defect type displayed in the defect type area 61, is displayed so as to extend from the left edge to the right edge of the display area 63. Adjacent to the right of the indicator 651 showing the number of correctly classified image data, an indicator for incorrectly classified defect types (e.g., indicator 652) is displayed, extending further to the right. If there are multiple defect types associated with image data incorrectly classified for a particular defect type, an indicator for each defect type is displayed adjacent to the right. For each defect type, the indicator (e.g., indicator 661) showing the number of image data associated with correctly classified defect types aligns with the position of the mark 641 in the center of the display area 631 when the number matches the correct number. Therefore, the user can easily recognize whether they have missed or overlooked anything based on the positional relationship between the right edge of indicator 651 and the mark 641. Note that the number of image data for each defect type included in the image data subject to classification differs depending on the defect type, while the mark 641 showing the correct number is located in the center of the display area 631. As a result, the number of image data that can be displayed in the display area for each defect type row differs, and the length (unit length) of each indicator displayed in each defect type row also differs. Thus, the display area 631 is designed to display an indicator with a length corresponding to a number exceeding the number of correct answers for each defect type.Here, display area 631 corresponds to the display area of ​​the present invention. Mark 641 corresponds to the mark of the present invention. Indicators 651 to 656 correspond to the classification basic imaging information indices of the present invention, the length of indicators 651 to 656 corresponds to the configuration corresponding to the number of classification basic imaging information of the present invention, and the display color (shaded in Figure 5) of indicators 651 to 656 corresponds to the configuration corresponding to the basic defect type associated with the classification basic imaging information of the present invention. The number 642 for the correct answer corresponds to the number of correct basic imaging information of the present invention. The number represented by the length of indicators 651 to 656 corresponds to the number of classification basic imaging information of the present invention.

[0056] Furthermore, to the right of the display area 631, there are count display fields 66 and 67 that display the number of image data classified for each defect type. Count display field 66 displays the number 662, which indicates the number of image data correctly classified for the defect type displayed in the defect type field 61, and a colored square 661 representing the defect type (correct type) pre-associated with each image data. Count display field 67 displays a colored square 671 and the number 672, which represent the correct type of image data incorrectly classified for the defect type displayed in the defect type field 61. Finally, at the far right of the classification result confirmation screen 6, there is a total field 68, which displays the total number 681 of image data classified for each defect type. The numbers 662 and 672 in count display fields 66 and 67, and the sum of the numbers 681 in the total field 68, correspond to the number of classification basic imaging information items in this invention.

[0057] Figure 5 shows that, according to the classification logic, there are a total of 11 image data entries classified as metal, as indicated in the total column 68. Furthermore, the count display columns 66 and 67 and indicators 651 and 652 show that among the image data classified as metal, 8 entries have a correct type of metal, and 3 entries have a correct type of pinhole. [Display area] The number of correct answers displayed in the center of 63, 642, is 8, and the indicator 651, which indicates image data with the correct type being metallic, has reached the position of the inverted triangle mark 641. Therefore, the user can clearly see that all image data with the correct type being metallic has been correctly classified as metallic. Furthermore, the user can clearly see, through the color coding of indicators 651 and 652, and the count display fields 66 and 67, that in addition to image data with the correct type being metallic, image data with the correct type being pinholes is also incorrectly classified as metallic in this classification logic. In other words, this classification logic is over-recognizing metallics, specifically by classifying image data with the correct type being pinholes as metallic. Since the user has designated metallics as a critical defect, overlooking metallics is unacceptable, but over-recognizing them may be acceptable in some cases. The display also shows which image data types are being over-recognized, so this information can be used to easily judge the appropriateness of the classification logic, specifically the combination of features and thresholds.

[0058] In Figure 5, two image data items were classified as pinholes by the classification logic, and the indicator 653 has not reached the central inverted triangle mark 641, so the user can clearly see that there have been oversights regarding pinholes. Of the five image data items whose correct type is pinhole, the user can also clearly see that the three items that were not correctly classified were mistakenly classified as metal, as described above, by the number of correct items 642 in the center of the display area 63, the indicators 653 and 631 for each defect type row, and the display in the count display fields 66 and 67.

[0059] Figure 5 shows that, according to the classification logic, there are 8 image data items classified as foreign objects in the total column. Furthermore, the count display columns 66 and 67 and indicators 654 and 655 indicate that the classified foreign object data includes 3 images with the correct type being "foreign object" and 5 images with the correct type being "wrinkle." Since the number of correct answers displayed in the center of the display area is 3, and indicator 654, which indicates image data with the correct type being "foreign object," has reached the inverted triangle position, the user can clearly see that all image data with the correct type being "foreign object" has been correctly classified as such. Additionally, the user can clearly see, through the color coding of indicators 654 and 655 and the count display columns 66 and 67, that, in addition to image data with the correct type being "foreign object," image data with the correct type being "wrinkle" has also been classified as foreign objects by this classification logic. Since there are 3 correct foreign object identifications, the display area for the foreign object indicator only has space for 6 items. There is no display area 631 to display indicator 655, which is the length of 5 image data items where the correct type is wrinkles. Therefore, a right-pointing arrow 69 indicates that there is a portion of indicator 655 that exceeds the display area 631 and is not displayed. This allows the user to understand that the classification logic correctly classifies foreign objects, but over-exposure is occurring.

[0060] In Figure 5, the user can clearly see that, based on this classification logic, there are 10 image data items whose correct type is wrinkles, of which 5 are correctly classified, while the other 5 are incorrectly classified as foreign objects.

[0061] As described above, according to the classification result confirmation screen 6 shown in Figure 5, it is immediately clear from the bar-shaped indicator 651, etc., and the number of correct answers 642 and marks 641 provided in the display area 631, which indicate which defect types are important defects, whether each defect type, including important defects, has been correctly classified, and whether there have been any oversights or over-exposures. The specific number can also be grasped from the display in the number display fields 66 and 67. Furthermore, since the indicator 651, etc., which shows the number of image data, is color-coded according to the correct type, the breakdown of correct types of image data classified into each defect type according to the classification logic can be clearly recognized.

[0062] Figure 5 shows a classification result confirmation screen that displays the results of classifying image data using a single classification logic. However, since the classification results will differ depending on the classification logic, that is, the features and thresholds used to classify the image data, the classification result confirmation screens for multiple classification logics can be displayed on the display device 37 in parallel or by switching between them. By comparing the classification results, it is possible to easily consider which classification logic is more appropriate depending on the purpose of the inspection.

[0063] In the classification result confirmation screen 6 shown in Figure 5, the length of the bar-shaped indicator displayed within the horizontally elongated display area 631 indicates the number of image data. Such an indicator is an example of the classification basic imaging information index of the present invention, and the classification basic imaging information index, its form according to the number of classification basic information, and its form according to the basic defect type are not limited thereto.

[0064] <Note 1> A basic information storage unit (35) stores basic information, which includes imaging information obtained by imaging an object under inspection (T), basic imaging information associated with any of the defect types, and basic defect types which are the defect types associated with each of the basic imaging information. A classification condition setting unit (32) sets classification conditions for classifying imaging information obtained by imaging the object under inspection (T) into one of the aforementioned defect types, A basic defect type classification unit (33) that classifies the basic imaging information according to the classification conditions, A display unit (37) displays a classification result confirmation screen (6) which includes the number of classified basic imaging information, which is the basic imaging information classified into a predetermined defect type, the basic defect type associated with the classified basic imaging information, and the number of correct basic imaging information, which is the number of basic imaging information from the target basic imaging information that is associated with the predetermined defect type, by classifying target basic imaging information, which is at least a part of the basic imaging information, according to the classification conditions, A classification condition setting support device (3) equipped with the following. [Explanation of Symbols]

[0065] 3: Inspection and Management Equipment Department 6:Classification result confirmation screen 32: Classification Condition Setting Section 33: Correct Data Classification Unit 35: Storage device 37:Display section T: Object under inspection

Claims

1. A basic information storage unit stores basic information, which includes basic imaging information obtained by imaging an object under inspection, and which includes basic imaging information associated with any of the defect types, and basic defect types that are the defect types associated with each of the basic imaging information. A classification condition setting unit sets classification conditions for classifying imaging information obtained by imaging the object under inspection into one of the aforementioned defect types, A basic defect type classification unit that classifies the basic imaging information according to the classification conditions, A classification result confirmation screen generation unit generates a classification result confirmation screen that includes the number of classified basic imaging information, which is the target basic imaging information classified into a predetermined defect type, the basic defect type associated with the classified basic imaging information, and the number of correct basic imaging information, which is the number of basic imaging information associated with the predetermined defect type among the target basic imaging information, by classifying target basic imaging information, which is at least a part of the basic imaging information, according to the classification conditions, A display unit that displays the classification result confirmation screen, Equipped with, The classification result confirmation screen has a display area for each predetermined defect type that displays a classification basic imaging information index, which has a length corresponding to the number of classification basic imaging information items and is displayed in a manner corresponding to the basic defect type associated with the classification basic imaging information, and the display area includes a mark indicating that the number of classification basic imaging information items displayed by the classification basic imaging information index matches the number of correct basic imaging information items.

2. The classification condition setting support device according to claim 1, characterized in that the classification result confirmation screen includes a display area for displaying the classification basic imaging information index corresponding to a number exceeding the number of correct basic imaging information items in the display area.

3. A classification condition setting support device according to claim 1 or 2, comprising a basic information generation unit that captures images of an object to be inspected that has defects and displays the acquired imaging information on the display unit, and accepts input of the defect type associated with the imaging information.

4. The classification condition setting unit accepts the specification of weights for the type of defect to be classified. A classification condition setting support device according to any one of claims 1 to 3, comprising at least one of a severity designation unit, a target defect designation unit that accepts the designation of the defect type to be classified in the basic defect type classification unit, and a feature quantity designation unit that accepts the designation of feature quantities acquired based on the imaging information and used for classification in the basic defect type classification unit.

5. The classification condition setting support device according to claim 4, characterized in that the classification condition setting unit sets the feature quantity and the threshold value of the feature quantity based on the designation received through at least one of the importance designation unit, the target defect designation unit, and the feature quantity designation unit.

Images